Advances in Artificial Systems for Logistics Engineering III 3031361148, 9783031361142
This book comprises high-quality refereed research papers presented at the 3rd International Conference on Artificial In
401 78 68MB
English Pages 1106 [1107] Year 2023
Preface
Organization
Contents
Advances of Computer Algorithms and Methods
AI Chatbots for Banks: Evolving Trends and Critical Issues
1 Introduction
2 Insights into the State-of-the-Art
2.1 Drawing upon AI in the Financial Arena
2.2 Spotlight on Virtual Assistants and Chatbots
2.3 A Closer Look at AI-Powered Chatbots
3 Illustrative Case Studies
3.1 Erica by Bank of America Corporation (BofA)
3.2 Cora by the Royal Bank of Scotland plc (RBS)
3.3 Stella, Widdy and MArIO from Italy
4 Most Critical Issues for Discussion
5 Recommendations for Improved Use of Banking Chatbots
6 Conclusion
References
Traffic Flow Characteristics and Vehicle Road Coordination Improvement in Subterranean Interweaving
1 Introduction
2 Experiment and Result
2.1 Vehicle Experiment
2.2 Monitor Video Recognition
2.3 Trajectory Data Analysis
3 Further Analysis
3.1 Space Headway Analysis
3.2 Speed Analysis
3.3 Result Analysis
4 Solution
4.1 Some Solutions and Their Disadvantages
4.2 Introducing IBeacon Technology
4.3 Key Problems
5 Conclusion
References
Influential Factors and Implementation Path of Talent Digital Evaluation Based on ISM Model: Taking Electric Power Enterprises as an Example
1 Introduction
2 Construction of Digital Influence Factor Set for Talents in Power Transmission Industry
3 Analysis of Impact Factors of Power Transmission Industry
3.1 Adjacency Matrix
3.2 Accessibility Matrix
3.3 Result Derivation
4 Construction of Interpretation Model for Power Transmission Industry
5 Research Results and Enlightenment
5.1 Research Results
5.2 Research Enlightenment
5.3 Research Conclusion
References
Knowledge Associated with a Question-Answer Pair
1 Introduction
2 Knowledge of the Reactive Agent
3 Interrogative Formula of the “Search Instruction” Type Question
4 Knowledge Structure in Question-Answer Transactions
4.1 The Language of Ternary Description
4.2 Representation of Declarative Knowledge in Question and Answer
5 Conclusion
References
Research on Low Complexity Differential Space Modulation Detection Algorithm
1 Introduction
2 Maximum Likelihood Detection Algorithm
2.1 ML Detection Algorithm Process
2.2 Complexity Analysis of ML Detection Algorithm
3 TDC Detection Algorithm
3.1 TDC Detection Algorithm Process
3.2 Complexity Analysis of TDC Detection Algorithm
4 Simulation Analysis
4.1 Simulation Analysis of Bit Error Rate Performance
4.2 Complexity Simulation Analysis
5 Conclusion
References
Mathematical Model of the Process of Production of Mineral Fertilizers in a Fluidized Bed Granulator
1 Introduction
2 Review of Mathematical Modelling Methods of Processes of Dehydration and Granulation in the Fluidized Bed
3 Mathematical Model of the Process of Production of Mineral Fertilizers in a Fluidized Bed Granulator
4 The Numerical Results Analysis of the Mathematical Model
5 Conclusion
References
Simulation Study on Optimization of Passenger Flow Transfer Organization at Nanning Jinhu Square Metro Station Based on Anylogic
1 Introduction
2 Analysis of Transfer Facilities in Nanning Jinhu Square Metro Station
2.1 General Layout
2.2 Analysis of Advantages and Disadvantages of Channel Transfer
2.3 Analysis of Transfer Facilities in Jinhu Square Station
3 Results and Analysis of Passenger Flow Transfer at Nanning Jinhu Square Metro Station
3.1 Passenger Flow Data Analysis
3.2 Station Simulation Modeling
3.3 Station Simulation Modeling
3.4 Bottleneck Analysis of Passenger Flow Transfer at Stations Based on Simulation Model
4 Optimization and Simulation of Passenger Flow Transfer Organization at Jinhu Square Metro Station
4.1 Optimization Scheme
4.2 Operation Results of Optimized Simulation Model
5 Conclusion
References
Engine Speed Measurement and Control System Design Based on LabVIEW
1 Instruction
2 Hardware System Design
2.1 Rotate Speed Sensor
2.2 Data Acquisition Card
3 Software System Design
3.1 Program Block Diagram Design
3.2 Front Panel Design
4 Experiment of Measurement and Control System
4.1 Experimental Conditions
4.2 Experimental Procedure
4.3 Analysis of Experimental Results
5 Conclusion
References
Research and Design of Personalized Learning Resources Precise Recommendation System Based on User Profile
1 Introduction
2 Research on Recommendation of Educational Resources
2.1 User Portrait
2.2 Research on Personalized Recommendation of Educational Resources
3 Key Links to Building a Learner Portrait Model
4 Precise Recommendation of Personalized Learning Resources Based on User Portraits
4.1 Technical Framework
4.2 Recommendation Policies
5 Precision Recommendation System Architecture for Online Learning Resources
5.1 Data Layer
5.2 Data Analysis Layer
5.3 Learner Analysis Module
5.4 Resource Analysis Module Includes Resource Portrait and Resource Similarity Analysis
6 Conclusion
References
Mixed Parametric and Auto-oscillations at Nonlinear Parametric Excitation
1 Introduction
2 System Model and Equations of Motion
3 Construction of Solutions of Equations
4 Stability of Stationary Movements
5 Calculations
6 Discussion and Conclusion
References
Spectrum Analysis on Electricity Consumption Periods by Industry in Fujian Province
1 Introduction
2 Basic Idea of Spectral Analysis
3 Data Preprocessing
4 Study on Periods by Single-Spectrum Analysis
4.1 Big Industry
4.2 Non-general Industry
4.3 Residential Living
4.4 Non-residential Lighting
4.5 Commercial Electricity
4.6 Agricultural Electricity
5 Conclusion
References
Determination of the Form of Vibrations in Vibratory Plow’s Moldboard with Piezoceramic Actuator for Maximum Vibration Effect
1 Introduction
2 Literature Review
3 Materials and Methods
4 Experiments and Results
5 Conclusions
References
Evaluating Usability of E-Learning Applications in Bangladesh: A Semiotic and Heuristic Evaluation Approach
1 Introduction
2 Related Works
3 Methodology
3.1 Application Selection
3.2 Heuristic Evaluation
3.3 Semiotic Evaluation
4 Results
5 Conclusion
References
Perceptual Computing Based Framework for Assessing Organizational Performance According to Industry 5.0 Paradigm
1 Introduction
2 Framework for Assessing Organizational Performance
2.1 Outline of Assessment Process
2.2 Representation of Words as Type-2 Fuzzy Sets
2.3 Computing with Words in the Perceptual Computer
3 Subdomain of Assessing Functional State of Employees
3.1 Assessing Functional State as Part of Human-Centric Approach
3.2 Components of the Perceptual Computer
3.3 Results for a Model Exercise
4 Conclusions and Further Research
References
Petroleum Drilling Monitoring and Optimization: Ranking the Rate of Penetration Using Machine Learning Algorithms
1 Introduction
2 Related Literature
2.1 Traditional ROP Models
2.2 Data Science Models
3 Proposed Methodology
3.1 Maintaining the Integrity of the Specifications
3.2 Understanding the Problem and Data Collection
3.3 Methodology
3.4 Data Modelling and Evaluation
3.5 Evaluation of the Random Forest Regressor
3.6 Feature Importance and Ranking
4 Conclusion
References
Application of Support Vector Machine to Lassa Fever Diagnosis
1 Introduction
2 Related Works
3 Methodology
3.1 Data Collection
3.2 Data Preprocessing
3.3 Input Variable Transformation and Target Variable
3.4 Use Case Diagram of the System
3.5 The Support Vector Machine Algorithm
3.6 Proposed Evaluation
3.7 Exploratory Analysis
3.8 Evaluation Metrics
4 Conclusion
References
A Novel Approach to Bat Protection IoT-Based Ultrasound System of Smart Farming
1 Introduction
2 Related Work
3 Methodology of Circuit Implementation
3.1 Model Selection Diagram
3.2 System Circuit Implementation
3.3 Architecture Model Implementation
4 Experimental Result and Discussion
5 Conclusion
References
Synthesis and Modeling of Systems with Combined Fuzzy P + I - Regulators
1 Problem Setting
2 Analysis of Recent Research and Publications
3 Presentation of the Main Research Material
4 Conclusions
References
Protection of a Printing Company with Elements of Artificial Intelligence and IIoT from Cyber Threats
1 Introduction
2 Analysis of Recent Research and Publications
3 Presentation of the Main Research Material
3.1 Building an Expert Simulator and Training ANNs by Conventional Means
3.2 Learning ANNs for Recognition from Adversarial Attacks
3.3 Comparison of System Performance
4 Conclusion
References
AMGSRAD Optimization Method in Multilayer Neural Networks
1 Introduction
1.1 AMSGrad Optimization Method
2 Methodology
2.1 Software Implementation
2.2 Logistics Function
3 Application of the AMSgrad Optimization Method
3.1 AMSGrad and Different Number of Epochs
3.2 Analysis of the Block Structure in the AMSGrad
4 Summary and Conclusion
References
Regional Economic Development Indicators Analysis and Forecasting: Panel Data Evidence from Ukraine
1 Introduction
2 Literature Review
3 Method of Panel Data Modelling in Economic Research
4 Results: Economic Development of the Regions of Ukraine Model Specification
5 Discussion: Panel Data Model of Ukraine Regions Economic Development Application
6 Summary and Conclusion
References
An Enhanced Performance of Minimum Variance Distortionless Response Beamformer Based on Spectral Mask
1 Instruction
2 Mvdr Beamformer
3 The Proposed Method
4 Experiments
5 Conclusion
References
Modelling Smart Grid Instability Against Cyber Attacks in SCADA System Networks
1 Introduction
2 Related Works
3 Datasets and Methods
3.1 Attack-Instability Modelling
3.2 Decentral Smart Grid Control (DSGC) Model
3.3 Dataset
4 Experimental Setup
5 Results and Discussion
6 Conclusion
References
Program Implementation of Educational Electronic Resource for Inclusive Education of People with Visual Impairment
1 Introduction
2 Literature Analysis
3 Object, and Research Methods
4 Results
5 Conclusion
References
A Version of the Ternary Description Language with an Interpretation for Comparing the Systems Described in it with Categorical Systems
1 Introduction
2 Ternary Description Language Version with Explicit Interpretation
3 Conclusion
References
A Hybrid Centralized-Peer Authentication System Inspired by Block-Chain
1 Introduction
2 Motivation and Goals
3 System Architecture and Design
3.1 User Sign up
3.2 Initial User Validation
3.3 Certificate from Central Server
3.4 Live Message #1
3.5 Certificate from peers & Live message #2
3.6 Confirmation with Central Side
3.7 Creation of the Authenticated Certificate
4 Scenarios
4.1 No Certificate from the Central Side
4.2 Peers Might be Offline
5 Evaluation and Discussion
5.1 Message Size Before and After the Execution of Some Functions
5.2 Number of Hashing Functions and Encryption – Decryption Functions
6 System Security Analysis
6.1 Identity Fraud
6.2 Network Spoofing
6.3 Replays in Certificates
7 Conclusion
References
Heuristic Search for Nonlinear Substitutions for Cryptographic Applications
1 Introduction
2 Related Works
3 Bijective 8-bit S-boxes
4 The Walsh-Hadamard Transform and Nonlinearity of Substitution
5 New Heuristics for Finding Nonlinear Substitutions
6 Experimental Results
7 Discussion of Results and Conclusions
References
Dangerous Landslide Suspectable Region Forecasting in Bangladesh – A Machine Learning Fusion Approach
1 Introduction
2 Literature Review
3 Methodology
3.1 Data Collection
3.2 Data Preprocessing
3.3 Machine Learning Models
4 Experiment and Results
5 Discussion
6 Conclusions
References
New Cost Function for S-boxes Generation by Simulated Annealing Algorithm
1 Instruction
2 Related Works
3 Methods
4 New Cost Function and Experimental Results
5 Comparison of Results and Discussion
6 Conclusion
References
Enriched Image Embeddings as a Combined Outputs from Different Layers of CNN for Various Image Similarity Problems More Precise Solution
1 Introduction
2 Tasks to Solve and State-of-the-Art Methods Overview
2.1 Near Duplicate Images
2.2 Multi-Angle Views
2.3 Schematical Layouts
2.4 Existing Approaches and the State-of-the-Art Overview
3 Methodology. Image Embedding Constructing and Using to Solve the Tasks
3.1 Image Feature Extractor – Embedding Builder
3.2 Embeddings Comparison, Search Optimization, and an Appropriate Storage
3.3 Decision-Making Unit
4 Real-World Tasks Application Results and Comparison
5 Summary and Conclusion
References
A Novel Approach to Network Intrusion Detection with LR Stacking Model
1 Introduction
1.1 Stacking Model
2 Methodology
2.1 Dataset and Preprocessing
2.2 Feature Extraction and Model Training
2.3 Experimental Setup and Sequence
3 Result Analysis
4 Conclusion and Future Work
References
Boundary Refinement via Zoom-In Algorithm for Keyshot Video Summarization of Long Sequences
1 Introduction
2 Related Work
3 Algorithm Description
4 Experiments and Results
4.1 Datasets
4.2 Evaluation Protocol
4.3 Implementation Details and Setup
4.4 Results
4.5 Ablation Study
4.6 Study on Importance of Introduced Components
4.7 Comparison with D-KTS at Higher Framerates
5 Conclusions
References
Solving Blockchain Scalability Problem Using ZK-SNARK
1 Introduction
2 ZK-SNARK Technology to Solve the Scalability Problem
3 Our Solution
3.1 Aggregation
3.2 Recursion
4 Time and Measurement Results
4.1 Aggregation Method
4.2 Recursion Method
4.3 Discussion
5 Conclusion
References
Interactive Information System for Automated Identification of Operator Personnel by Schulte Tables Based on Individual Time Series
1 Introduction
2 Materials and Methods
2.1 Schulte Tables Methodology
2.2 Operator Activity Modeling
2.3 Multidimensional Average Method
3 Results
4 Discussion
5 Summary and Conclusion
References
DIY Smart Auxiliary Power Supply for Emergency Use
1 Introduction
1.1 Problem Statement
1.2 Analysis of Recent Research and Publications
1.3 Purpose of the article
2 Development Requirements
3 Research Results
3.1 Possibility of Use
3.2 Advantages and Disadvantages of Auxiliary Power Supply
3.3 The Architecture of the Developed Device (MVP)
3.4 Remote Control and Indication
3.5 System Load Testing
4 Summary and Conclusion
References
A Computerised System for Monitoring Water Activity in Food Products Using Wireless Technologies
1 Introduction
2 Related Work
3 Methodology and Regulatory Principles for Determining Water Activity in Food Products
3.1 Regulation of the Water Activity Indicator in the Main International Regulatory Documents
3.2 Main Methods and Instruments for Water Activity Measuring
3.3 Analysis of the Main Criteria for Choosing a Method of Measuring Water Activity
4 A Specialised Computer System for Monitoring Water Activity
4.1 Hardware design
4.2 Software Design
5 Conclusion
References
Reengineering of the Ukrainian Energy System: Geospatial Analysis of Solar and Wind Potential
1 Introduction
2 Materials and Methods
3 Results
4 Summary and Conclusion
References
Complex Approach for License Plate Recognition Effectiveness Enhancement Based on Machine Learning Models
1 Introduction
2 Problem Statement
3 Problem Solving
3.1 Hardware Component
3.2 Analysis of DSS for Pattern Recognition Problem Solving
3.3 An Approach to the Automation of a DSS Training Task
3.4 Creation of a Set of Training Input Data Using the Proposed Approach
3.5 Software Validation in Real Conditions
3.6 Interaction with External Detection Systems
4 Conclusion
References
The Analysis and Visualization of CEE Stock Markets Reaction to Russia’s Invasion of Ukraine by Event Study Approach
1 Introduction
2 Literature Review
3 Methodology
3.1 Research Data
3.2 Shock Estimation by “Price Spread – Volume Volatility” Indicators
3.3 Shock Estimation by Price Changing
3.4 ESG-based Analysis
4 Results
4.1 Markets’ Overall Visualization of Shock: Indexes Handling
4.2 Shock Estimation by “PRICE Spread – Volume Volatility” Indicators
4.3 Shock Estimation by Price Changing
4.4 Investigation of Shock Transition by Stocks with Different ESG Scores
5 Summary and Conclusion
References
The Same Size Distribution of Data Based on Unsupervised Clustering Algorithms
1 Introduction
2 Methodology
3 Opportunities of Existing Unsupervised Clustering Algorithms to Distribute Data into Clusters of Equal Size
3.1 Hierarchical Clustering
3.2 K-means Clustering
3.3 Bisecting K-means Clustering
3.4 DBSCAN Clustering
4 Clustering Algorithm Based on Square of Vector Modulus
5 Results
6 Conclusion
References
Investigation of Microclimate Parameters in the Industrial Environments
1 Introduction
1.1 Defining of the Peculiarities of Resource Support for the Technological Process of Poultry Breeding
2 The Comparative Analysis of the Insulation Isotherms Experimental Results
2.1 Solving Model Based on Graphic Dependencies
3 Processing Results by Geometric Method Using Complex Drawing
4 Summary and Conclusion
References
Detection of Defects in PCB Images by Separation and Intensity Measurement of Chains on the Board
1 Instruction
2 PCB Image Preprocessing Algorithms
2.1 Distributed Cumulative Histogram
2.2 K-Means Clustering Slgorithm
2.3 Flood-Fill Algorithm
2.4 Thinning Algorithm
3 Selection and Separation of Chains
3.1 Connectivity Defects
3.2 Separation of Chains and Connectivity Defect Detection
3.3 Measurement of Intensity Defects
3.4 Evaluation by Distributed Cumulative Histogram
4 Summary and Conclusion
References
Tropical Cyclone Genesis Forecasting Using LightGBM
1 Introduction
2 Literature Review
3 Methodology
3.1 Problem Formulation
3.2 Dataset
3.3 Feature Selection
3.4 LightGBM Algorithm
3.5 Train with LightGBM
4 Result and Discussion
5 Conclusion
References
Optimization of Identification and Recognition of Micro-objects Based on the Use of Specific Image Characteristics
1 Instruction
2 Methodology for Identification and Recognition of Microobjects Based on the Use of Redundant Information Structures
3 Optimal Image Identification with Detection and Correction of Distorted Points Based on Wavelet Filtering
4 Analysis of the Results of the Study
5 Conclusion
References
Development and Comparative Analysis of Path Loss Models Using Hybrid Wavelet-Genetic Algorithm Approach
1 Instruction
2 Literature Review
2.1 COST231 Model
3 Methodology
3.1 Method of Data Collection
3.2 Proposed Hybrid Wavelet-GA Method
4 Results
5 Discussion
6 Conclusion
References
Mathematical Advances and Modeling in Logistics Engineering
Risk Assessment of Navigation Cost in Flood Season of the Upper Reaches of the Yangtze River Based on Entropy Weight Extension Decision Model
1 Introduction
2 Risk Assessment Index System of Navigation Cost in Flood Season in the Upper Reaches of the Yangtze River
2.1 Impact of Flood Period on the Navigation Cost of Ships in the Upper Reaches of the Yangtze River
2.2 The Risk Assessment System of Ship Navigation Cost in Flood Season in the Upper Reaches of the Yangtze River
3 Extension Evaluation Model of Risk Entropy Weight for Navigation Cost in Flood Season in the Upper Reaches of the Yangtze River
3.1 Evaluation Object, Classical Domain and Section Domain of the Model
3.2 Correlation Function and Correlation Degree of the Model
3.3 Comprehensive Correlation Degree and Risk Level of the Model
4 Test of Navigation Cost Risk Assessment Model in Flood Period of Upper Yangtze River Based on Entropy Weight Extension Decision Model
4.1 Data Source of Test Case
4.2 Optimization of Evaluation Index Date
4.3 Determination of Case Evaluation Index Weight
4.4 Determine of the Classical Domain and Matter Element to be Evaluated
5 Conclusion
References
Influencing Factors and System Dynamics Analysis of Urban Public Bicycle Projects in China Based on Urban Size and Demographic Characteristics
1 Introduction
2 Relevant Research on Public Bike
3 Analysis on Characteristics of Chinese Cities
3.1 City Size and Process of Urbanization
3.2 Traffic Mode Division
3.3 Demographic Characteristics
3.4 Bicycle Infrastructure and Right of Way Characteristics
3.5 Regime Structure
4 System Dynamics Analysis on the Relationship Between Bicycle Travel and Public Bicycle
4.1 Reason for Using the System Dynamics Method
4.2 System Dynamics Analysis of the Relationship Between Bicycle and Public Bicycle
5 Conclusion
References
Cross-border Logistics Model Design e-Tower Based on Blockchain Consensus Algorithm
1 Introduction
2 Common Consensus Algorithms in Blockchain Technology
2.1 PoW Algorithm
2.2 PoS Algorithm
2.3 DPoS Algorithm
2.4 PBFT Algorithm
3 e-Tower Design for Shenzhen ZK Company Cross-border Logistics
3.1 Overall Architecture Design
3.2 Composition of Underlying P2P Network
3.3 Application of PBFT Algorithm
3.4 Transaction Process
3.5 Application in Cross-Border Customs Clearance
4 Privacy and Security Verification of e-Tower
5 Conclusion
References
Research on the Service Quality of JD Daojia's Logistics Distribution Based on Kano Model
1 Introduction
2 The Connotation of JD Daojia and Kano Model
2.1 JD Daojia
2.2 Kano Model
3 JD Daojia Service Quality Improvement Model Based on Kano Model
3.1 Questionnaire Design and Data Collection
3.2 Data processing of Better-Worse coefficient
3.3 Data Analysis and Conclusion
4 Simulation Case JD Daojia's Service Quality Improvement Strategy
5 Conclusion
References
CiteSpace Based Analysis of the Development Status, Research Hotspots and Trends of Rural E-Commerce in China
1 Introduction
2 Research Design
2.1 Research Methods
2.2 Data Source
3 Analysis of Research Status
3.1 Time Distribution Analysis
3.2 Author Statistics and Cooperative Network Analysis
3.3 Analysis of Paper Publishing Journals and Cooperation Networks
4 Research Hotspot and Trend Analysis
4.1 Keyword Co-occurrence Analysis
4.2 Keyword Emergence Analysis
4.3 Analysis Results and Development Strategies of Rural E-Commerce
5 Research Conclusions and Prospects
5.1 Research Conclusion
5.2 Prospect of Future Research on Rural E-Commerce
References
The Optimization and Selection of Deppon Logistics Transportation Scheme Based on AHP
1 Introduction
1.1 Status Analysis of Deppon Logistics Transportation Scheme
2 Deppon Logistics Transportation Plan Optimization and Selection Analysis
2.1 Introduction to the Analytic Hierarchy Process
2.2 The Mathematical Model of Analytic Hierarchy Process
2.3 Analysis of Influencing Factors of Deppon Logistics Transportation Scheme
2.4 Application of Analytic Hierarchy Process in Deppon Logistics Transportation Scheme Optimization
3 Solution and Application of Numerical Examples
3.1 General Situation of Logistics Transportation in a Certain Area of Wenzhou
3.2 Examples of Three Logistics Transportation Schemes
3.3 Comparative Analysis of Three Logistics Transportation Schemes
4 Introduction of Calculation
4.1 Model
4.2 Weight Calculation
4.3 Hierarchical Sorting and Order
5 Conclusion
References
Pharmacological and Non-pharmacological Intervention in Epidemic Prevention and Control: A Medical Perspective
1 Introduction
1.1 Selection of Journals
1.2 Chronology of Growth in Epidemic Prevention and Control Research
2 Pharmacological Intervention
3 Non-pharmacological Intervention
4 Limitations and Directions for Future Research
5 Conclusions
References
Passenger Flow Forecast of the Section of Shanghai-Kunming High-Speed Railway from Nanchang West Station to Changsha South Station
1 Introduction
2 Influencing Factors and Forecasting Methods of Passenger Flow
2.1 Analysis of Influencing Factors of Passenger Flow Forecast
2.2 Commonly Used Methods of Passenger Flow Forecasting and Their Applicability Analysis
3 Passenger Flow Forecasting for the Section of Shanghai-Kunming High-Speed Railway from Nanchang West Station to Changsha South Station
3.1 An Overview of the Shanghai-Kunming High-speed Railway
3.2 A Passenger Flow Forecasting Method Based on the Artificial Neural Network
3.3 Passenger Flow Forecast Based on the Exponential Smoothing Method
3.4 Passenger Flow Forecast Based on the Time Series Method
3.5 Comparative Verification of the Forecasting Effect
4 Forecast of Passenger Flow in Coming Years Based on the Artificial Neural Network
4.1 Input Data Forecast in the Forecast of Passenger Flow in Coming Years
4.2 Results of Passenger Flow Forecast in the Coming Years
5 Conclusion
References
The Effect of Labor Rights on Mental Health of Front-Line Logistics Workers: The Moderating Effect of Social Support
1 Introduction
2 Literature Review and Hypothetical Inference
2.1 Labor Rights and Mental Health
2.2 The Moderating Effect of Social Support
3 Data Collection
3.1 Questionnaire Design and Chosen Scale
3.2 Respondents
4 Data Analysis and Research Hypothesis Testing
4.1 Data Analysis
4.2 Research Hypothesis Testing
5 Conclusion and Management Suggestions
5.1 Conclusion and Discussion
5.2 Management Suggestions
References
An Investigation into Improving the Distribution Routes of Cold Chain Logistics for Fresh Produce
1 Introduction
2 Research Status at Home and Abroad
2.1 Domestic Research Status
2.2 Research Status Abroad
3 Characteristics of Cold Chain Logistics of Fresh Agricultural Products
3.1 High Requirements for Logistics Efficiency
3.2 Strict Temperature Control Requirements
3.3 Wide Distribution of Supply Chain
3.4 High Cost
4 Current Situation Analysis of Cold Chain Logistics of Fresh Agricultural Products
4.1 Lack of Perfect Cold Chain Logistics System
4.2 Lack of Specialized Talents in Cold Chain Logistics
4.3 Backward Cold Chain Infrastructure
5 Optimization of Cold Chain Logistics Distribution Path of Fresh Agricultural Products
5.1 Problem Description
5.2 Using Mileage Saving Method to Optimize Distribution Path
6 Conclusion
References
Development of Vulnerability Assessment Framework of Port Logistics System Based on DEMATEL
1 Introduction
2 Literature Review
2.1 Vulnerability
2.2 Port Logistics System
2.3 Vulnerability of Port Logistics System
2.4 Vulnerability Assessment Methods
3 Port Logistics Vulnerability Assessment Model
3.1 Assessment Framework
3.2 DEMATEL Method
4 Case Study
5 Conclusion
References
Application Prospect of LNG Storage Tanks in the Yangtze River Coast Based on Economic Model
1 Introduction
2 The Role of LNG Tank Container in the Storage and Transportation Market
2.1 Basic Concept of LNG Tank Container
2.2 Current Situation of LNG Tank Container Storage and Transportation in China
2.3 Function Analysis of LNG Tank Container
3 Economic Analysis of LNG Tank Container Storage and Transportation
3.1 LNG Transport Ship Type Selection
3.2 Construction of Economic Measurement Model
3.3 Calculation of Storage and Transportation Economy
4 Application Prospect of LNG Tank Container Along the Yangtze River
4.1 Analysis of Domestic LNG Application Prospect
4.2 LNG Market Demand Along the Yangtze River
4.3 Prediction of LNG Tank Container Market Scale Along the Yangtze River
4.4 Suggestions for the Development of LNG Tank Transportation Along the Yangtze River
5 Conclusion
References
Multi-depot Open Electric Truck Routing Problem with Dynamic Discharging
1 Introduction
2 Problem Description and Mathematical Formulation
2.1 Problems’ Description and Mathematical Notation
2.2 Electric Truck Power Consumption Model
2.3 Mathematical Models for DD-MDOEVRPTW
3 Hybrid Genetic Algorithm for DD-MDOEVRPTW
3.1 Coding Design
3.2 Population Initialization
3.3 Constraint Processing and Fitness Functions
3.4 Genetic Manipulation
3.5 Insert Local Search
4 Numerical Experiments and Analysis
5 Conclusion and Outlook
References
Analysis on the Selection of Logistics Distribution Mode of JD Mall in the Sinking Market
1 Introduction
2 Overview of Relevant Theories
2.1 Sinking Market
2.2 The Sinking Market E-commerce Theory
2.3 Related Theories of E-commerce Logistics Distribution
3 Analysis on the Current Situation of JD Mall in the Sinking Market
3.1 The Reason Why JD Mall Entered the Sinking Market
3.2 Analysis on the Existing Logistics Distribution Mode of JD Mall
3.3 Analysis of Problems Existing in Logistics Distribution Mode
4 Analysis on the Logistics Distribution Model Selection of JD Mall in the Sinking Market
4.1 Establishment of Evaluation Index System for Logistics Distribution Mode Selection
4.2 Index Weight Determination and Consistency Testing at All Levels
4.3 Construction of Fuzzy Hierarchy Model for Logistics Distribution Mode of JD Mall in Sinking Market
4.4 Final Evaluation Results and Analysis
5 Conclusion
References
Research on Cruise Emergency Organization Based on Improved AHP-PCE Method
1 Introduction
1.1 Characteristics of Major Cruise Emergencies
2 Cruise Ship Emergency Organization System for Major Emergencies
2.1 Cruise Ship Emergency Management Capability for Major Emergencies
2.2 Analysis of Factors Affecting the Emergency Response System for Major Cruise Ship Emergencies
3 Evaluation Study of Cruise Ship Emergency Response Organizations for Major Emergencies
3.1 Index Selection Criteria
3.2 Construction of Index System
3.3 Fuzzy Hierarchical Analysis
4 Cruise Ship Major Emergency Response Organization Evaluation
4.1 Calculation of Indicator Weights
4.2 Fuzzy Evaluation of Emergency Organization for Cruise Major Emergencies
5 Conclusion
References
Fresh Agricultural Products Supplier Evaluation and Selection for Community Group Purchases Based on AHP and Entropy Weight VIKOR Model
1 Introduction
1.1 Evaluation System of Green Suppliers of Fresh Agricultural Products
2 Methodology
2.1 Hierarchical Analysis to Determine Subjective Weights
2.2 Entropy Weighting Method to Determine Objective Weights
2.3 Calculate the Composite Weights
2.4 VIKOR Method Judging Model
3 Empirical Research
3.1 Determination of Indicator Weights
3.2 VIKOR Method to Determine the Optimal Option
4 Analysis and Suggestions
5 Related Work
5.1 Determination of Indicator Weights
5.2 Evaluation Criteria
5.3 Evaluation Methods
6 Conclusion and Future Work
References
Logistics of Fresh Cold Chain Analysis of Joint Distribution Paths in Wuhan
1 Introduction
2 Relevant Conceptions
2.1 Theory of Cold Chain Logistics
2.2 Joint Distribution
2.3 Vehicle Distribution Routing Issues
3 Analysis of the Current Situation and Problems of Common Distribution of Fresh Food Cold Chain Logistics in Wuhan
3.1 Fresh Cold Chain Logistics Development in Wuhan
3.2 Wuhan Cold Chain Logistics Infrastructure Improvement
4 Application of the Mileage Saving Method
4.1 The Principle of the Mileage Saving Method
4.2 Conditions of Application of the Mileage Saving Method
4.3 Model Construction for the Mileage Saving Method
5 Design of a Common Distribution Route Optimization Scheme
5.1 Data Selection
5.2 Draw a Simplified Road Map
5.3 Optimal Design of Paths
6 Conclusion
References
Optimization of Logistics Distribution Route Based on Ant Colony Algorithm – Taking Nantian Logistics as an Example
1 Introduction
2 Problem Description
2.1 Description of Problems Faced by Enterprises
2.2 Modeling and Optimization of Nantian Logistics Distribution Path
3 Model Establishment
3.1 Determine Coordinate Points
3.2 Assumptions
3.3 Parameter Setting
4 Model Solution and Result Analysis
4.1 Model Construction
4.2 Model Solution Results
5 Result Analysis
6 Conclusion
References
Optimization Research of Port Yard Overturning Operation Based on Simulation Technology
1 Introduction
2 Theoretical Review
2.1 Box-Turning Operation and Its Impact
2.2 Simulation Platform for Container Rail-Water Intermodal
3 Case Analysis
3.1 Introduction to Huangshi Newport
3.2 There Are Problems in the Box-Turning Operation of Huangshi Newport
4 Simulation and Optimization of Simulation Platform for Container Rail-Water Intermodal
4.1 Theoretical Analysis
4.2 Simulation Experiment of Container Rail-Water Intermodal Platform
5 Countermeasures and Suggestions
5.1 Basic Idea of Yard Operation Optimization
5.2 Basic Principles for Optimization of Yard Operation
5.3 Reducing the Turning Rate
5.4 Specific Methods to Reduce Overturning Speed
6 Conclusion
References
A Review of Epidemic Prediction and Control from a POM Perspective
1 Introduction
1.1 Selection of Journals
1.2 Research Trend
2 Epidemic Simulation and Prediction
2.1 Compartmental Model
2.2 Statistical Model
2.3 Hybrid Model
3 Epidemic Control Measures
4 Medical Materials Production in an Epidemic
5 Conclusions and Future Research Direction
References
Application of SVM and BP Neural Network Classification in Capability Evaluation of Cross-border Supply Chain Cooperative Suppliers
1 Introduction
2 Introduction to Evaluation Model
2.1 BP Neural Network Model
2.2 Support Vector Machines Model
3 Establishment of Supplier Competency Evaluation System
4 Classification Experiment Design
4.1 Experimental Design of BP Neural Network Classification
4.2 Experimental Design of SVM Classification
4.3 Experimental Analysis
5 Conclusions
References
Research on Port Logistics Demand Forecast Based on GRA-WOA-BP Neural Network
1 Introduction
2 Model Approach
2.1 Grey Relational Analysis
2.2 GRA-WOA-BP Neural Network Model
3 Port Logistics Demand Impact Indicator System Construction
3.1 Influencing Factor Indicator Selection
3.2 Input and Output Variables are Determined
4 Analysis of Simulation Experiments
4.1 Neural Network Parameter Setting
4.2 Analysis of Model Results
5 Conclusion
References
Evaluation and Optimization of the Port A Logistics Park Construction Based on Fuzzy Comprehensive Method
1 Introduction
2 Overview of Relevant Theories
2.1 Fuzzy Comprehensive Method
2.2 Characteristics of Logistics Park Construction Project
2.3 Risk Characteristics of Logistics Park Construction Projects
3 Analysis on Risk Factors of Port a Logistics Park Construction
3.1 Analysis on the Environment of Port a Logistics System
3.2 Analysis on Risk Factors of Logistics Park Construction
4 Evaluation of Port a Logistics Park Construction Based on Fuzzy Comprehensive Method
4.1 Construction of Risk Evaluation Index System for Logistics Park
4.2 Establish Indicators and Evaluation System
4.3 Effect Fuzzy Evaluation Matrix
4.4 Fuzzy Evaluation Calculation
5 Optimization of Port a Logistics Park Construction
5.1 Improve the Service Mode of Diversified Logistics Parks
5.2 The Logistics Park is Upgraded to Intelligence
5.3 Actively Seek Government Support
6 Conclusion
References
Advances in Technological and Educational Approaches
OBE Oriented Teaching Reform and Practice of Logistics Information System Under the Background of Emerging Engineering Education
1 Introduction
2 Reform Objectives and Ideas of Course Teaching
2.1 Ideas for Curriculum Teaching Reform
2.2 Determination of Course Teaching Objectives
3 Content Reconstruction and Organization Implementation of Curriculum Teaching
3.1 Reconstruction of Course Teaching Content
3.2 Organization and Implementation of Teaching Process
4 Process Based Multi Indicator Comprehensive Course Evaluation and Analysis of Course Goal Achievement
5 Conclusion
References
Teaching Practice of “Three Integration” Based on Chaoxing Learning Software – Taking the Course of “Complex Variable Function and Integral Transformation” as an Example
1 Introduction
2 Three Major Problems in Course Teaching
2.1 Lack of Activity in Large Class Teaching
2.2 Low Integration of Teaching Content and Ideological and Political Education
2.3 Mathematics Teaching and Professional Application Derailed
3 Practice of Innovative Teaching
3.1 Innovative Thinking of “Three Integration”
3.2 Processing Teaching Content
3.3 Improve Teaching Methods and Means
3.4 Improve Teaching Evaluation Methods
4 Effect of Innovative Teaching
4.1 Improvement of Students’ Learning Autonomy
4.2 The Number of Students Winning Awards Has Increased Significantly
4.3 The Students’ Learning Feedback is Constantly Praised
5 Conclusion
References
Transformation and Innovation of E-Commerce Talent Training in the Era of Artificial Intelligence
1 Introduction
2 The Requirements for Business Talents in the Era of Artificial Intelligence and the Orientation of Their Training Transformation
2.1 Requirements for Business Talent in the Era of Artificial Intelligence
2.2 Path Selection of Business Talent Training in the Era of Artificial Intelligence
3 Influencing Factors of School-Enterprise Collaboration in Cultivating E-Commerce Talents in the Era of Artificial Intelligence
3.1 The Basic Situation of E-commerce Education in Higher Vocational Colleges
3.2 The Development of Cooperative Education between New Business Schools and Higher Vocational Colleges
4 School-Enterprise Collaborative Education Strategy for E-Commerce Talents in the Era of Artificial Intelligence
4.1 Cultivate E-commerce Talents’ Philosophical Thinking Under the Background of Artificial Intelligence
4.2 The Systematic Construction of “Central-Local” Legislative Hierarchy of School-Enterprise Collaborative Education
4.3 Improve the Social Training System and Realize Education and Social Equity
5 Conclusion
References
Talent Training Mode Based on the Combination of Industry-Learning-Research Under the Background of Credit System Reform
1 Introduction
2 Literature Review
2.1 Research on Credit System
2.2 Research on the Cooperation Between Industry, University and Research
2.3 Research Review at Home and Abroad
3 Analysis of Talent Training Mode Under the Background of Foreign Credit System
3.1 Characteristics of Talent Training Mode Under the Credit System in the United States
3.2 Characteristics of Talent Training Mode Under the Background of British Credit System
3.3 Characteristics of Talent Training Mode Under the Background of German Credit System
4 The Classification of Talent Training Modes of Industry-Learning-Research Cooperative Education Under the Credit System and the Problems to be Concerned
4.1 The Classification of Talent Training Modes of Industry-Learning-Research Cooperative Education Under the Credit System
4.2 Problems that Should Be Paid Attention to in Flexible Talent Cultivation Under the Background of Credit System Reform
5 Conclusion
References
Analysis of the Innovation Mechanism and Implementation Effect of College Students’ Career Guidance Courses Based on Market Demand
1 Introduction
2 The Practical Significance of the Innovation of Career Guidance Courses for College Students
2.1 Be Beneficial to Talent Reserve for Industrial Technological Revolution
2.2 Be Beneficial for College Students to Achieve Higher Quality Employment
2.3 Be Beneficial for Efficiency Improvement of Enterprise Human Resources
3 The Common Problems Existing in the Current Career Guidance Courses for College Students
3.1 Insufficient Attention from Colleges and the Curriculum is not Systematic
3.2 The Teaching Content is Boring and not Very Practical
3.3 Large Class Teaching, Lack of Individual Guidance
4 Analysis of the Innovation Mechanism of College Students’ Career Guidance Course Based on Market Demand
4.1 Help College Students Broaden Their Horizons and Develop a “Strong Heart” for Employment
4.2 Promote Graduates to Improve Themselves and Have a “Voice” in Career
4.3 Whole Process and Whole Staff Participation in Talent Cultivation, to Create a Sense of Professional “Atmosphere”
5 Implementation Effect Evaluation
6 Conclusion
References
Comparative Study on the Development of Chinese and Foreign Textbooks in Nanomaterials and Technology
1 Introduction
2 Selection of Chinese and Foreign Textbooks
3 Comparation of the Design of the Structure of Content in Chinese and Foreign Textbooks
3.1 Comparation of the Design of Knowledge System in Chinese and Foreign Textbooks
3.2 Comparation of the Design of Content of Examples and Exercises in Chinese and Foreign Textbooks
4 Comparation of the Morphology of Chinese and Foreign Textbooks
5 Conclusion
References
Practical Research on Improving Teachers’ Teaching Ability by “Train, Practice and Reflect” Mode
1 Introduction
2 Design and Implementation of the “Train, Practice and Reflect” Coupling Linkage Teaching Ability Training Program
2.1 Scheme Design
2.2 Program Implementation
3 Research Methods
3.1 Survey Subjects
3.2 Scale Design
4 Data Analysis and Discussion
4.1 Reliability Test
4.2 Average Value, Standard Deviation and Independent Sample T-Tests
5 Conclusion
References
College Foreign Language Teacher Learning in the Context of Artificial Intelligence
1 Introduction
2 Literature Review
2.1 Teacher Learning
2.2 College Foreign Language Teacher Learning in the Context of AI
3 Research Design
3.1 Subjects
3.2 Instruments
3.3 Data Collection and Analysis
4 Results and Discussion
4.1 Factors of Foreign Language Teacher Learning
4.2 Characteristics of Foreign Language Teacher Learning
4.3 The Effects of AI on Foreign Language Teacher Learning
5 Conclusion and Implications
References
The Innovation Integration Reform of the Course “Single Chip Microcomputer Principle and Application”
1 Introduction
2 Main Problems in the Course of Single Chip Microcomputer Principle and Application
3 Reform of the Course of the Integration of Specialization and Innovation in the Principle and Application of Single Chip Microcomputer
3.1 Design of Multi-function LED Stream Lamp
3.2 Design of Intelligent Car Competition Timer
3.3 Design of “Explorer” Smart Car
3.4 Design of Waveform Acquisition and Display System Based on PC Display
3.5 Matrix of Teaching Content Based on the Reform of Integration of Specialty and Creativity
4 Phased Achievements of the Reform of Integration of Professional and Creative Work
5 Conclusion
References
Discussion and Practice on the Training Mode of Innovative Talents in Economics and Management in Women’s Colleges
1 Introduction
2 Definition of Economic and Management Applied Innovative Talents
3 Analysis of the Elements of the Training Mode of Economic and Management Applied Innovative Talents
3.1 Talent Cultivation Concept
3.2 Professional Setting Mode
3.3 Teaching Evaluation Methods
4 Difficulties in the Implementation of Innovative Talent Training Mode in Economic Management
5 Taking the School of Business Administration of Shandong Women’s University as an Example
5.1 Thoughts on the Reform of Talent Training Mode
5.2 An Integrated Curriculum System Based on Spiral Ability Training
5.3 Establish an Effective Evaluation System for Training Innovative Talents in Engineering Application
6 Conclusion
References
Cultivation and Implementation Path of Core Quality of Art and Design Talents Under the Background of Artificial Intelligence
1 Introduction
2 Challenges Faced by Art and Design Talents in the Era of Artificial Intelligence
2.1 AI technology Promotes the Innovation of Art and Design Talents
2.2 Intensified Contradictions in the Employment Market of Art and Design Talents in the Context of Artificial Intelligence
3 Core Quality of Art and Design Talents
3.1 Requirements of Art and Design Talents for Core Quality Training
3.2 Elements of Core Quality of Art and Design Talents
3.3 Construction of Core Quality Training System for Art and Design Talents
4 Research on the Implementation Path of the Core Quality Training of Art and Design Talents
4.1 Overall Framework and Specific Content of Core Literacy
4.2 Teaching Design Strategy for Cultivating Core Literacy
4.3 Teaching Implementation Strategy of Core Literacy Cultivation
4.4 Teaching Evaluation Strategies for Cultivating Core Literacy
5 Conclusion
References
Reform and Innovation of International Logistics Curriculum from the Perspective of Integration of Industry and Education
1 Introduction
2 The Existing Problems of the Curriculum
2.1 The Teaching Contents Are Different from the Enterprise Practice
2.2 Concepts and Skills Are Cumbersome and Complex that Cognitive Points Are Overloaded
2.3 Fragmented Knowledge Points Are not Easy to Generate Deep Learning
3 Innovation Strategy
3.1 Conceptual Understanding
3.2 Modular Recombination
3.3 Project-based Implementation
4 Teaching Process Organization and Design
4.1 Facing the Industry, Docking Enterprises, Reconstructing the Course Content System
4.2 Three Steps to Promote, Three Steps to Transfer, the Implementation of Project-Based Teaching
4.3 Process Motivation, Result Evaluation, to Ensure the Learning Effect
5 Teaching Effect of Integration of Production and Education
6 Conclusion
References
An Analysis of Talent Training in Women’s Colleges Based on the Characteristics of Contemporary Female College Students
1 Introduction
2 Research on Characteristics of Women’s Colleges and Universities
2.1 Research on the Goal and Idea of Running a School
2.2 Research on the Characteristics of the Main Body of Running a School
2.3 Research on the Characteristics of Talent Training
2.4 Research on the Characteristics of Disciplines and Specialties
2.5 Research on the Characteristics of Curriculum Construction
3 Research Method
4 Analysis of Characteristics of Contemporary Female College Students
4.1 An Analysis of the Psychological Characteristics of Contemporary Female College Students
4.2 An Analysis of the Characteristics of Contemporary Female College Students
5 An Analysis of Talent Training in Women’s Colleges Based on the Characteristics of Contemporary Female College Students
5.1 Training Specifications Integrating Knowledge, Ability and Quality
5.2 Integrated Curriculum System with Spiral Ability Training as the Main Line
6 Conclusion
References
Solving Logistical Problems by Economics Students as an Important Component of the Educational Process
1 Introduction
1.1 Literature Review
1.2 Software Tools for Solving Logistics Tasks
2 Features of Solving Logistics Problems
2.1 Logistical Tasks in the Professional Activity of Economists
2.2 Implementation of the Mathematical Model
2.3 Analysis of the Results Obtained
3 An Example of a Lesson Dedicated to Solving Logistical Problems
3.1 Initial Data and Problem Statement
3.2 Task Solving and Interpretation of Results
3.3 Analysis of Alternative Solutions
4 Summary and Conclusion
References
Exploration and Practice of Ideological and Political Construction in the Course of “Container Multimodal Transport Theory and Practice” for Application-Oriented Undergraduate Majors—Taking Nanning University as an Example
1 Introduction
2 The Necessity of Ideological and Political Construction of “CMT” Course for Application-Oriented Undergraduate
3 Basic Thinking of Ideological and Political Construction of the Course “CMT”
3.1 Mining Ideological and Political Elements of Curriculum Modules
3.2 Weight Analysis of Curriculum Ideological and Political Elements
3.3 Weight Analysis Results of Curriculum Ideological and Political Elements
4 Ideological and Political Construction Path and Practice of CMT Course
4.1 Select Teachers with Outstanding Ideological and Political Ability to Teach, and Create a Capable Subject of Ideological and Political Construction of Curriculum
4.2 Compile Applicable and Visible Teaching Materials, Outlines and Courseware, and Build a Useful Course Construction Carrier Reflecting Ideological and Political Content
4.3 Plan and Design a Number of Curriculum Implementation Rules, Including Teacher Evaluation and Collaborative System, to Form a Guarantee Mechanism for Establishing Morality and Cultivating People
4.4 Add Value-Leading Indicators and Strengthen the Process Assessment and Evaluation of Ideological and Political Teaching
5 Conclusion
References
A Study on Learning Intention of Digital Marketing Micro Specialty Learners Under the Background of New Liberal Arts—Based on Structural Equation Model
1 Introduction
2 Theoretical Basis and Research Hypothesis
2.1 Theoretical Basis
2.2 Research Model and Hypothesis
3 Research Design
3.1 Research Object
3.2 Research Tools
4 Data Analysis
4.1 Common Method Deviation Test and Correlation Analysis
4.2 Structural Equation Test
4.3 Mediating Effect Test
5 Conclusions and Suggestions
References
Comparisons of Western and Chinese Textbooks for Advanced Electronic Packaging Materials
1 Introduction
2 Descriptions of the Chosen Textbooks
3 Comparisons Between Western and Chinese Textbooks
3.1 Format Structure
3.2 Knowledge System Design
3.3 Knowledge Point Introduction
3.4 Information Technology Application
4 Effects of Textbook Improvement
5 Conclusion
References
Innovation and Entrepreneurship Teaching Design in Application-Oriented Undergraduate Professional Courses – Taking the Transportation Enterprise Management Course as an Example
1 Introduction
1.1 The Development Status of Innovation and Entrepreneurship Education for College Students
1.2 The Relationship Between Application-Oriented Talent Training and Innovation and Entrepreneurship Teaching
1.3 Problems to be Solved in this Paper
2 The Role and Significance of Integrating Innovation and Entrepreneurship Teaching into the Course of Transportation Enterprise Management
2.1 Analysis of Students
2.2 The Promotion of Innovation and Entrepreneurship Teaching to Professional Courses Teaching
2.3 The Course Meets the Needs of Enterprise Operation and Management
3 Innovation and Entrepreneurship Elements Contained in the Course
3.1 Teaching Requirements for the Course
3.2 Main Teaching Contents of the Course
3.3 Analysis of Elements of Innovation and Entrepreneurship
4 Innovation and Entrepreneurship Teaching Design
4.1 Project Design
4.2 Comprehensive Practical Design
4.3 Teaching Design
4.4 Assessment Design
5 Conclusion
References
The Construction of University Teachers’ Performance Management System Under the Background of Big Data Technology
1 Introduction
2 Problems of Performance Management in Colleges and Universities Under the Background of Big Data
3 The Influence of Big Data on University Teachers’ Performance Evaluation Platform
4 Implementation of University Performance Evaluation Management Platform Based on Big Data
4.1 Architecture of University Performance Evaluation Management Platform
4.2 Construction of University Performance Data Warehouse
5 Conclusion
References
Curriculum Evaluation Based on HEW Method Under the Guidance of OBE Concept
1 Introduction
1.1 OBE
2 Hierarchy Entropy Weight Method Model
2.1 Hierarchy Entropy Weight Method Model Process
2.2 Assessment Criteria for Curriculum Under OBE Concept
2.3 Hierarchical Assessment Criteria
2.4 EWM Process
3 Curriculum Evaluation Under OBE Concept
3.1 Expert Selection
3.2 Weight of Assessment Criteria Base on EWM
4 Examples of Curriculum Evaluation
4.1 Curriculum Evaluation
4.2 Characteristics of Curriculum Evaluation Under OBE Concept
5 Conclusion
References
The Relevance of a Systematic Approach to the Use of Information Technologies in the Educational Process
1 Introduction
1.1 Literature Review
1.2 Directions of Digitalization of the Educational Process
2 A Systematic Approach to the Use of Information Technologies in the Educational Process
2.1 The Essence of the System Approach
2.2 Modern Information Technologies in the Educational Process
3 Application of Information Technologies in the Course “Economic Theory”
3.1 Use of Internet Resources
3.2 Use of Data Analysis and Visualization Technologies
3.3 Use of Knowledge Assessment Technologies
4 Summary and Conclusion
References
Construction and Practice of “CAD/CAM Foundation” Course Based on Learning Outcome
1 Introduction
2 Background and Course Analysis
2.1 Course Objective
2.2 Course Content
2.3 Learning Basis and Learning Effect of Previous Sessions
3 Practice of Curriculum Reform
3.1 Curriculum Planning Based on Outcome Objectives
3.2 Mobilize the Initiative of Students
3.3 Outcome Evaluation
4 Results and Discussion
5 Conclusion
References
Research and Practice of Ideological and Political Education in the Context of Moral Education and Cultivating People
1 Introduction
2 Problems in Curriculum Ideology and Politics
3 Research on the Path of Ideological and Political Teaching Reform of Automobile Electric Appliances Course
3.1 Reform of ideological and Political Teaching System of Automobile Electric Appliances Course
3.2 Ideological and Political Teaching Content Reform of Automobile Electric Appliances Course
3.3 Reform of Ideological and Political Teaching Methods of Automobile Electric Appliances Course
3.4 Ideological and Political Teaching Evaluation Reform of Automobile Electric Appliances Course
4 Effect of Ideological and Political Teaching Reform of Automobile Electric Appliances Course
5 Conclusion
References
A Quantitative Study on the Categorized Management of Teachers’ Staffing in Colleges and Universities
1 Introduction
2 Research Methods
2.1 Methods of Empirical Research
2.2 Methods of Comparative Analysis
2.3 Methods of Mathematical Statistics
3 Prediction Results
4 Analysis and Suggestions
4.1 Entry and Exit Mechanism of Teachers
4.2 Enrichment of Staffing Methods
4.3 Establish and Improve Relevant Supporting Measures
5 Conclusion
References
Course Outcomes and Program Outcomes Evaluation with the Recommendation System for the Students
1 Introduction
2 Related Works
3 Methodology
3.1 Course Outcome (CO)
3.2 Program Outcomes (PO)
3.3 CO/PO Evaluation Process
4 Design and Implementation
4.1 Modules of the System
4.2 Module Description
5 Result and Discussion
6 Recommendation for the Students
7 Conclusion
References
Methodology of Teaching Educational Disciplines to Second (Master’s) Level Graduates of the “Computer Science” Educational Program
1 Introduction
1.1 Problem Statement
1.2 Literature Review. Analysis of Recent Research and Publications
1.3 Highlighting Aspects that are not Sufficiently Studied
1.4 Purpose of the Article
1.5 Research Objectives (Goals)
2 Research Methods
2.1 Research Tools
2.2 Reliability and Accuracy of Results
2.3 Methodological Basis of the Study
3 Research Results
3.1 Theoretical Foundations of the Construction of Teaching Methods
3.2 Peculiarities of the Methodology, Application of Teaching Methods and Tools
3.3 Discussion of Research Results
4 Summary and Conclusion
4.1 Expanding the Boundaries of the Scientific Field
4.2 Scientific Novelty. Scientific Justification
4.3 Practical Use
4.4 Prospects for Further Research and Study
References
Professional Training of Lecturers of Higher Educational Institutions Based on the Cyberontological Approach and Gamification
1 Introduction
1.1 Problem Statement
1.2 Literature Review. Analysis of Recent Research and Publications
1.3 Highlighting Aspects that are Understudied
1.4 Purpose of the Article
1.5 Research Objectives (Goals)
2 Research Methods
2.1 Research Tools
2.2 Reliability and Accuracy of Results
2.3 Methodological Basis of the Study
3 Research Results
3.1 Theoretical Foundations of the “Cyberontological Approach”, the “Cyberpedagogy” Component
3.2 Practice of Applying the Cyberontological Approach in Educational Institutions
3.3 The Problem in the Use of Higher Educational Institutions with Specific Learning Conditions in the Educational Process
3.4 Discussion of Research Results
4 Summary and Conclusion
4.1 Expanding the Boundaries of the Scientific Field
4.2 Scientific Novelty. Scientific Justification
4.3 Practical Use
4.4 Prospects for Further Research and Study
References
Exploring the Perceptions of Technical Teachers Towards Introducing Blockchain Technology in Teaching and Learning
1 Introduction
2 Related Work
3 Methodology of Study
4 Results and Discussion
5 Conclusion
References
Author Index
Recommend Papers
![Advances in Artificial Systems for Medicine and Education IV (Advances in Intelligent Systems and Computing) [1st ed. 2021]
3030671321, 9783030671327](https://ebin.pub/img/200x200/advances-in-artificial-systems-for-medicine-and-education-iv-advances-in-intelligent-systems-and-computing-1st-ed-2021-3030671321-9783030671327.jpg)
![Advances in Computer Science for Engineering and Education III [1st ed.]
9783030555054, 9783030555061](https://ebin.pub/img/200x200/advances-in-computer-science-for-engineering-and-education-iii-1st-ed-9783030555054-9783030555061.jpg)
![Advances in Artificial Systems for Power Engineering II (Lecture Notes on Data Engineering and Communications Technologies) [1st ed. 2022]
3030970639, 9783030970635](https://ebin.pub/img/200x200/advances-in-artificial-systems-for-power-engineering-ii-lecture-notes-on-data-engineering-and-communications-technologies-1st-ed-2022-3030970639-9783030970635.jpg)
- Author / Uploaded
- Zhengbing Hu
- Qingying Zhang
- Matthew He
- Similar Topics
- Science (general)
- International Conferences and Symposiums
File loading please wait...
Citation preview
Lecture Notes on Data Engineering and Communications Technologies 180
Zhengbing Hu Qingying Zhang Matthew He Editors
Advances in Artificial Systems for Logistics Engineering III
Lecture Notes on Data Engineering and Communications Technologies Series Editor Fatos Xhafa, Technical University of Catalonia, Barcelona, Spain
180
The aim of the book series is to present cutting edge engineering approaches to data technologies and communications. It will publish latest advances on the engineering task of building and deploying distributed, scalable and reliable data infrastructures and communication systems. The series will have a prominent applied focus on data technologies and communications with aim to promote the bridging from fundamental research on data science and networking to data engineering and communications that lead to industry products, business knowledge and standardisation. Indexed by SCOPUS, INSPEC, EI Compendex. All books published in the series are submitted for consideration in Web of Science.
Zhengbing Hu · Qingying Zhang · Matthew He Editors
Advances in Artificial Systems for Logistics Engineering III
Editors Zhengbing Hu Faculty of Applied Mathematics National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute” Kyiv, Ukraine
Qingying Zhang College of Transportation and Logistics Engineering Wuhan University of Technology Wuhan, China
Matthew He Halmos College of Arts and Sciences Nova Southeastern University Fort Lauderdale, FL, USA
ISSN 2367-4512 ISSN 2367-4520 (electronic) Lecture Notes on Data Engineering and Communications Technologies ISBN 978-3-031-36114-2 ISBN 978-3-031-36115-9 (eBook) https://doi.org/10.1007/978-3-031-36115-9 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
The development of artificial intelligence (AI) systems and their applications in various fields is one of the modern science and technology’s most pressing challenges. One of these areas is AI and logistics engineering, where their application aims to increase the effectiveness of AI generation and distribution for the world’s population’s life support, including tasks such as developing industry, agriculture, medicine, transportation, and so on. The rapid development of AI systems necessitates an increase in the training of an increasing number of relevant specialists. AI systems have a lot of potential for use in education technology to improve the quality of training for specialists by taking into account the personal characteristics of these specialists as well as the new computing devices that are coming out. As a result of these factors, the 3rd International Conference on Artificial Intelligence and Logistics Engineering (ICAILE2023), held in Wuhan, China, on March 11–12, 2023, was organized jointly by Wuhan University of Technology, Nanning University, the National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Huazhong University of Science and Technology, the Polish Operational and Systems Society, Wuhan Technology and Business University, and the International Research Association of Modern Education and Computer Science. The ICAILE2023 brings together leading scholars from all around the world to share their findings and discuss outstanding challenges in computer science, logistics engineering, and education applications. Out of all the submissions, the best contributions to the conference were selected by the program committee for inclusion in this book. March 2023
Zhengbing Hu Qingying Zhang Matthew He
Organization
General Chairs Q. Y. Zhang Ivan Dychka
Wuhan University of Technology, China National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Ukraine
Online conference Organizing Chairs Z. B. Hu C. L. Wang Y. Wang
National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Ukraine Anhui University, China Wuhan University of Science and Technology, China
Program Chairs Q. Y. Zhang Matthew He G. E. Zhang
Wuhan University of Technology, China Nova Southeastern University, USA Nanning University, China
Publication Chairs Z. B. Hu Q. Y. Zhang Matthew He
National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Ukraine Wuhan University of Technology, China Nova Southeastern University, USA
Publicity Chairs Y. Z. Pang Q. L. Zhou
Nanning University, China Shizuoka University, Japan
viii
Organization
O. K. Tyshchenko Vadym Mukhin
University of Ostrava, Czech Republic National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Ukraine
Program Committee Members Margherita Mori X. H. Tao Felix Yanovsky D. Y. Fang Ivan Izonin L. Luo Essa Alghannam L. Xu H. F. Yang Y. C. Zhang A. Sachenko S. Gnatyuk X. J. Zhou Oleksandra Yeremenko Rabah Shboul W. B. Hu Yurii Koroliuk O. K. Tyshchenko H. L. Zhong J. L. Zhang G. K. Tolokonnikova
University of L’Aquila, Italy School of Intelligent Manufacturing of Nanning University, China Delft University of Technology, Netherlands Beijing Technology and Business University, China Lviv Polytechnic National University, Ukraine Sichuan University, China Tishreen University, Syria Southwest Jiaotong University, China Waseda University, Japan Hubei University, China Kazimierz Pułaski University of Technology and Humanities in Radom, Poland National Aviation University, Ukraine Wuhan Textile University, China Kharkiv National University of Radio Electronics, Ukraine Al-albayt University, Jordan Wuhan University, China Chemivtsi Institute of Trade and Economics, Ukraine University of Ostrava, Czech Republic South China University of Technology, China Huazhong University of Science and Technology, China FNAT VIM of RAS, Moscow, Russia
Contents
Advances of Computer Algorithms and Methods AI Chatbots for Banks: Evolving Trends and Critical Issues . . . . . . . . . . . . . . . . Margherita Mori and Lijing Du
3
Traffic Flow Characteristics and Vehicle Road Coordination Improvement in Subterranean Interweaving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Enshi Wang, Bihui Huang, and Bing Liu
14
Influential Factors and Implementation Path of Talent Digital Evaluation Based on ISM Model: Taking Electric Power Enterprises as an Example . . . . . . Wei Luo, Jiwei Tang, Saixiao Huang, and Yuan Chen
25
Knowledge Associated with a Question-Answer Pair . . . . . . . . . . . . . . . . . . . . . . Igor Chimir
35
Research on Low Complexity Differential Space Modulation Detection Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shuiping Xiong and Xia Wu
45
Mathematical Model of the Process of Production of Mineral Fertilizers in a Fluidized Bed Granulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bogdan Korniyenko and Andrii Nesteruk
55
Simulation Study on Optimization of Passenger Flow Transfer Organization at Nanning Jinhu Square Metro Station Based on Anylogic . . . . . Yan Chen, Chenyu Zhang, Xiaoling Xie, Zhicheng Huang, and Jinshan Dai
65
Engine Speed Measurement and Control System Design Based on LabVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chengwei Ju, Geng E. Zhang, and Rengshang Su
79
Research and Design of Personalized Learning Resources Precise Recommendation System Based on User Profile . . . . . . . . . . . . . . . . . . . . . . . . . . Tingting Liang, Zhaomin Liang, and Suzhen Qiu
90
Mixed Parametric and Auto-oscillations at Nonlinear Parametric Excitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alishir A. Alifov
101
x
Contents
Spectrum Analysis on Electricity Consumption Periods by Industry in Fujian Province . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Huawei Hong, Lingling Zhu, Gang Tong, Peng Lv, Xiangpeng Zhan, Xiaorui Qian, and Kai Xiao Determination of the Form of Vibrations in Vibratory Plow’s Moldboard with Piezoceramic Actuator for Maximum Vibration Effect . . . . . . . . . . . . . . . . . Sergey Filimonov, Sergei Yashchenko, Constantine Bazilo, and Nadiia Filimonova Evaluating Usability of E-Learning Applications in Bangladesh: A Semiotic and Heuristic Evaluation Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . Samrat Kumar Dey, Khandaker Mohammad Mohi Uddin, Dola Saha, Lubana Akter, and Mshura Akhter Perceptual Computing Based Framework for Assessing Organizational Performance According to Industry 5.0 Paradigm . . . . . . . . . . . . . . . . . . . . . . . . . Danylo Tavrov, Volodymyr Temnikov, Olena Temnikova, and Andrii Temnikov Petroleum Drilling Monitoring and Optimization: Ranking the Rate of Penetration Using Machine Learning Algorithms . . . . . . . . . . . . . . . . . . . . . . . Ijegwa David Acheme, Wilson Nwankwo, Akinola S. Olayinka, Ayodeji S. Makinde, and Chukwuemeka P. Nwankwo Application of Support Vector Machine to Lassa Fever Diagnosis . . . . . . . . . . . Wilson Nwankwo, Wilfred Adigwe, Chinecherem Umezuruike, Ijegwa D. Acheme, Chukwuemeka Pascal Nwankwo, Emmanuel Ojei, and Duke Oghorodi A Novel Approach to Bat Protection IoT-Based Ultrasound System of Smart Farming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Md. Hafizur Rahman, S. M. Noman, Imrus Salehin, and Tajim Md. Niamat Ullah Akhund Synthesis and Modeling of Systems with Combined Fuzzy P + I Regulators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bohdan Durnyak, Mikola Lutskiv, Petro Shepita, Vasyl Sheketa, Nadiia Pasieka, and Mykola Pasieka Protection of a Printing Company with Elements of Artificial Intelligence and IIoT from Cyber Threats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Bohdan Durnyak, Tetyana Neroda, Petro Shepita, Lyubov Tupychak, Nadiia Pasieka, and Yulia Romanyshyn
109
120
129
141
152
165
178
187
197
Contents
AMGSRAD Optimization Method in Multilayer Neural Networks . . . . . . . . . . . S. Sveleba, I. Katerynchuk, I. Kuno, O. Semotiuk, Ya. Shmyhelskyy, S. Velgosh, N. Sveleba, and A. Kopych
xi
206
Regional Economic Development Indicators Analysis and Forecasting: Panel Data Evidence from Ukraine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Larysa Zomchak, Mariana Vdovyn, and Olha Deresh
217
An Enhanced Performance of Minimum Variance Distortionless Response Beamformer Based on Spectral Mask . . . . . . . . . . . . . . . . . . . . . . . . . . . Quan Trong The and Sergey Perelygin
229
Modelling Smart Grid Instability Against Cyber Attacks in SCADA System Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . John E. Efiong, Bodunde O. Akinyemi, Emmanuel A. Olajubu, Isa A. Ibrahim, Ganiyu A. Aderounmu, and Jules Degila Program Implementation of Educational Electronic Resource for Inclusive Education of People with Visual Impairment . . . . . . . . . . . . . . . . . . Yurii Tulashvili, Iurii Lukianchuk, Valerii Lishchyna, and Nataliia Lishchyna
239
251
A Version of the Ternary Description Language with an Interpretation for Comparing the Systems Described in it with Categorical Systems . . . . . . . . G. K. Tolokonnikov
261
A Hybrid Centralized-Peer Authentication System Inspired by Block-Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wasim Anabtawi, Ahmad Maqboul, and M. M. Othman Othman
271
Heuristic Search for Nonlinear Substitutions for Cryptographic Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oleksandr Kuznetsov, Emanuele Frontoni, Sergey Kandiy, Oleksii Smirnov, Yuliia Ulianovska, and Olena Kobylianska Dangerous Landslide Suspectable Region Forecasting in Bangladesh – A Machine Learning Fusion Approach . . . . . . . . . . . . . . . . . . . . Khandaker Mohammad Mohi Uddin, Rownak Borhan, Elias Ur Rahman, Fateha Sharmin, and Saikat Islam Khan New Cost Function for S-boxes Generation by Simulated Annealing Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oleksandr Kuznetsov, Emanuele Frontoni, Sergey Kandiy, Tetiana Smirnova, Serhii Prokopov, and Alisa Bilanovych
288
299
310
xii
Contents
Enriched Image Embeddings as a Combined Outputs from Different Layers of CNN for Various Image Similarity Problems More Precise Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Volodymyr Kubytskyi and Taras Panchenko
321
A Novel Approach to Network Intrusion Detection with LR Stacking Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mahnaz Jarin and A. S. M. Mostafizur Rahaman
334
Boundary Refinement via Zoom-In Algorithm for Keyshot Video Summarization of Long Sequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alexander Zarichkovyi and Inna V. Stetsenko
344
Solving Blockchain Scalability Problem Using ZK-SNARK . . . . . . . . . . . . . . . . Kateryna Kuznetsova, Anton Yezhov, Oleksandr Kuznetsov, and Andrii Tikhonov Interactive Information System for Automated Identification of Operator Personnel by Schulte Tables Based on Individual Time Series . . . . . . . . . . . . . . . Myroslav Havryliuk, Roman Kaminskyy, Kyrylo Yemets, and Taras Lisovych DIY Smart Auxiliary Power Supply for Emergency Use . . . . . . . . . . . . . . . . . . . . Nina Zdolbitska, Mykhaylo Delyavskyy, Nataliia Lishchyna, Valerii Lishchyna, Svitlana Lavrenchuk, and Viktoriia Sulim
360
372
382
A Computerised System for Monitoring Water Activity in Food Products Using Wireless Technologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Oksana Honsor and Roksolana Oberyshyn
393
Reengineering of the Ukrainian Energy System: Geospatial Analysis of Solar and Wind Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Iryna Doronina, Maryna Nehrey, and Viktor Putrenko
404
Complex Approach for License Plate Recognition Effectiveness Enhancement Based on Machine Learning Models . . . . . . . . . . . . . . . . . . . . . . . . Yakovlev Anton and Lisovychenko Oleh
416
The Analysis and Visualization of CEE Stock Markets Reaction to Russia’s Invasion of Ukraine by Event Study Approach . . . . . . . . . . . . . . . . . . Andrii Kaminskyi and Maryna Nehrey
426
The Same Size Distribution of Data Based on Unsupervised Clustering Algorithms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Akbar Rashidov, Akmal Akhatov, and Fayzullo Nazarov
437
Contents
Investigation of Microclimate Parameters in the Industrial Environments . . . . . Solomiya Liaskovska, Olena Gumen, Yevgen Martyn, and Vasyl Zhelykh Detection of Defects in PCB Images by Separation and Intensity Measurement of Chains on the Board . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Roman Melnyk and Ruslan Tushnytskyy Tropical Cyclone Genesis Forecasting Using LightGBM . . . . . . . . . . . . . . . . . . . Sabbir Rahman, Nusrat Sharmin, Md. Mahbubur Rahman, and Md. Mokhlesur Rahman
xiii
448
458
468
Optimization of Identification and Recognition of Micro-objects Based on the Use of Specific Image Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Isroil I. Jumanov and Rustam A. Safarov
478
Development and Comparative Analysis of Path Loss Models Using Hybrid Wavelet-Genetic Algorithm Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ikechi Risi, Clement Ogbonda, and Isabona Joseph
488
Mathematical Advances and Modeling in Logistics Engineering Risk Assessment of Navigation Cost in Flood Season of the Upper Reaches of the Yangtze River Based on Entropy Weight Extension Decision Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jun Yuan, Peilin Zhang, Lulu Wang, and Yao Zhang Influencing Factors and System Dynamics Analysis of Urban Public Bicycle Projects in China Based on Urban Size and Demographic Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiujuan Wang, Yong Du, Xiaoyang Qi, and Chuntao Bai
503
514
Cross-border Logistics Model Design e-Tower Based on Blockchain Consensus Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shujun Li, Anqi He, Bin Li, Fang Ye, and Bin Cui
524
Research on the Service Quality of JD Daojia’s Logistics Distribution Based on Kano Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yajie Xu and Xinshun Tong
536
CiteSpace Based Analysis of the Development Status, Research Hotspots and Trends of Rural E-Commerce in China . . . . . . . . . . . . . . . . . . . . . . Yunyue Wu
547
xiv
Contents
The Optimization and Selection of Deppon Logistics Transportation Scheme Based on AHP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Long Zhang and Zhengxie Li Pharmacological and Non-pharmacological Intervention in Epidemic Prevention and Control: A Medical Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . Yanbing Xiong, Lijing Du, Jing Wang, Ying Wang, Qi Cai, and Kevin Xiong Passenger Flow Forecast of the Section of Shanghai-Kunming High-Speed Railway from Nanchang West Station to Changsha South Station . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cheng Zhang, Puzhe Wei, and Xin Qi
558
573
583
The Effect of Labor Rights on Mental Health of Front-Line Logistics Workers: The Moderating Effect of Social Support . . . . . . . . . . . . . . . . . . . . . . . . Yi Chen and Ying Gao
598
An Investigation into Improving the Distribution Routes of Cold Chain Logistics for Fresh Produce . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mei E. Xie, Hui Ye, Lichen Qiao, and Yao Zhang
608
Development of Vulnerability Assessment Framework of Port Logistics System Based on DEMATEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yuntong Qian and Haiyan Wang
618
Application Prospect of LNG Storage Tanks in the Yangtze River Coast Based on Economic Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jia Tian, Hongyu Wu, Xi Chen, Xunran Yu, and Li Xv
628
Multi-depot Open Electric Truck Routing Problem with Dynamic Discharging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xue Yang and Ning Chen
640
Analysis on the Selection of Logistics Distribution Mode of JD Mall in the Sinking Market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Weihui Du, Xiaoyu Zhang, Saipeng Xing, and Can Fang
651
Research on Cruise Emergency Organization Based on Improved AHP-PCE Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Long Zhang and Zhengxie Li
664
Contents
Fresh Agricultural Products Supplier Evaluation and Selection for Community Group Purchases Based on AHP and Entropy Weight VIKOR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gong Feng, Jingjing Cao, Qian Liu, and Radouani Yassine
xv
681
Logistics of Fresh Cold Chain Analysis of Joint Distribution Paths in Wuhan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Weihui Du, Donglin Rong, Saipeng Xing, and Jiawei Sun
696
Optimization of Logistics Distribution Route Based on Ant Colony Algorithm – Taking Nantian Logistics as an Example . . . . . . . . . . . . . . . . . . . . . . Zhong Zheng, Shan Liu, and Xiaoying Zhou
708
Optimization Research of Port Yard Overturning Operation Based on Simulation Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Qian Lin, Yang Yan, Ximei Luo, Lingxue Yang, Qingfeng Chen, Wenhui Li, and Jiawei Sun A Review of Epidemic Prediction and Control from a POM Perspective . . . . . . Jing Wang, Yanbing Xiong, Qi Cai, Ying Wang, Lijing Du, and Kevin Xiong
719
734
Application of SVM and BP Neural Network Classification in Capability Evaluation of Cross-border Supply Chain Cooperative Suppliers . . . . . . . . . . . . Lei Zhang and Jintian Tian
745
Research on Port Logistics Demand Forecast Based on GRA-WOA-BP Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zhikang Pan and Ning Chen
754
Evaluation and Optimization of the Port A Logistics Park Construction Based on Fuzzy Comprehensive Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xin Li, Xiaofen Zhou, Meng Wang, Rongrong Pang, Hong Jiang, and Yan Li
764
Advances in Technological and Educational Approaches OBE Oriented Teaching Reform and Practice of Logistics Information System Under the Background of Emerging Engineering Education . . . . . . . . . . Yanhui Liu, Jinxiang Lian, Xiaoguang Zhou, and Liang Fang Teaching Practice of “Three Integration” Based on Chaoxing Learning Software – Taking the Course of “Complex Variable Function and Integral Transformation” as an Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Huiting Lu and Xiaozhe Yang
777
787
xvi
Contents
Transformation and Innovation of E-Commerce Talent Training in the Era of Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lifang Su and Ke Liu Talent Training Mode Based on the Combination of Industry-Learning-Research Under the Background of Credit System Reform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shanyong Qin and Minwei Liu
801
811
Analysis of the Innovation Mechanism and Implementation Effect of College Students’ Career Guidance Courses Based on Market Demand . . . . Jingjing Ge
822
Comparative Study on the Development of Chinese and Foreign Textbooks in Nanomaterials and Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yao Ding, Jin Wen, Qilai Zhou, Li Liu, Guanchao Yin, and Liqiang Mai
833
Practical Research on Improving Teachers’ Teaching Ability by “Train, Practice and Reflect” Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jing Zuo, Yujie Huang, and Yanxin Ye
844
College Foreign Language Teacher Learning in the Context of Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jie Ma, Pan Dong, and Haifeng Yang
854
The Innovation Integration Reform of the Course “Single Chip Microcomputer Principle and Application” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chengquan Liang
865
Discussion and Practice on the Training Mode of Innovative Talents in Economics and Management in Women’s Colleges . . . . . . . . . . . . . . . . . . . . . . Zaitao Wang, Ting Zhao, Xiujuan Wang, and Chuntao Bai
876
Cultivation and Implementation Path of Core Quality of Art and Design Talents Under the Background of Artificial Intelligence . . . . . . . . . . . . . . . . . . . . Bin Feng and Weinan Pang
886
Reform and Innovation of International Logistics Curriculum from the Perspective of Integration of Industry and Education . . . . . . . . . . . . . . . Xin Li, Meng Wang, Xiaofen Zhou, Jinshan Dai, Hong Jiang, Yani Li, Sida Xie, Sijie Dong, and Mengqiu Wang An Analysis of Talent Training in Women’s Colleges Based on the Characteristics of Contemporary Female College Students . . . . . . . . . . . . Ting Zhao, Zaitao Wang, Xiujuan Wang, and Chuntao Bai
899
911
Contents
Solving Logistical Problems by Economics Students as an Important Component of the Educational Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nataliya Mutovkina Exploration and Practice of Ideological and Political Construction in the Course of “Container Multimodal Transport Theory and Practice” for Application-Oriented Undergraduate Majors—Taking Nanning University as an Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Shixiong Zhu, Liwei Li, and Zhong Zheng
xvii
921
931
A Study on Learning Intention of Digital Marketing Micro Specialty Learners Under the Background of New Liberal Arts—Based on Structural Equation Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yixuan Huang, Mingfei Liu, Jiawei You, and Aiman Magde Abdalla Ahmed
944
Comparisons of Western and Chinese Textbooks for Advanced Electronic Packaging Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Li Liu, Guanchao Yin, Jin Wen, Qilai Zhou, Yao Ding, and Liqiang Mai
954
Innovation and Entrepreneurship Teaching Design in Application-Oriented Undergraduate Professional Courses – Taking the Transportation Enterprise Management Course as an Example . . . . . . . . . . . Yan Chen, Liping Chen, Hongbao Chen, Jinming Chen, and Haifeng Yang
963
The Construction of University Teachers’ Performance Management System Under the Background of Big Data Technology . . . . . . . . . . . . . . . . . . . . Fengcai Qin and Chun Jiang
974
Curriculum Evaluation Based on HEW Method Under the Guidance of OBE Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Chen Chen and Simeng Fan
983
The Relevance of a Systematic Approach to the Use of Information Technologies in the Educational Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Nataliya Mutovkina and Olga Smirnova
995
Construction and Practice of “CAD/CAM Foundation” Course Based on Learning Outcome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1006 Ming Chang, Wei Feng, Zhenhua Yao, and Qilai Zhou Research and Practice of Ideological and Political Education in the Context of Moral Education and Cultivating People . . . . . . . . . . . . . . . . . . 1016 Geng E. Zhang and Liuqing Lu
xviii
Contents
A Quantitative Study on the Categorized Management of Teachers’ Staffing in Colleges and Universities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1028 Zhiyu Cui Course Outcomes and Program Outcomes Evaluation with the Recommendation System for the Students . . . . . . . . . . . . . . . . . . . . . . . . 1039 Khandaker Mohammad Mohi Uddin, Elias Ur Rahman, Prantho kumar Das, Md. Mamun Ar Rashid, and Samrat Kumar Dey Methodology of Teaching Educational Disciplines to Second (Master’s) Level Graduates of the “Computer Science” Educational Program . . . . . . . . . . . 1054 Ihor Kozubtsov, Lesia Kozubtsova, Olha Myronenko, and Olha Nezhyva Professional Training of Lecturers of Higher Educational Institutions Based on the Cyberontological Approach and Gamification . . . . . . . . . . . . . . . . . 1068 Oleksii Silko, Lesia Kozubtsova, Ihor Kozubtsov, and Oleksii Beskrovnyi Exploring the Perceptions of Technical Teachers Towards Introducing Blockchain Technology in Teaching and Learning . . . . . . . . . . . . . . . . . . . . . . . . . 1080 P. Raghu Vamsi Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1091
Advances of Computer Algorithms and Methods
AI Chatbots for Banks: Evolving Trends and Critical Issues Margherita Mori1(B) and Lijing Du2,3 1 University of L’Aquila (Retd), L’Aquila, Italy
[email protected]
2 School of Management, Wuhan University of Technology, Wuhan 430072, China 3 Research Institute of Digital Governance and Management Decision Innovation,
Wuhan University of Technology, Wuhan 430072, China
Abstract. The paper aims at providing a conceptual framework for analyzing evolutionary trends in the financial arena that have to do with banking on chatbots designed to take advantage of most recent developments in the domain of Artificial Intelligence (AI). Attention is focused on the positive effects that are associated with its applications and, at the same time, on critical issues that include potential risks and limitations, such as those faced by physically impaired bank customers. The starting point is an overview of how chatbots have changed the business areas of banking and financial services, as well as what can be expected in terms of future strategic shifts and behavioral changes of both banks and their customers. The next step is to conduct a more in-depth investigation of the role that AIpowered chatbots have begun to play - and are likely to continue to play - in the areas under scrutiny, particularly in the banking industry. Conclusions are based on success stories that should be replicated not only to improve customer experience and service support, but also to help modernize the banking industry, which is the most traditional subset of the financial sphere of the economy. Keywords: AI applications · Banking and financial services · Chatbots
1 Introduction The recent hype around ChatGPT and its competitors leads to shed unprecedented light upon cutting-edge technologies that keep accelerating the pace of significant developments in the financial arena and that include artificial intelligence (AI). Actually, technological advances have fueled financial innovation worldwide in the last decades: it has involved new types of financial firms (such as fintechs, that combine finance and technology), new financial products (such as digital payments and cryptoassets), and new ways of doing financial business (such as those reshaping banking into neobanking); as such, what seemed incredible a few months ago can be perceived as foundational, even in the historically change-resistant banking industry. Within this framework, AI applications have rapidly gained momentum. Among others, financial institutions have begun to use them to improve their customer experiences © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 3–13, 2023. https://doi.org/10.1007/978-3-031-36115-9_1
4
M. Mori and L. Du
by enabling frictionless, 24/7 interactions while saving money; certain AI applications have been widely disseminated in the financial industry, particularly by banks, which are thought to offer the greatest cost savings opportunities through front- and middleoffice applications. Chatbots, for example, enable two-way communication by allowing an online conversation via text or text-to-speech as a cost-effective alternative to direct contact with live human agents. These considerations pave the way for emphasizing the role of AI-powered chatbots in the financial arena, particularly in the banking sector, which is gradually becoming more self-service oriented in order to meet the needs of digitally savvy customers. Given the wider adoption of these first-response tools that greet, engage, and serve customers in a friendly and familiar way, it sounds appealing to overview the implied advantages and risks: therefore, this paper is aimed at providing a conceptual framework for analyzing the challenges that banks – as well as other financial institutions – have to take by embracing AI chatbots and by ultimately making conversational banking an automated process; conclusions draw upon success stories that are worth replicating not only to benefit those directly involved, but also to contribute to evolving trends in the financial sphere of the economy. The roadmap of this research is shown in Fig. 1. AI Chatbots for Banks: Evolving Trends and Critical Issues
1. Introduction
2. Insights into the State-of-the-Art 2.1 Drawing upon AI in the Financial Arena 2.2 Spotlight on Virtual Assistants and Chatbots 2.3 A Closer Look at AI-powered Chatbots
3. Illustrative Case Studies
4. Most Critical Issues for Discussion
3.1 Erica by Bank of America Corporation
5.Recommendati ons for Improved Use of Banking Chatbots
3.2 Cora by the Royal Bank of Scotland plc 3.3 Stella, Widdy and MArIO from Italy
Fig. 1. The roadmap of this research
2 Insights into the State-of-the-Art Due to the underlying cross-cultural issues, it seems appropriate to explore the most recent literature on several interrelated topics: on one hand, AI applications need to be investigated in order to devise those that are most useful to banks and other financial institutions; on the other hand, it is a matter of scrutinizing the distinguishing features of virtual assistants and chatbots. Finally, attention should be paid to how they can benefit from AI to better satisfy the needs on both the demand and the supply side.
AI Chatbots for Banks: Evolving Trends and Critical Issues
5
2.1 Drawing upon AI in the Financial Arena According to a shared view, AI is one of the technologies that can fundamentally change industries and banking makes no exception. Areas of major interest tend to be identified with investing, financial crime reporting, and complying with environmental, social and governance (shortly ESG) criteria: for example, AI can help banks to know their customers better and to provide them with suggestions that better align with their needs and preferences; algorithms can continuously analyze the portfolios of wealth management clients for risks while searching for the “next best offer” and can recommend a switch when its expected benefits exceed its costs [1]. As far as AI applications designed to help combating financial crime, they analyze transactions and record suspicious cases. For instance, various criteria are applied to each movement of capital, such as the amount, the currency, and the country to which the money either is going to or coming from: if a criterion does not match the typical patterns, the algorithm reports the anomaly to the account manager for further inquiry; as the feedback increases, the AI-powered model learns how to classify transactions correctly and ultimately reports only those that imply a real threat of a crime. Furthermore, the ability to quickly process large amounts of data makes AI models attractive to provide banks with the support they need to comply with the Taxonomy Regulation and delegated acts that have been adopted by the European Commission since 2020: these rules have requested European Union (EU) banks to sort out which transactions can be deemed “environmentally sustainable”, in an attempt at reducing greenwashing risks; in a few words, “the implementation of the Taxonomy signals an enhancement of mandatory sustainability reporting in the EU by driving capital towards activities that are ‘irrefutably’ green” [2]. Therefore, banks have to cope with a lot of new data from their corporate customers to properly classify their transactions and adhere to the evolving rules that fall within the scope of the climate and energy targets set by the EU for 2030 and that reflect its aim to be climate-neutral by 2050. 2.2 Spotlight on Virtual Assistants and Chatbots The increasing reliance on virtual offices – and to home banking – has initially shaped the recourse to virtual assistants, who are self-employed workers specialized in providing administrative services from a remote location: these independent contractors usually work from a home office and can access the necessary documents remotely; tasks that are typically performed by virtual assistants include scheduling appointments, making phone calls, and managing email accounts. However, the specific duties to be performed depend on the terms of each contract, that in some cases may involve clerical and bookkeeping tasks, while in others may be centered upon posting regular updates to social media or writing articles for a blog [3]. Intelligent virtual assistants (IVAs) look like the most conspicuous way in which AI has modernized so far – and can still upgrade – the financial system by adding “intelligent” features to the human version of these assistants and by ultimately contributing to the development of the so-called “intelligent banks” [4]: to make short a long story, IVAs are engineered entities residing in software that interfaces with humans in a human way; the embedded technology incorporates elements of interactive voice response and
6
M. Mori and L. Du
other modern AI applications to deliver full-fledged “virtual identities” that are enabled to converse with users [5]. As such, IVAs can perform tasks or provide services for an individual that are based on commands or questions and that may detect both intent and sentiment. IVAs tend to be compared to chatbots (as shown in Table 1), which deploy AI and natural language processing (NLP) to simulate human-like conversations with users via chat: the key task to be performed by chatbots is to understand user questions and automate responses to them with instant messages; in other words, chatbots are designed to make it easier for users to find the information they are searching for by responding to their requests and questions – through text input, audio input, or both – without the need for human intervention [6]. By way of comparison, due to their advanced natural language understanding and artificial emotional intelligence, IVAs can automate both complicated and repetitive tasks whereas chatbots mainly rely on rule-based conversational AI and tend to be associated with easier deployment [7]. Table 1. Comparison between virtual assistants and chatbots Virtual Assistants
Chatbots
Follow commands and complete daily tasks for users
Interact with people and answer their questions
Perform daily tasks, such as setting alarms, making phone calls, etc
Help people interact with websites and solve problems
Wide range of functions and terminology
Only understand some specific terms
More interactivity
Low interactivity
2.3 A Closer Look at AI-Powered Chatbots Actually, some AI-enhanced chatbots use complex algorithms that allow to provide more detailed responses: for instance, the deep learning capabilities of AI chatbots enable interactions to become more accurate over time, thus allowing to build a web of appropriate responses via the interactions between the chatbots themselves and humans; not occasionally conversational chatbots are programmed and trained in NLP, which enables them to facilitate a 2-way communication with the customers and their banks thanks to machine learning and AI. Accordingly, Conversational Banking is considered a smart way to boost customer loyalty by offering quick responses to clients’ queries and AI chatbots are thought of making up one of the most promising strategic areas for banks “to win the satisfaction vote of their loyal customers” [8]. By common opinion, AI chatbots bring about several benefits not only to the banks that rely on them but also to the banking sector as a whole: for instance, these strategic tools allow for improved service to customers by answering their questions round the clock, contribute to boost employee productivity by solving minor issues posed by customers, and can assist them with paying down debt, checking account status, and getting credit score insights; AI chatbots do not only serve customers but can also support
AI Chatbots for Banks: Evolving Trends and Critical Issues
7
employees, for instance to schedule meetings and send messages. Personalized marketing is another area of interest, as these chatbots can be used to suggest customer-tailored investment options and offer promotions based upon available customer’s data, such as the user’s location, preferences and interests [9]. However, as previously stated, “the success of banking chatbots will be effective when customers are satisfied with the chatbots and engage in using them,” and efforts have been made to investigate both the antecedents and consequences of customer brand engagement in using banking chatbots: the antecedents include interactivity, time convenience, compatibility, complexity, observability, and trialability; and the consequences deal with satisfaction with the brand experience. “Trialability, compatibility, and interactivity positively influence customer brand engagement through a chatbot, thereby influencing satisfaction with the brand experience and customer brand usage intention,” according to clear empirical evidence [10].
3 Illustrative Case Studies In sight of fully exploiting the potential of banking on AI chatbots, the financial arena – and particularly the banking industry – can be conceived as a huge laboratory with both physical and virtual features that provide useful insights. Interesting case studies emerge while surfing the web, since most banks have embraced the trend towards digitalization by setting up online facilities, even when traditional branches continue to be operating: therefore, these chatbots are getting part of the overall strategies developed within the banking industry and offer a promising avenue to both engage their actual customers and attract new ones; a look at the websites of major banks sounds like a reinforcement, as AI-powered chatbots are often deployed for connecting and communicating with customers, besides internally. 3.1 Erica by Bank of America Corporation (BofA) Not surprisingly, BofA is credited towards becoming the industry leader in the market segment populated with AI chatbots, thanks to Erica that was introduced in 2018 and improved in early 2023 to accentuate its human touch. Erica is marketed as personalized, proactive, and predictive, which has helped it become one of the most frequently used virtual banking assistants, assisting 32 million customers [11]: this is described as “a powerful virtual assistant that helps you stay on top of your finances” due to its ability to consider a variety of data within BofA, such as cash flows, balances, individual transaction history, and upcoming bills; for example, you may “be alerted when merchant refunds post to your account”, “receive bill reminders when payments are scheduled to be made”, and “review weekly updates on monthly spending”. Moreover, Erica is available 24/7 to help you make the most of your money: this virtual financial assistant provides insights and guidance to help put cash to work, assists customers with exploring investment strategies, and refers them to a specialist to discuss different opportunities; support is also provided to Merrill Investments and hence BofA customers who hold Merrill Investments accounts can resort to Erica to access quotes, track performances, place orders, as well as to get connected with Merrill advisors from
8
M. Mori and L. Du
the Mobile Banking app. However, Erica has been made exclusively available in the Mobile Banking, which may not allow to satisfy all needs, including those stemming from potential (rather than actual) customers. It is sufficient to “enter your U.S. mobile number and we’ll text you a link to download the Bank of America Mobile Banking app so you can get started,” though plans are in the works to provide services available through Erica in Online Banking in the future. As part of an ongoing improvement process, this AI bot can now use more methods to quickly answer customer questions, including a live chat with a BofA specialist: looking ahead, it appears relevant that Erica learns from customer conversations and has a dedicated team that is constantly working to expand the capabilities offered; to emphasize this point, “Erica is only available in English, but it is expected to learn Spanish” [12]. 3.2 Cora by the Royal Bank of Scotland plc (RBS) Heading to Europe, it was July 20, 2018 when RBS, part of the NatWest Group plc, posted on Facebook an introduction to Cora: it is a digital chatbot that has been continuously trained to provide instant answers to a quite large set of banking queries and that is “here to help you 24/7”, no matter whether customers are using the app, are logged into digital banking or are getting access to the institutional website [13]. As shown on it, both actual and potential customers are invited to “meet Cora, your digital assistant”. Several tasks can be performed thanks to Cora, that are listed together their technicalities: for instance, you can “change your details”, since “Cora can update your address and help you get your phone number, email address or name updated too”; you can also retrieve your “balance and transactions”, provided that “in your app or Digital Banking, Cora can securely see your spending to search your transactions”. This tool can even help if “you want a debit or credit card refund”, as the section centered upon “retail disputes” showcases. The “how to chat with Cora” section suggests to “keep your question simple and straight to the point” and provides a few examples: “I want to set up a standing order”, “How do I make a payment” and “How do I download a statement”; during the interaction, “Cora might ask you a few questions to make sure you get the right answer, and may bring in a human colleague to help if necessary”. Last, but not least, this chatbot is part of the marketing activities carried out by RBS, as Cora has been nominated for Innovation of the Year and potential users are encouraged to “vote for Cora at the British Bank Awards 2023” so as to get a chance to win £1000 in this Awards’ prize draw (though “T&C’s apply”, besides being this initiative reserved to “UK Residents” and “over 18s only”). 3.3 Stella, Widdy and MArIO from Italy Personification stands out as the underlying asset of several Italian banks that have adopted AI chatbots, such as Stella by Banca Sella Group, Widdy by Banca Widiba SpA and MArIO by UBI Banca (Unione di Banche Italiane SpA). Stella dates back to 2009 as a chatbot that was adopted to support customers engaged in home banking, and evolved in 2017 as an updated version that took advantage of a thorough analysis of the interactions recorded up to that point: this empirical approach has allowed to identify the
AI Chatbots for Banks: Evolving Trends and Critical Issues
9
areas that seem most convenient to focus on and has enabled to improve the quality of the responses that customers are provided with; in the pandemic era, Stella has helped to devise new areas that it could deal with, thus paving the way to the development of new skills, such as those regarding loans, in sight of better and sooner satisfying customers’ needs by both human and virtual service assistance [14]. Widdy is a conversational digital employee who is capable of sophisticated understanding of complex issues and who not only assists customers but also learns from these interactions: Widiba chose Teneo to build its intelligent online assistant, which has been in use since January 2014 to provide high-quality service and create a positive customer experience, and that can be easily updated with new knowledge and capabilities. Widdy can assist users with every step on the Widiba website, such as walking them through all stages of opening a new account step by step, and if users suspend their activity and return to it later, Widdy can recognize what steps have already been completed, allowing the account to be finalized without wasting the individual customer’s time. Widiba recently expanded and upgraded its use of the Teneo platform, and can now provide “predictive help” to customers who email the contact center via the website, using natural language understanding to automatically answer their questions even before they hit the send button [15]. As far as MArIO, it has been defined as the smartest banking bot in Italy [16]: MArIO stands for Migration Artificial Intelligence Operator and was adopted in 2017 in order to support the process developed by UBI Banca – a leading Italian retail bank, acquired by Intesa Sanpaolo SpA in 2020 – to integrate 3 other banks (Nuova Banca dell’Etruria e del Lazio SpA, Nuova Cassa di Risparmio di Chieti SpA and Nuova Banca delle Marche SpA); MArIO was the first AI chatbot deployed for internal customer support in an Italian bank and allowed for great user engagement (with about 4 thousand bank employees involved). Actually, lots of improvements have been achieved once in production, with the number of different answers grown from 89 to 227 and the number of topics recognized increased from 121 to 609, which led to consider extending MArIO bot for day-to-day customer support.
4 Most Critical Issues for Discussion Unquestionable empirical evidence surfaces, which leads to identify several reasons why AI-powered chatbots should be more and more widely adopted by banks but also to grasp critical issues to be addressed without delay, in sight of bridging the gap between supply and demand factors, as well as between theory and practice. Some of these issues stem from questions that sound all but surprising and that sometimes are mentioned in the FAQs, just as in the case of Erica: they include questions and answers such as is the chatbot secure and private? Yes, your interactions with it are protected by the same industry-leading privacy and security features as the mobile app and Online Banking. Can someone else but the customers activate the chatbot while using their phones? No, customers need to be authenticated through the app in order to be able to use the chatbot itself. Other issues can be referred to in terms of digital accessibility: on one hand, the chatbots under investigation tend not to be available to potential (versus actual) customers
10
M. Mori and L. Du
of the banks involved, which may confine them into a “missing out” situation, provided that struggling to retain their customer base should be combined with trying to enlarge it by attracting new customers; on the other hand, chatbots that are only available on digital banking apps exclude both actual customers who prefer to otherwise avail themselves of their digital banking facilities and potential customers who want to learn more of the targeted banks by accessing their institutional websites. Success stories include the Spanish-speaking virtual assistant that was launched in 2018 by Banpro (Grupo Promerica) – Nicaragua’s largest bank, headquartered in Managua – in cooperation with Canadian banking technology firm Finn.ai and that has been made available via Facebook Messenger, a channel reportedly being used daily at that time by 50% of Nicaraguans [17]. Critical issues may also surface with regard to diversity, equity and inclusion (or DE&I) as closely linked values that imply support to different groups of individuals, including people of different races, ethnicities, religions, abilities, genders, and sexual orientations. Actually, valuable opportunities may be missed by underscoring the needs that pertain to people with disabilities and that could be eventually satisfied by specific AI chatbots: with disability being part of the diversity conversation activity, it sounds like a must – rather than an option – to adapt these strategic tools to make them fully operational even in the case of special circumstances and situations, as it is the case when users are physically (for instance, visually) impaired or when they are affected by invisible illnesses, as it may happen in the case of neurodiversity; understanding neurodivergent people (just like those who are affected by a quite low stress tolerance) means embracing and being open to other ways of learning and collaborating, which leads to propose AI chatbots trained to support people with cognitive differences, that reflect natural variations in how the brain is wired. Above all, the major critical issues are illustrated in Fig. 2. Secure and private
Critical issues
Restrictions on digital accessibility Diversity, equity and inclusion
Fig. 2. The critical issues
5 Recommendations for Improved Use of Banking Chatbots Focusing on recommendations, relevant issues can be referred to linguistic minorities, such as Spanish speaking bank customers in several parts of the United States, who cannot interact with AI chatbots in English. No wonder that some of these tools can be trained to work in several languages, as shown by Nordea, the leading Nordic bank that
AI Chatbots for Banks: Evolving Trends and Critical Issues
11
covers Sweden, Denmark, Norway and Finland: this bank has been using several virtual agents in production on the Boost.ai platform, operating in each market’s local language (plus English in some countries); to this end, a comprehensive conversational AI strategy has been developed to scale customer service across the four markets served – for a total of over 9 million private customers and more than 500,000 active corporate customers – with a varied channel mix of customer service touchpoints (via in-person, phone, email and both live and automated chat) [18]. Improvements can also be made by solving some critical but frequent problems, since being smart – as AI-powered chatbots promise to be – should not allow for answers like “I did not understand” and banks cannot afford reputational risks, encompassing the risk of losing their credibility by relying upon poor performing chatbots: same questions may be written in many different ways, misspellings may occur not occasionally, items may be referred to with different names; moreover, questions may be very – even too – complex, or may be out of scope. Solutions include the ability to use partial understanding and to tell the user how to rephrase, as well as to recognize and be explicit on out of scope. It must also be acknowledged that the implementation of chatbot technology is evolving rapidly in the banking system, just like across all other industries, while customer acceptance is lagging behind: from a theoretical point of view, it has been highlighted “the importance of perceived compatibility and perceived usefulness in the adoption of banking chatbot technology”, with “awareness of the service” postively impacting “perceived ease of use” and “perceived privacy risk” and indirectly affecting “usage intention of banking chatbots through perceived usefulness”; moreover, “perceived ease of use influences perceived usefulness, and perceived compatibility has an effect on both perceived ease of use and perceived usefulness” [19]. Therefore, it makes sense to arrange promotional campaigns designed to encourage customers to rely on AI chatbots, as shown by Zenith Bank Plc – a large financial service provider in Nigeria and Anglophone West Africa – that has launched a marketing initiative between October 3 and December 30, 2022 to reward users of its Whatsapp Banking through ZiVA (Zenith Intelligent Virtual Assistant) “your personal banking assistant on WhatsApp” [20].
6 Conclusion To summarize, the successful implementation of AI-powered chatbots in the banking industry should be part of comprehensive strategies: these tools can assist in sensing, comprehending, acting, and learning, thus envisioning a system that can perceive the world around them, analyze, and understand the information they receive, take actions based on that understanding, and improve their own performance by learning from what happened; recommended strategies have the potential to support and drive growth, in sight of a more sustainable and inclusive financial industry. Several trends in digital engagement have accelerated during the corona pandemic, and the bleak picture to be confronted should push all financial institutions to take advantage of AI technology by fully utilizing its potential, particularly in providing distinct customer experiences. Needless to say, AI is unlikely to replace skilled human workers, and banks, by the way, require a dedicated team to manage and improve their chatbots: because computational algorithms lack the ability to reason cognitively, it appears that humans will
12
M. Mori and L. Du
always play a role in the emotional aspects of bank-customer relationships; at the same time, human agents are required to monitor the training of AI models to ensure that banks do not run the risk of providing unsuitable solutions to their customers or, worse, that the models do not create unintentional bias against certain types of consumers. As a result, as technology advances, AI can be expected to augment – but not replace – the human elements of the banking relationship. All in all, customer demand for both digital banking and human interaction requires banks to find the right balance between them, not only to preserve their market shares but also to try to increase them. To this end, making AI chatbots available to some extent to potential customers (besides the full operability reserved to actual ones) may prove a win-win strategy: success stories sound like a reinforcement to acknowledge the key role that these tools can play, though critical issues should not be underestimated; likewise, recommendations stemming from use cases developed should be carefully checked and eventually complied with, in order to fully exploit the potential of AI applications in the market segment under scrutiny, thereby contributing to shape sustainable pathways to prosperity.
References 1. Bremke, K., Cavus, M., Mindt, M., et al.: How AI is changing banking. Frankfurt am Main: Deutsche Bank AG (2023). https://www.db.com/what-next/digital-disruption/betterthan-humans/how-artificial-intelligence-is-changing-banking/index. Accessed 13 Feb 2023 2. Pettingale, H., de Maupeou, S., Reilly, P.: EU Taxonomy and the Future of Reporting. Harvard Law School Forum on Corporate Governance. Harvard College, Cambridge, MA (April 4, 2022). https://corpgov.law.harvard.edu/2022/04/04/eu-taxonomy-and-the-future-of-report ing/. Accessed 13 Feb 2023 3. Will, K.: What is a Virtual Assistant, and What Does One Do?. Investopedia, New York, NY. https://www.investopedia.com/terms/v/virtual-assistant.asp. Accessed 15 Feb 2023 4. Jubray, E., Graham, T., Ryan, E.: The Intelligent Bank - Redefining Banking with Artificial Intelligence. Accenture, Dublin (2018). https://www.accenture.com/t00010101t000000z__ w__/gb-en/_acnmedia/pdf-71/accenture-banking-aw-jf-web.pdf. Accessed 19 Feb 2023 5. Intelligent Virtual Assistant. Techopedia. Edmonton, AB: Janalta Interactive (June 28, 2022). https://www.techopedia.com/definition/31383/intelligent-virtual-assistant. Accessed 15 Feb 2023 6. What is a chatbot? Armonk, NY: IBM. https://www.ibm.com/topics/chatbots. Accessed 15 Feb 2023 7. Dilmegani, C.: Chatbot vs Intelligent Virtual Assistant; Use Cases Comparison. AIMultiple, Tallin (January 20, 2023). https://research.aimultiple.com/chatbot-vs-intelligent-virtual-ass istant/. Accessed 15 Feb 2023 8. Sayiwal, S.: Chatbots in Banking Industry: A Case Study. Journal of Emerging Technologies and Innovative Research (JETIR) 7(6), 1498–1502 (2020). https://www.jetir.org/papers/JET IR2006221.pdf. Accessed 19 Feb 2023 9. Shashank, B.R., Rashmi, R.: A Review of Chatbots in the Banking Sector. Int. J. Eng. Res. Technol. (IJERT), 10(06), 428–430 (2021). https://www.ijert.org/research/a-review-of-cha tbots-in-the-banking-sector-IJERTV10IS060203.pdf. Accessed 15 Feb 2023 10. Hari, H., Iyer, R., Sampat, B.: Customer brand engagement through chatbots on bank websites – examining the antecedents and consequences. Int. J. Human-Comp. Interac. 38(13), 1212–1227 (2022). https://doi.org/10.1080/10447318.2021.1988487
AI Chatbots for Banks: Evolving Trends and Critical Issues
13
11. McNamee, J.: Bank of America adds a human touch to its virtual assistant, Erica. Insider Intelligence. NY, NY: Insider Inc (December 14, 2022). Erica becomes a little more human Insider Intelligence Trends, Forecasts & Statistics. Accessed 16 Feb 2023 12. Personalized. Proactive. Predictive. Charlotte, NC: Bank of America Corporation (2023). https://promotions.bankofamerica.com/digitalbanking/mobilebanking/erica. Accessed 16 Feb 2023 13. Meet Cora – Here to help you 24/7. Edinburgh: The Royal Bank of Scotland plc (2023). https://www.rbs.co.uk/support-centre/cora.html. Accessed 15 Feb 2023 14. Gaudiuso, R.: Intelligenza artificiale e assistenti virtuali: come cambia la relazione con i clienti. Sella Insights. Biella: Banca Sella Holding SpA (February 21, 2022) (in Italian). https://sellainsights.it/-/intelligenza-artificiale-e-assistenti-virtuali-come-cambia-la-rel azione-con-i-clienti. Accessed 16 Feb 2023 15. Walker, T.: Widdy (Widiba). Amsterdam: Chatbors.org (2023). https://www.chatbots.org/cha tbot/widdy/. Accessed 16 Feb 2023 16. Vitale, A., Boido, M.: The Story Behind the Smartest Banking Chatbot in Italy!, Chatbot Summit, Milan (January 31, 2018). https://www.slideshare.net/pmontrasio/il-pi-intelligentechatbot-bancario-in-italia. Accessed 16 Feb 2023 17. Banpro and Finn.ai roll out virtual banking assistant in Central America. Retailer Banker International (March 7, 2018). https://www.retailbankerinternational.com/news/banpro-finnai-roll-ai-powered-virtual-banking-assistant-central-america/. Accessed 18 Feb 2023 18. Case Study Nordea. Stavanger: Boost.ai (2023). https://www.boost.ai/case-studies/nordeaemploys-comprehensive-conversational-ai-strategy-to-scale-customer-service. Accessed 18 Feb 2023 19. Alt, M., Vizeli, I., Sapl ˘ acan, ˘ Z.: Banking with a chatbot – a study on technology acceptanc. Studia Universitatis Babe¸s-Bolyai Oeconomica 66(1), 13–35 (2021). https://doi.org/10.2478/ subboec-2021-0002 20. Terms and Conditions. Lagos: Zenith Bank Plc (2022). https://www.zenithbank.com/images/ TERMS_AND_CONDITIONS.pdf. Accessed 18 Feb 2023
Traffic Flow Characteristics and Vehicle Road Coordination Improvement in Subterranean Interweaving Enshi Wang1(B) , Bihui Huang1 , and Bing Liu2 1 CCCC Second Expressway Consultant Co. Ltd., Wuhan CCCC Traffic Engineering CO. Ltd.,
Wuhan 430050, China [email protected] 2 School of Transportation and Logistics Engineering, Wuhan University of Technology, Wuhan 430063, China
Abstract. Subterranean interchange accidents are frequent, which frequently occur in weaving segments. Among them, rear-end collisions are their main types of accidents, and rear-end collisions are one of the typical phenomena of traffic instability. In view of this, this project adopts the method of empirical analysis to obtain the characteristics of traffic flow in the subterranean interchange weaving segment. Through the Lyapunov index to determine its stability, it is found that the stability of the weaving segment is significantly reduced, resulting in multiple rear-end collision accidents. The main causes of instability are: The limited space for the layout of the traffic signs and the lack of GPS information lead to inadequate access to the driver’s route information, which makes it easier to make mistakes in decision-making, emergency braking, and frequent sudden lane changes, and imposes random strong interference on the traffic flow. Based on this, the project improves traffic flow stability by reducing random strong interference: Pushing position information to the navigation system through the Beacon base station to achieve self-position information compensation, inducing the drivers to change lanes reasonably through the driving instructions, to reduce random strong interference, and at the same time rerouting and pushing routes for real-time relocation of vehicles. Keywords: Traffic safety · Subterranean interchange · Weaving segment · Traffic flow stability · Information compensation · Vehicle road coordination
1 Introduction As the urban underground road gradually forms a network, a large number of subterranean interchanges will be formed, and the subterranean interchange will be the key node of the road network, which will affect the operation efficiency of the overall road network, and its traffic safety issues need more attention. According to statistics, the annual average traffic accident of Xiamen Wanshishan-Zhonggushan Mountain subterranean interchange is as high as 42, and the proportion of rear-end accidents is 94% © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 14–24, 2023. https://doi.org/10.1007/978-3-031-36115-9_2
Traffic Flow Characteristics and Vehicle Road Coordination Improvement
15
[1]. Weaving segment is the most common accidents area, accounting for 42.3% of the accidents. At present, the safety of the weaving segments on the ground has been attached much importance. Many scholars have proposed improvement programs from the aspects of ramp control, driver characteristics, traffic flow characteristics, etc. For example, Aries van Beinum mainly revealed that lane changes caused by merging and diverging vehicles create most turbulence [2]; Mohamed Abdel-Aty’s research showed the reduction of upstream speed is the key to improve the safety of weaving segment and lower upstream variable speed limit would better improve the safety of the whole weaving segment [3]; David Sulejic proposed a optimization technique for the lane- changing concentration [4]; Ling Wang studied crash likelihood using real-time crash data with the objection, identified hazardous conditions, but the above research mainly focuses on the ground [5]. Different from the ground, the underground environment has obvious disadvantages in terms of light, line shape and information conditions. It is more likely to interfere with traffic flow, and the system stability is poor, which in turn increases the accident frequency and intensity. Therefore, the underground environment needs further exploration. Traffic flow stability is an important indicator to measure the anti-jamming capability of the system [6, 7]. According to relevant research, improving the stability of the traffic flow system can significantly improve the traffic safety level [8, 9]. On this basis, research is carried out to obtain the driving behavior characteristic data of subterranean interchange weaving segment through real vehicle experiment and surveillance video. The stability is judged by Lyapunov exponent, and the traffic loss is caused by the characteristics of underground traffic environment [10]. The Beacon base station disposed on the road side is used to construct a navigation system to provide position information for the drivers, improve the stability of the traffic flow from the perspective of vehicle road coordination, and improve the traffic safety level of the subterranean interchange. This article is divided into five parts. In the second part, through the real vehicle experiment and monitoring video, acquire the vehicle characteristics. The third part analysis collected data and draw some conclusions. The fourth part proposes improvement measures, that is, using iBeacon technology to build a navigation system to improve traffic flow stability. Finally, the fourth part summarizes and discusses the results of this paper.
2 Experiment and Result In the research, we got the trajectory, speed and space headway data through two ways: vehicle experiment and Monitor Video Recognition. 2.1 Vehicle Experiment Carry out vehicle experiments on the road at the of Wanshishan and Zhonggushan underground intersection in Xiamen from August 4, 2017 to August 12, 2017. This intersection is the first subterranean interchange project in China. The subterranean interchange tunnels consist of 7 tunnels on the left and right of Wanshishan Tunnel, A and B tunnels of Zhonggushan Tunnel and three ramps A, B and C connecting the upper and lower
16
E. Wang et al.
tunnels. The tunnels have six bifurcations, one parallel section, four through sections, 15 kinds of tunnel clearance sections, the maximum excavation depth is 26.4 m and the excavation height is 15.4 m. It is designed according to the urban expressway standard 2 lanes, the design speed is 60 km/h, the single tunnel net width is 11.5 m, the net height is 5.0 m, the left tunnel is 2850.5 m, the right tunnel is 2805 m. The space inside the ramp of the interchange is designed 8.5 m according to the width of the single lane, and the design speed is 30 km/h. The A ramp is annular and plane curve radius is 100 m. Figure 1 is the subterranean intersection. 1) A ramp: Zhonggushan Tunnel from south to east to Wanshishan Tunnel right turn ramp. 2) B ramp: Wanshishan Tunnel from east to north to Zhonggushan Tunnel right turn ramp. 3) C ramp: Zhonggushan Tunnel from north to east to Wanshishan Tunnel right turn annular ramp.
Fig. 1. Experimental location
The experiment selected 5 men and 2 women drivers as the experimenters, driving for more than 3 years as same as less than 3 years. Table 1 shows 5 periods to collect traffic flow statistics in different draffic characteristics and visual environments. Table 2 is the equipment and collected data. The choice of experimental path is to avoid drivers from taking repeated paths and eliminating adaptability. Figure 2 shows the experiment path. Firstly, starting from the A tunnel, drove along the A-c-C-d path, drove through the diversion into the blue ramp and out of the d-tunnel; then, drove along the d-D-b-B-d-D-c path, halfway through the Yellow ramp to the main line after the convergence of c-tunnel. Finally, it traveled along the c-C-d-D-b-B-c-c-a path, crossed the shunt lane into the red ramp, then ran out of the c-tunnel, after drove through the C portal, it returned to a portal along the C-a path to complete the operation of a loop. The time required to complete an annular is nearly 1 h, a driver need to complete six times a day of the above six circuits.
Traffic Flow Characteristics and Vehicle Road Coordination Improvement
17
Table 1. Five periods to carry out the experiment Number
Time
Traffic Flow Character
1
7:00 ~ 9:00
Morning Rush Hour
2
9:00 ~ 11:00
Morning Flat Peak
3
15:00 ~ 17:00
Afternoon Flat Peak
4
17:00 ~ 19:00
Evening Rush Hour
5
21:00 ~ 22:30
Free Flow
Table 2. Specific collected data Number
Component
Collected Data
1
Mobileye ME630 System
Lane Departure, Headway, Front Vehicle Speed
2
Vehicle OBD Data Parsing
Steering Wheel Angle and Speed, Speed, Throttle, Brake Pedal
3
RT2500 Inertial Navigation System
Vehicle Coordinates, Trajectory
4
Illuminance Instrument
Visual Environment
5
Tachograph
Traffic Video Recording
Fig. 2. Experiment path
2.2 Monitor Video Recognition Through MATLAB image recognition programming, analyze the monitor video of the test section. And through the background-based target detection method, target recognition and tracking of vehicles were made to finally get the vehicle trajectory data.
18
E. Wang et al.
2.3 Trajectory Data Analysis Figure 3 is the trajectory characteristics of the vehicle at the diversion point combined with real vehicle experiments and monitoring video. The curve represents the vehicle trajectory, and the bar graph represents the vehicle occupancy. Among them, up to 26.7% of the vehicles suddenly change lanes 50 m before the split nose, and bring great obstruction to the traffic flow.
Fig. 3. Vehicle trajectory
3 Further Analysis 3.1 Space Headway Analysis According to the analysis of the experimental data, the relationship between the actual space headway(S) and stopping sight distance (ST) is shown in Fig. 4. Among them, up to 86% of the space headway is smaller than the safe distance. Overall, in the subterranean interweaving section, the space headway is close, and the risk of following the vehicle is large.
Fig. 4. Space headway and stopping sight distance
The maximum Lyapunov exponent in chaos theory is used to judge the stability of traffic flow [11]. If the index is positive a, the system is unstable, and when it is negative, the system is stable. Taking the time-varying data of the vehicle as the input, using the wolf reconstruction method and being calculated by MATLAB, the result is shown in Fig. 6. It can be seen that the Lyapunov exponent of space headway is greater than 0, so the system is unstable.
Traffic Flow Characteristics and Vehicle Road Coordination Improvement
19
Fig. 5. Traffic accident type distribution
Fig. 6. Lyapunov Index of space headway
According to analysis the space headway and the traffic accident (Fig. 5), we found that the stability of the interweaving section system was poor, and the space headway was significantly lower than the safety distance, resulting in frequent rear-end collisions. Further analysis of the reasons for the small space headway, by consulting the relevant literature, we found the following explanation: Saffarian pointed out that when the visibility is extremely low, the driver will tend to reduce the headway to ensure that the front car is within the visible distance [12]. Leeuwen studied the effects of vertical information loss on driving behavior through perceptual occlusion paradigms and simulation experiments [13]. It was found that when the driver only presented the near-end visual region (similar to the underground intercommunication feature), the driver’s gaze point was even more Focusing on the front car, it is closer to the car. It is also found through driving simulation studies that when the visibility is very low, the driver usually chooses a smaller headway. 3.2 Speed Analysis The experimental data was collated and analyzed, and the overspeed ratio was obtained as shown in Fig. 7. According to the analysis we found that the overspeed ratio was higher than 20% in each period, so the speed risk was high. Calculating the Lyapunov exponents of the space headway, the value of each time period was greater than 0, so the system was in an unstable state (see Fig. 8).
20
E. Wang et al.
Fig. 7. Overspeed ratio
Fig. 8. Lyapunov Index of speed
It was found that overspeed vehicles accounts for higher than 20% in each period. Hogendoorn’s and Champion’s research showed that the low contrast of visual stimuli leads to Drivers’ underestimating the speed of the car, which results in overspeed [14, 15]. Therefore, the low illuminance of the subterranean environment and the high concentration of pollutants can be the reason of overspeed phenomenon. 3.3 Result Analysis Through the trajectory analysis, it was found that the emergency braking and sudden lane change occurred frequently in the subterranean interchange. According to the carfollowing theory, the emergency braking and sudden lane change of the preceding vehicle can generate strong stimulation to drivers of following vehicles, which leads to drivers’ drastic driving behaviors [16]. Those behaviors destroy the stability of the system. When the system is unstable, the emergency braking or sudden lane change behavior of the preceding vehicle can cause the disturbance to propagate in the reverse direction of traffic, increasing the probability of the rear-end collision. Further analysis of the subterranean environment, the reasons for this phenomenon are as follows: (1) The horizon is limited in subterranean environment, resulting in
Traffic Flow Characteristics and Vehicle Road Coordination Improvement
21
inadequate signs and information. As a result, it is difficult to effectively guide the driving route and drivers unfamiliar with the road cannot change lanes in advance; (2) GPS in the subterranean space doesn’t work, so vehicles that rely on navigation cannot obtain effective navigation information, and it is difficult to determine an accurate driving route; (3) Low subterranean space illumination and high concentration of pollutants decrease the contrast of visual stimuli in the space, so drivers cannot identify traffic signs in time and eventually miss the best lane change position.
4 Solution 4.1 Some Solutions and Their Disadvantages In traditional navigation, GPS positioning, WIFI positioning, base station positioning, etc. are usually used. In the subterranean tunnel, GPS positioning is invalid because of the lack of GPS signals. The WIFI positioning works well for indoor positioning, but it is expensive and cannot be used independently and permanently. The difficulty of subterranean environment positioning has led people to resort to other ways to obtain route information, mainly through two. One is the inferred navigation, that is, the inertial navigation. The principle is to collect the acceleration of the moving object and automatically perform the integral operation to obtain the instantaneous velocity and instantaneous position data of the moving object for navigation. However, since the sampling frequency is generally high (tens or even hundreds of times in one second), the error is easily to accumulate, so inertial navigation is only suitable for use in a short time; the other is to obtain information through the traffic signage. However, due to the limited subterranean space and signage layout, it is unable to provide sufficient traffic information. 4.2 Introducing IBeacon Technology In order to solve the above problems, we designed roadside facilities to meet the driver’s information needs, that is, road coordination. The iBeacon technology was chosen, which is a low-energy Bluetooth technology that works like the previous Bluetooth technology [17]. The Beacon base station continuously transmits signals in an area of tens of meters, and the device receiving the signal can respond accordingly. UUID, minor, and major are some of the parameters in iBeacon signal packets. They are used to identify Beacon devices and pass information [18]. The iBeacon positioning was used for indoor navigation and indoor preferential information push because of the small consumption of Beacon equipment, but it has not been involved in subterranean navigation. In this research, this technology was used to improve the navigational dilemma in places where GPS doesn’t work like subterranean tunnels. 4.3 Key Problems The specific method is as follows: the subterranean navigation module based on the iBeacon technology sends the location information by the Beacon base station, and
22
E. Wang et al.
the drivers’ device pushes the driving route in real time after receiving the location information to realize the compensation of the route information in the subterranean environment (as shows in Fig. 9). To achieve this goal, a set of methods was designed to convert the received beacon information into navigation instructions to be presented to the driver. In order to remind the lane change and turn in advance, the Beacon base station was arranged at a certain distance before the lane change line and the intersection. After the device obtains the position information, voice prompt is performed to realize the effect of promptly changing the lane change.
Fig. 9. How does the Beacon work
Another key issue is the placement to place beacon base stations, which affects the accuracy and effectiveness of navigation. Flat layout: The base station was divided into two types by their function: positioning base station and broadcasting base station. The positioning base station was disposed every 20 m for path navigation, and the distance between broadcasting base station and intersection, S was calculated from the distance formula S = V85% ∗ T . T is the time a navigation activity cost. For example, the characteristic driver’s perception reaction time and braking time total 0.7s. In addition, the voice prompt time is 1.5 s (considering the broadcast speed and content), so T is 2.2 s. Combined with operating speed as 50 km/h(13.9/s), S is 31.97 m. Therefore, it is laid at a distance of 30 m from the intersection to provide the driver with voice prompt (see Fig. 10). Section layout: Considering the signal coverage of the Beacon base station, for the two-lane section, the Beacon base station was placed at the top of the tunnel; for the three-lane section, Beacon base stations was placed on both sides of the tunnel to ensure the effect (see Fig. 11).
Fig. 10. Flat layout
A test has been carried out to evaluate the performance of the Beacon system. Beacon interval of broadcasting was set as 100 ms and Beacon base station covered an area of 20 m [19]. The experimenter drove toward Beacon base station from 30 m away and recorded the period from the time that the car 50 m was away from Beacon to the time
Traffic Flow Characteristics and Vehicle Road Coordination Improvement
23
Fig. 11. Section layout
that the signal was received. Finally, the average interval was 0.2 s, negligible compared with the time one navigation activity cost.
5 Conclusion In the above study, through the analysis of the data, it was concluded that up to 26.7% of the vehicles suddenly changing lane 50 m before the split nose. By analyzing the environment of the weaving segments, it was proposed that the lack of route information results in the driver unable to reasonably change the lane in advance, which leads to accidents. In the actual environment, the driver’s own conditions, such as psychological load, driving experience, road line type and traffic marking settings, may affect the occurrence of traffic accidents, this will be the goal of future research. In terms of solutions, the above application for iBeacon technology is not yet mature. In order to achieve good results, there must be a unified communication protocol standard to represent various traffic information, and the accuracy and post-care of Beacon base stations require further research. In addition, there are many short-range wireless communication technologies, such as UWB technology, Z-Wave technology, and wireless Mesh network. It is also the direction of the next step to test which one is more suitable for the compensation of underground intercommunication information.
References 1. Li, X., Yan, X., et al.: A rear-end collision risk assessment model based on drivers’ collision avoidance process under influences of cell phone use and gender—A driving simulator based study[J]. Accid. Anal. Prev. 97, 1–18 (2016) 2. van Beinum, A., Farah, H., et al.: Driving behaviour at motorway ramps and weaving segments based on empirical trajectory data. Transportation Research Part C: Emerging Technologies 92, 426–441 (2018) 3. Abdel-Aty, M., Wang, L.: Implementation of variable speed limits to improve safety of congested expressway weaving segments in microsimulation. Transportation Research Procedia 27, 577–584 (2017) 4. Sulejic, D., Jiang, R., et al.: Optimization of lane-changing distribution for a motorway weaving segment. Transportation Research Procedia 21, 227–239 (2017) 5. Wang, L., Abdel-Aty, M., et al.: Real-time crash prediction for expressway weaving segments. Transportation Research Part C: Emerging Technologies 61, 1–10 (2015)
24
E. Wang et al.
6. Derbel, O., Peter, T., et al.: Modified intelligent driver model for driver safety and traffic stability improvement. IFAC Proceedings Volumes 46(21), 744–749 (2013) 7. Peng, G., Liu, C., et al.: Influence of the traffic interruption probability on traffic stability in lattice model for two-lane freeway. Physica A 436, 952–959 (2015) 8. Mhaskar, P., El-Farra, N.H., et al.: Stabilization of nonlinear systems with state and control constraints using Lyapunov-based predictive control. Syst. Control Lett. 55(8), 650–659 (2006) 9. Haddad, J., Geroliminis, N.: On the stability of traffic perimeter control in two-region urban cities. Transportation Research Part B: Methodological 46(9), 1159–1176 (2012) 10. Lee, S., Heydecker, B.G., et al.: Stability analysis on a dynamical model of route choice in a connected vehicle environment. Transportation Research Part C: Emerging Technologies 94, 67–82 (2018) 11. Johansson, M., Rantzer, A.: Computation of piecewise quadratic lyapunov functions for hybrid systems. European Control Conference (ECC) (1997). https://doi.org/10.23919/ecc.1997.708 2399 12. Saffarian, M., Happee, R., et al.: Why do drivers maintain short headways in fog? A drivingsimulator study evaluating feeling of risk and lateral control during automated and manual car following. Ergonomics 55(9), 971–985 (2012) 13. van Leeuwen, P.M., Happee, R., et al.: Vertical field of view restriction in driver training: A simulator-based evaluation. Transport. Res. F: Traffic Psychol. Behav. 24, 169–182 (2014) 14. Hogendoorn, H., Alais, D., et al.: Velocity perception in a moving observer. Vision. Res. 138, 12–17 (2017) 15. Champion, R.A., Warren, P.A.: Contrast effects on speed perception for linear and radial motion. Vision. Res. 140, 66–72 (2017) 16. Wagner, P.: Analyzing fluctuations in car-following. Transportation Research Part B: Methodological 46(10), 1384–1392 (2012) 17. Varsamou, M., Antonakopoulos, T.: A bluetooth smart analyzer in iBeacon networks. In: 2014 IEEE Fourth International Conference on Consumer Electronics Berlin (ICCE-Berlin), pp. 288–292. IEEE (2014) 18. Lin, X.-Y.: A mobile indoor positioning system based on iBeacon technology. Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Annual Conference, 4970–4973 (2015) 19. de Henares, A.: Towards accurate indoor localization using iBeacons, fingerprinting and particle filtering. International Conference on Indoor Positioning and Indoor Navigation (IPIN) 2016, 4–7 (2016)
Influential Factors and Implementation Path of Talent Digital Evaluation Based on ISM Model: Taking Electric Power Enterprises as an Example Wei Luo, Jiwei Tang(B) , Saixiao Huang, and Yuan Chen CHN Energy Dadu River Big Data Services Co., Ltd., Chengdu 610000, China [email protected]
Abstract. The digital talent evaluation index system of electric power enterprises is usually affected by multiple factors. Clarifying the relationship between various factors is of great significance for the digital transformation and upgrading of enterprises and talent storage and allocation. Therefore, based on the ISM model, this paper constructs the adjacency matrix with the power enterprise as the representative framework, calculates the accessibility matrix with the help of Matlab programming, and finally analyzes the implementation path of the impact factor. In the study, the factors that affect the digitalization of enterprise talents are summarized into 20 sub - factors, which are divided into six progressive levels: top level factor, core factor, critical factor, hub factor, key factor, and bottom - level factor according to the hierarchical factors. This research based on hierarchical structure can provide guidance and reference for the future digital evaluation of enterprise talents. Keywords: Power transmission enterprises · Digitization · Talent characteristics · ISM model · Accessibility matrix
1 Introduction The digital economy has become an essencial driving force for the transformation and upgrading of industrial structures and the promotion of high-quality economic development [1–3]. According to the White Paper on China’s Digital Economy Development (2021), the scale of China’s digital economy has reached 39.2 trillion yuan, with an annual growth rate of 9.7% [4]. As the mainstay of the national economy, the digital construction of state-owned enterprises is crucial to the innovative development of the national economy [5]. In the report of the 20th National Congress of the Communist Party of China, the General Secretary pointed out that we should speed up the construction of enterprise digitalization, promote the deep integration of digital economy and the real economy, and build a digital industrial cluster with international competitiveness [6]. The digital transformation and upgrading of enterprises is an inevitable requirement for the implementation of the decision and deployment of the CPC Central Committee, and also a new concept, strategy and measure for talent construction in the new era [7, 8]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 25–34, 2023. https://doi.org/10.1007/978-3-031-36115-9_3
26
W. Luo et al.
To deepen the talent supply - side structural reform, the construction of talent digitalization is an essential basis for improving the innovation ability of enterprises [9, 10]. In the context of the digital economy, the information base of the digital economy has impacted the development of the entire industry, resulting in “disruptive” innovation and the subversion of the organizational structure and business model [11–13]. On the one hand, information exchange, data flow and knowledge diffusion pose new challenges to the construction of enterprise human resources; On the other hand, in the digital wave, how to better select, cultivate, evaluate and motivate employees through human resource management practice has become a problem that enterprise managers need to consider [14]. Although the digital talent construction of enterprises has been put forward as an important strategy at the national level, from the content of the study, most of the research only stays at the level of policy formulation, or a more macro level of industry development, while ignoring the theoretical level of digital talent realization mechanism and logical analysis [15]. In particular, there is a lack of talent in digital model research focusing on specific industries and specific industrial types [16]. In addition, as a unique resource of the enterprise, human resources itself is complex and have an important impact on the business activities of the enterprise [17]. Therefore, the research on the digital evaluation index system of talents needs to consider not only the general ability items of skills, but also the professional ability items of talents, and build a systematic and comprehensive analysis framework [18]. Based on the above considerations, this paper takes the electric power industry as the representative to explore the digital talent impact factor set in the electric power industry, and uses the Interpretative Structural Model theory to explore the impact factors of the digital talent evaluation of power transmission enterprises and research on the implementation path, providing countermeasures and suggestions for building a reasonable digital talent evaluation standard in the electric power industry.
2 Construction of Digital Influence Factor Set for Talents in Power Transmission Industry According to the theory of system dynamics and game theory, the transformation of economic and social systems is the result of the interaction of complex elements, and the transformation and upgrading of systems are usually closely related to policy environment, market environment, technological development and other factors. Abroad, Afha and others analyzed the development process of European power system based on socio - technical theory [19]. Wang Jifeng and others pointed out that the application of information technology has always played an essential role in promoting the development of China’s power system, and the digitalization of power system runs through the whole process of production, operation and sales of power enterprises [20]. In fact, the process of digital transformation of electric power enterprises is essentially the transformation and upgrading of labor force based on the digitalization of talents, which is affected by the knowledge, ability and attitude of employees [21]. Factor analysis is one of the important methods [22, 23]. As a consequence, to determine the critical factors of talent digitalization in power enterprises, this part uses the literature research method and expert consultation method to sort out the factors that affect talent digitalization. The specific analysis contents are shown in Table 1.
Influential Factors and Implementation Path of Talent Digital Evaluation
27
Table 1. Influential factor set of talent digitalization Level I indicators
Level II indicators Level III indicators
Level I indicators
Level II indicators
Level III indicators
Knowledge and skills
Compliance knowledge
Ability
Innovation
Awards S15
Legal knowledge S1 Rules and regulations S2
Patents/Property S16
Network risk awareness S3 Managerial skills
Executive force
Performance appraisal S17
Loyalty
Personal honor S18
Responsibility
Annual performance S19
Training S6
Execution effect
Performance result S17
ERP application S7
Service awareness
Ratio of service objects S20
Computer operation level S8
Compressibility
Psychological resilience S21
Aggressiveness
Part-time work S22
Major S4 Degree S5
Informatization skills
Ability
Attitude
Communication ability
Cross department work S9
Writing ability
Press Publicity S10
Professional judgment
Risk control ability S23
Official writing S11
Refine on
Labor awards S24
Thesis and books S12
Technical Writing Writing technical manuals S25
Learning ability
Professionalism
Professional Certificate S13 Highest education S14
In order to improve the reliability and validity of digital talent evaluation elements, 30 experts were invited to screen 25 factors in combination with Delphi technology and SPOT discussion method, remove the factors that do not meet the requirements and repeated factors, and adjust the factors with low correlation. Specifically, the, characteristics S10 and S11 in Table 1 are merged into F10 , i.e. Official document writing; Delete factor S14 , S17 , S18 , S21 , S22 andS25 ; New factor F14 - improvement of on-the-job education is added.
28
W. Luo et al.
Finally, a talent digitalization evaluation index system database consisting of 20 factors (F1 − F20 ) is obtained, as shown in Table 2 for details. Among them, knowledge consists of three secondary hands, namely, “compliance knowledge, management skills and informatization skills”, including eight tertiary evaluation indicators; Competence consists of four secondary indicators, which are “communication ability, writing ability, innovation ability and learning ability”, including seven secondary indicators in total; Attitude is composed of three secondary indicators: “conscientious performance, executive ability and service awareness”; The specialty consists of two secondary indicators, see Table 2. Table 2. Digital talent evaluation index system base Level I indicators
Level II indicators
Level III indicators
Level I indicators
Level II indicators
Level III indicators
Knowledge and skills
Compliance knowledge
Legal knowledge F1
Ability
Innovation
Patents/Property F12
Rules and regulations F2
Awards F13
Network risk awareness F3 Managerial skills
Major F4 Degree F5
Informatization skills
on-the-job education F14 Professional Certificate F15
Responsibility
Annual assessment ranking F16
Training F6
Execution effect
Performance Appraisal Results F17
ERP application F7
Service awareness
Ratio of service objects F18
Professional judgment
Risk control ability F19
Refine on
Labor awards F20
Computer operation level F8 Ability
Learning
Communication ability
Cross department work F9
Writing ability
Official writing F10
Attitude
Professionalism
Thesis and books F11
3 Analysis of Impact Factors of Power Transmission Industry Interpretative Structural Model (ISM) is a theoretical system used to explain the relationship between systems and among various factors within a plan, and is widely used in system dynamics. Specifically, the operation of ISM is divided into three steps: first, invite an expert group to analyze the correlation between factors and build an Adjacency
Influential Factors and Implementation Path of Talent Digital Evaluation
29
Matrix; second, phase change is carried out based on Matlab to create the Accessibility Matrix; finally, an interpretation model is formed through hierarchical extraction. 3.1 Adjacency Matrix Adjacency matrix is used to test whether there is correlation between factors. In general, if adjacency matrix A is expressed as a square array of i × j: ⎡
⎤ a11 . . . a1j A = ⎣ ... ... ...⎦ ai1 . . . aij
(1)
Then, the value range of aij in (1) is: aij =
0 Si and Sj has no direct relationship 1, Si and Sj has direct relationshop
(2)
For this reason, in this paper, the initial relational matrix can be obtained by consulting the experts in Table 2 again through a 0–1 matrix questionnaire. Table 3. Accessibility Matrix satisfying conditions (1) - (3) F1
F2
F3
F4
F5
F6
F7
F8
F9
F10
F11
F12
F13
F14
F15
F16
F17
F18
F19
F20
F1
0
0
1
0
1
0
0
0
0
0
1
0
0
0
1
0
0
0
0
0
F2
1
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
1
0
1
0
F3
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
F4
0
0
0
0
0
1
0
0
0
0
0
1
0
0
0
1
0
0
0
0
F5
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
F6
0
1
0
0
0
0
0
1
0
0
0
0
0
0
1
0
0
0
0
1
F7
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
F8
0
0
0
0
1
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
F9
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
F10
0
0
0
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
1
0
F11
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
1
1
0
0
F12
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
1
1
0
0
F13
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
1
1
1
0
1
F14
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
1
1
0
0
F15
0
1
0
0
0
0
0
0
0
0
1
0
1
0
0
0
0
1
0
0
F16
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
F17
0
0
0
1
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
F18
0
0
0
0
1
0
0
0
1
0
0
0
0
1
0
0
0
0
1
0
F19
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
F20
1
0
0
0
0
0
0
0
0
0
1
0
0
0
0
1
0
0
0
0
30
W. Luo et al.
3.2 Accessibility Matrix The reachable matrix is a skeleton matrix (1) measured through the Boolean operation. The reachable matrix reflects whether there is an influence path and the degree of one element Si in the adjacency matrix for another element Sj . According to the definition of accessibility matrix, if there is: (A + I )k−1 = (A + I )k = (A + I )k+1 = M
(3)
Then, M = (A + I )k+1 is the reachable matrix of adjacency matrix A. 3.3 Result Derivation In this study, the eye(·) function and isequal (·) function are used in MATLAB for programming calculation, and the 20 factors in total, from F1 to F20 in Table 2 are considered, the accessibility matrix satisfying conditions (1) - (3) can be obtained as shown in Table 3.
4 Construction of Interpretation Model for Power Transmission Industry In the reachable matrix (Table 3), if the reachable set is assumed to be C(Fi ), it represents the factor set from row i to column j; The antecedent set is P(Fj ), representing the factor set from column j to row i; The symbol of intersection of reachable set C(Fi ) and antecedent set P(Fj ) is: C(Fi ) P(Fj ). If, the accessibility matrix is decomposed into hierarchical relations, and if the factors of each row and lattice meet the conditions (4) and (5): P(Fj ) (4) C(Fi ) = C(Fi ) P(Fj ) = C(Fi )
P(Fj )
(5)
Then, when condition (4) is met, factor Fi can be considered as the upper - level factor; while if condition (5) is met, then Fi is a low - level factor. Considering the upper level factors, when they (the upper - level factors) appear, the corresponding rows and columns should be crossed out in the accessibility matrix, and then new upper - level factors should be found from the remaining rows and columns. This cycle continues until all the upper elements have been cleared. In this study, through the above methods, it can be got the factor set that affects the digital evaluation of talents in the power transport industry, as shown in Table 4. Specifically, there are 6 levels, including L1 = {F3 , F7 , F8 , F17 }; L2 = {F1 , F9 , F19 }; L3 = {F2 , F10 , F16 , F18 }; L4 = {F6 , F13 }; L5 = {F11 , F12 }; L6 = {F4 , F5 , F14 , F15 , F20 }. Combining Table 2 and Table 4, an explanatory structure model diagram is drawn, shown in Fig. 1.
Influential Factors and Implementation Path of Talent Digital Evaluation
31
Table 4. Influential Factors of Accessibility Matrix for Digital Evaluation of Talents in Electric Power Transportation Industry Hierarchy
Factor set
L1
F3 , F7 , F8 ,F17
L2
F1 , F9 ,F19
L3
F2 , F10 , F16 ,F18
L4
F6 , F13
L5
F11 , F12
L6
F4 , F5 , F14 , F15 ,F20
Fig. 1. ISM - based influencing factor system
5 Research Results and Enlightenment 5.1 Research Results Through Matlab and Boolean operation, this paper divides the factor set that affects the digital evaluation of talents in the electric power transportation industry into six progressive levels, specifically: (1) Top - level factors: ERP application, computer operation level, performance appraisal results, and network risk awareness; (2) Core factors: legal knowledge, number of cross departmental work, risk control ability;
32
W. Luo et al.
(3) Important factors: rules and regulations, official document writing, annual performance ranking, service object ratio; (4) Key factors: training and awards; (5) Principal factors: patents/knowledge, papers/works; (6) Underlying factors: major, academic degree, labor awards, in-service education promotion, professional qualification certificate. 5.2 Research Enlightenment The index system that affects the digital evaluation of talents in the electric power transportation industry includes six progressive levels. According to the interaction between factors, this paper proposes that the digital level of mastery in enterprises can be improved from the following aspects: (1) Strengthen the informatization level training of enterprise staff and strengthen value recognition. From the top level, among the four main factors, ERP application, computer operation level and network risk awareness are closely related to enterprise information systems. On the one hand, the application of information system in the enterprise improves the production and operation efficiency of the enterprise; On the other hand, how to improve the practical application rate of information systems has become a problem that enterprise managers need to consider. Therefore, it is a feasible method to build an excellent corporate atmosphere and culture and improve the value recognition of employees on the application of information technology through the training of employees’ informatization level. (2) Optimize the assessment indicators and improve the staff’s sense of participation. It can be seen from Fig. 1 that among the influencing factors at the middle level (L2-L5 in Fig. 1), performance appraisal, annual performance ranking, risk management and control, and the ratio of service objects play an important role. This shows that the construction of digitalization of enterprise talents requires extensive participation of employees. In the performance evaluation indicators, the appraising weight of digitalization - related standards should be increased, such as “technical competition awards, ERP security sharing, frequency of use of OA system”, to improve the participation of employees in digitalization construction. In addition, in the assessment indicators, it is also necessary to improve the weight of indicators related to information security and operational security to ensure the steady progress of enterprise digital construction. (3) Strengthen infrastructure construction and improve the internal digital talent training system of the enterprise. From the perspective of the underlying factors, the non-external factors such as the significant and academic degree of the staff have had a fundamental and fundamental impact on the digital talent system. Therefore, in the enterprise, it is necessary and capable to distinguish and compare the differences between the two types of employees by identifying “excellent employees” and “ordinary employees” under the digital talents, forming a mutual matching support mechanism for “excellent ordinary” employees, improving the operating ability of ordinary employees in computer operation, network security, enterprise ERP application, and thus indirectly improve the level of digital talents in the enterprise.
Influential Factors and Implementation Path of Talent Digital Evaluation
33
5.3 Research Conclusion The digital talent construction of enterprises is a very complex project, which is affected by the interaction of multiple system elements. With the help of the ISM model, taking electric power transportation enterprises as an example, this paper preliminarily explores the influencing factors of the evaluation of enterprise talent digitalization and deduces a six - level progressive influencing factor system, which can guide the development of enterprise talent digitalization evaluation scale in the future. However, it should be noted that the results of this study are obtained by electric power transport enterprises, and whether the study is applicable to other enterprises needs further analysis. Acknowledgment. This project is supported by Application of Talent Management Wisdom Based on ERP (E654900065).
References 1. Aliyev, A.G.: Technologies ensuring the sustainability of information security of the formation of the digital economy and their perspective development directions. Int. J. Info. Eng. Elect. Bus. (IJIEEB) 14(5), 1–14 (2022) 2. Aliyev, A.G., Shahverdiyeva, R.O.: Scientific and methodological bases of complex assessment of threats and damage to information systems of the digital Economy. Int. J. Info. Eng. Elect. Bus. (IJIEEB) 14(2), 23–38 (2022) 3. Ravenelle, A.J.: The digital economy. Contemporary Sociology: A Journal of Reviews 50(5), 416–417 (2021) 4. Wei, Z.: Analysis on the key factors and guarantees of enterprise digital transformation-taking the science and technology innovation board listed companies as an example. People’s Forum Academic Frontier (18), 70–78 (2022). (in Chinese) 5. Qiwei, Z., Xin, L., Donghong, L.: Research framework on multiple functions and openness of enterprise digital transformation. Journal of Xi’an Jiaotong University (Social Science Edition) 42(03), 10–19 (2022). (in Chinese) 6. Jinping, X.: Hold high the great banner of socialism with Chinese characteristics and work together to build a socialist modern country in an all-round way[EB/OL] [2022-10-16]. https:// www.12371.cn/2022/10/25/ARTI1666705047474465.shtml 7. Jing, X., Ping, Z.: Can digitalization realize the “quality improvement and increment” of enterprise green innovation-Based on the resource perspective. Scientific research, 1–19. (in Chinese) 8. Xia, F., Yun, Z., Ping, Z.: Research on the difficulties and countermeasures of financial talents’ data literacy training in the digital economy era. China University Teaching 09, 23–27 (2022). (in Chinese) 9. Yanping, L., Le, L., Xiang, H.: Digital human resource management: integration framework and research outlook. Sci. Technol. Progr. Countermeas. 38(23), 151–160 (2021) (in Chinese) 10. Pei, Z., Yong, S.: Spatial pattern and influencing factors of coordinated development of information infrastructure and integrated infrastructure. Economic Issues Exploration 10, 94–104 (2022). (in Chinese) 11. Liyan, C.: Research on innovation choices and destructive innovation paths of latecomers. Harbin University of Technology, Harbin (2013). (in Chinese) 12. Jianhua, H., Yongduan, G., Lijun, C.: Digital platform enterprise network contract psychological contract: content, measurement and service performance impact verification. Business Economics and Management 03, 5–15 (2022). (in Chinese)
34
W. Luo et al.
13. Yumei, L., Luyan, M., Yi, Z.: Research on the impact of innovative use behavior of enterprise information system - based on social impact theory and individual innovation traits. Science and Technology Management Research 39(10), 177–184 (2019). (in Chinese) 14. Roberts, C., Geels, F.W.: Conditions and intervention strategies for the deliberate acceleration of sociotechnical transitions: lessons from a comparative multi-level analysis of two historical case studies in dutch and danish heating. Technology Analysis & Strategic Management 31(9), 1081–1103 (2019) 15. Järvi, K., Khoreva, V.: The role of talent management in strategic renewal. Employee Relations 42(1), 75-89 (2020) 16. Laia, Y., Ishizakab, A.: The application of multi-criteria decision analysis methods into talent identification process: a social psychological perspective. J. Bus. Res. 109, 637–647 (2020) 17. Yijie, W., Wenjun, M., Jianli, F.: Impact factors and realization path of carbon neutralization target under new development pattern. Resource Development and Market 38(08), 915– 920+985 (2022) (in Chinese) 18. Ping, L., Yi, Z.: Wang yijie research on the stability of low carbon technological innovation alliance under punishment mechanism - based on the perspective of stochastic evolutionary game. Resource Development and Market 38(1), 16–22 (2022). (in Chinese) 19. Afha, B., Scc, D., Edce, F., Bp, G., et al.: From global to national scenarios: bridging different models to explore power generation decarbonizatio based on insights from socio-technical transition case studies. Technol. Forecast. Soc. Chang. 151, 119882 (2020) 20. Jifeng, W., Peng, L., Jinzhao, L., Yufei, S.: The course and development trend of power system digitalization. China Southern Power Grid Technology 15(11), 1–8 (2021). (in Chinese) 21. Liu, P., Zhang, Y., Xiong, Z., Wang, Y., Qing, L.: Judging the emotional states of customer service staff in the workplace: A multimodal dataset analysis. Frontiers in Psychology 13, 1001885. https://doi.org/10.3389/fpsyg.2022.1001885 22. Ahuja, S., Kaur, P., Panda, S.N.: Identification of influencing factors for enhancing online learning usage model: evidence from an Indian University. Int. J. Edu. Manage. Eng. (IJEME), 9(2), 15–24 (2019) 23. Matthew, O., Buckley, K., Garvey, M.: A framework for multi-tenant database adoption based on the influencing factors. Int. J. Info. Technol. Comp. Sci. (IJITCS) 8(3), 1–9 (2016)
Knowledge Associated with a Question-Answer Pair Igor Chimir(B) Odessa State Environmental University, Lvovskaya Street 15, Odessa 65016, Ukraine [email protected]
Abstract. The article is devoted to the modeling and presentation of declarative knowledge, which exchanged by participants of a question-answer dialogue. The knowledge of the reactive (responding) agent of the dialogue is considered from the point of view of the classical epistemological understanding of knowledge. It is shown in what cases the epistemological formula “the subject R knows that the proposition p is not complete and must be supplemented with a question for which the proposition p is the true answer. Further, the article examines the logical connection between the question and the relevant answer in the context of the interrogative formula of the question, considered in erotetic logic. Declarative knowledge expressed by the structural component of the interrogative formula, which is called the subject of the question, is proposed to be modeled using linguistic-independent semantic entities in the categories of the Ternary Description Language. The use of the Ternary Description Language makes it possible to construct models of the subject of the question that do not depend on a specific natural language. The final part of the article describes a set of eight patterns that can be used to represent declarative knowledge associated with a question-answer pair. Keywords: Dialogue · Declarative knowledge · Ternary Description Language
1 Introduction The words “conversation” and “dialogue” are often understood as synonyms. However, in what follows we will prefer the word “dialogue”, and the participants in the dialogue will be called dialogue agents. We are primarily interested in the dialogue that takes place between an artificial dialogue agent (for example, a chatbot) and a human (a chatbot user). In the process of the dialogue, the dialogue agents form a dialogue transaction, which is an elementary complete cycle of knowledge exchange between the agents. Although there can be many participants in a dialogue, only two dialogue agents form a dialogue transaction. A dialogue agent, in the process of dialogue interaction with its partner, can perform one of two alternative roles: (1) the role of an active dialogue agent; (2) the role of a reactive dialogue agent. An active agent is an inquiring agent. The part of the dialogue © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 35–44, 2023. https://doi.org/10.1007/978-3-031-36115-9_4
36
I. Chimir
transaction that the active agent forms is not necessarily a single verbal question, but always has the status of an interrogative. In goal-oriented dialogues, the main motivation for a dialogue agent to play the role of an active agent is the lack of knowledge necessary to continue the dialogue. An agent plays the role of an active dialogue agent in case when he needs additional knowledge that he expects to receive from his dialogue partner. A reactive agent is a responding agent, and the part of a dialogue transaction that it forms has a response status with respect to the active agent. A dialogue agent who honestly plays the role of a reactive agent provides the partner with the knowledge that, from his point of view, is relevant to the active agent question. A chatbot can be viewed as an artificial agent participating in an unstructured verbal question-answering dialogue. Often a chatbot is a reactive agent and must be able to generate answers to human questions. In a dialogue transaction, the chatbot and its partner exchange knowledge. Thus, a chatbot is a knowledge-based system, and when designing it, it is important to rely on a model for representing knowledge in a question-answer transaction. Knowledge bases, which are used by modern chatbots, are connected in one way or another with natural language [1, 2]. Representing knowledge using natural language sentences seems to be the norm since the dialogue between the chatbot and its user is verbal. However, this should not be imperative. We will consider a possible way of representing declarative knowledge associated with a question-answer transaction in semantic categories that are invariant to a particular natural language. The article develops the direction of research published in [3]. In the initial part of the article, attention is focused on understanding the knowledge of the reactive agent of dialogue from the point of view of the classic epistemological formula “true and justified belief”. The subsequent part of the article analyzes the logical connection between the structure of the question and the structure of the relevant answer from the point of view of the interrogative formula of Belnap and Steel [4]. The structural component of the interrogative formula, called the subject of the question, is considered an indefinite relevant answer. The subject of the question, as a rule, is modeled by a set of propositions and is represented by sentences of a natural language. The article discusses the possibility of modeling the subject of the question using linguistically independent entities in the categories of the Language of Ternary Description [5–8]. The final part of the article describes the declarative knowledge models and patterns associated with a question-answer transaction.
2 Knowledge of the Reactive Agent What is the declarative knowledge that a reactive agent operates and how can it be understood and represented? Epistemologists propose to understand this knowledge as a proposition and call it knowledge-that. The following verbal formula is known: “knowledge-that is a Justified and True Belief (JTB)” [9]. According to this verbal formula, the necessary and sufficient conditions for the reactive agent R to know that the proposition p takes place can be formulated as follows: 1) proposition p is true, 2) reactive agent R believes that p, and
Knowledge Associated with a Question-Answer Pair
37
3) reactive agent R is justified in believing that p. Epistemology is a philosophical science, and therefore, when interpreting the formula JTB and conditions of the reactive agent R knowledge, that p, pays much attention to philosophical issues, for example, how the phrase “justified belief” should be understood. The only thing that is not questioned is the statement that the proposition must be true. False propositions are not considered knowledge. Thus, the truth of the proposition is understood in the absolute, and not in the relative (in relation to the reactive agent) sense. One can agree with this, however, one can find examples when a proposition that is true for one person is not the same for another person. For example, the proposition represented by the sentence “Tobacco smoking is a virtue” may be true for one reactive agent, but false for another. The knowledge of the agent R that p may be represented by a diagrammatic formula in the form of a UML class diagram using the relation of association between the classes R and p. The UML formula of knowledge-that is shown in Fig. 1.
Fig. 1. Diagrammatic formula of knowledge-that of a reactive agent
The multiplicity [1..1] of the pole of the association adjacent to the class R means that only one object of the class R participates in the association, and the multiplicity [1..*] of the pole adjacent to the class p means that one or more objects of the class p participate in the association. In other words, one reactive agent can know one or more propositions. The memory of a reactive agent cannot store all possible knowledge-that. This knowledge is formed by the reactive agent in the process of dialogue as a response to the question of the active agent. From the point of view of the way of forming knowledge-that by the reactive agent, all questions of the active agent can be divided into two classes: (1) questions of the “search instruction” type and (2) questions of the “task” type. A question of the “search instruction” type assumes that at the moment the question is received, the requested knowledge is already in the memory of the reactive agent, and the structural elements of the question position the memory to the required area. To form an answer to a question of the “search instruction” type, the use of the attention resource is not required. An example is the following question: “What is your name?” When the active agent, in order to gain access to the knowledge of the reactive agent, uses a question of the “search instruction” type and the requested knowledge is already in the memory of the reactive agent, the diagrammatic formula of knowledge-that shown in Fig. 1 is adequate. Let us show that this formula needs to be refined in the case when a question of the “task” type is used to gain access to the knowledge of a reactive agent. To obtain the knowledge that is requested using a question of the “task” type, the reactive agent must solve the task associated with this question. The answer is a variant of the solution of the task, obtained by the reactive agent. An example of a question “task” type is: “If Socrates was born in 469 BC, how old would Socrates be today?” It is clear that, most likely, the knowledge requested by this question is not stored in the
38
I. Chimir
memory of the reactive agent in a ready-made form, and the use of a mental resource is required for their formation. From the point of view of the declarative-procedural dichotomy of knowledge, knowledge-that is declarative. When forming declarative knowledge, which is the answer to a question of the “task” type, the reactive agent must use procedural knowledge. The procedural knowledge that a reactive agent uses depends on how a question of the “task” type is formulated. Let us illustrate the last statement by the example of convergent questions of the “task” type. Convergent questions are different questions that assume the same true answer. In other words, the same true proposition of the reactive agent can be the answer to different questions of the active agent. Schaffer analyzes various convergent questions, including the following two [10]. (1) Is there a goldfinch in the garden, or a raven? (2) Is there a goldfinch in the garden, or a canary? If it is true that there is a goldfinch in the garden, then the reactive agent should give the same answer to both questions: “There is a goldfinch in the garden.” However, in order to get this answer in the case of question (1), the reactive agent needs relatively simple procedural knowledge (knowledge that makes it possible to distinguish a goldfinch from a raven), and in order to get the same answer in the case of question (2), the reactive agent needs much more complex procedural knowledge (knowledge that makes it possible to distinguish a goldfinch from a canary). Thus, the diagrammatic formula for the knowledge-that of a reactive agent shown in Fig. 1, in the general case, is incomplete and must be supplemented by a question, the answer to which is knowledge-that of the reactive agent. Figure 2 shows a refined formula for the knowledge-that.
Fig. 2. A refined formula for the knowledge-that of a reactive agent
In Fig. 2, Que denotes a class of questions. The class diagram shown in Fig. 2 differs from the class diagram shown in Fig. 1 in that it contains one more relation of the association type between the class of propositions p and the class of questions Que. The multiplicity [1..*] of the pole of the association adjacent to the class Que means that one or more objects of the class Que participate in the association. In other words, the same proposition can be the answer to several different questions.
3 Interrogative Formula of the “Search Instruction” Type Question The logical connection between the question of the “search instruction” type and the answer to this question can be revealed by explicating the question formula proposed by Belnap and Steel [4]. The question formula, which Belnap and Steel consider, and which they call the interrogative formula, defines a question as a composition of two logical components: the request of the question and the subject of the question. In Fig. 3, the interrogative formula is presented in the form of a UML class diagram.
Knowledge Associated with a Question-Answer Pair
39
Fig. 3. The interrogative formula in the form of a UML class diagram
The class diagram in Fig. 3 represents the class of questions Que as a composite of two classes: the class of question’s requirements Req and the class of question’s subjects Subj. The multiplicity [1..1] of poles adjacent to the Req and Subj classes means that exactly one object of each of these classes participates in the composition. The names of the poles reflect the roles of objects of these classes in the composition. The pole adjacent to the Subj class is named “repository”, since the subject of the question points to that part of the reactive agent’s memory where the knowledge that includes the answer to the question is located. The pole adjacent to the Req class is named “constraints” since the requirement of the question sets constraints that the reactive agent must take into account when forming an answer. Belnap and Steel’s theory assumes that the knowledge determined by the subject of the question is propositions that are represented by sentences in natural language, and the requirement of the question may, for example, require that these sentences be considered as an alternative. Questions are classified with the cardinality of the subject of the question. The subject of the question can be: (1) a finite and small number of propositions; (2) an infinite or very large number of propositions. In the first case, the relevant questions are called whether-questions, and in the second case – which-questions. An example of a whether-question is the question of tobacco smoking, which was, allegedly, formulated by the English king James the first [4]: «Tobacco smoking: a vice, a virtue, a vagary, an extravagance, a cure for all ills?» The subject of this question consists of exactly five elements: “tobacco smoking is a vice”; “tobacco smoking is a virtue”; “tobacco smoking is a vagary”; “tobacco smoking is an extravagance”; “tobacco smoking is a cure for all ills”. An example of a which-question is the following question: «Which positive integer is the smallest prime greater than 45?» The subject of this question consists of an infinite number of elements and can be represented by two clauses with one variable. x is the smallest prime greater than 45. x is a positive integer. It is clear that whether-question is a special case of which-question, and whethersubject is a special case of which-subject. Belnap and Steel propose a notation for whether-subject, according to which whether-subject is represented by a clause with variables (such a clause is called the subject matrix) and a set of constraints that define the values of variables in the matrix (such constraints are called category conditions).
40
I. Chimir
Both the subject matrix and category conditions are represented by means of natural language. The answer to whether-question or which-question is part of the subject of the question that the reactive agent separates from the whole subject of the question in accordance with the constraints set by the requirement of the question. The subject of whetherquestion or which-question contains the answer to the question, therefore the subject of a question of the type “search instruction” can be understood as an indefinite answer, or an answer with some (often significant) degree of indefiniteness.
4 Knowledge Structure in Question-Answer Transactions The concept of “proposition” is not identical to the concept of “sentence” in natural language. A proposition can be represented by a natural language sentence. Representation of propositions in the form of sentences in natural language is convenient when writing articles on the interpretation and modeling of knowledge-that, but, from the author’s point of view, it limits the developers of artificial reactive agents using such kinds of models. If the knowledge-that used by an artificial reactive agent is tied to a specific natural language, then the agent can maintain a dialogue only in a specific language and is monolingual. In the case when the representation of the knowledge-that of a reactive agent is carried out in categories that are not related to a specific natural language, then there is a potential opportunity to design an artificial reactive agent that can support a dialogue in several languages and be multilingual. It is very important to be multilingual, especially for an artificial dialogue agent “dwelling” on the Internet. 4.1 The Language of Ternary Description It is advisable to represent declarative knowledge in a question-answer transaction by means, of the ontology which is based on non-linguistic entities. One of these means is the Language of Ternary Description (LTD), proposed by Uyemov [5–8]. The initial entity in LTD is an object, in the most general sense of the word. An object, depending on its place in the structure of knowledge, can exist in one of three forms: object-thing, object-property, and object-relation. The categories “thing” and “property” have the traditional meaning in the LTD, which is accepted in classical logic, and the meaning of the category “relation” differs from the traditional one. It is generally accepted to use the concept of “relation” to denote the mutual influence of several things, or the relationship between things. In the context of the LTD, a relation is understood as what constitutes a thing, or the relationship that takes place in a thing. In other words, the relation in LTD is, in some sense, another name for the internal logical organization of a thing. The binary association of an object-thing with an object-property generates two prototypes for representing entities in LTD. 1) The name of the first prototype is “a thing, which possesses a property”, and the formal notation is: (∗)∗.
Knowledge Associated with a Question-Answer Pair
41
2) The name of the second prototype is: “a property, which attributed to a thing”, and the formal notation is: (∗))∗. The binary association of an object-thing with an object-relation generates two more prototypes for representing entities in LTD. 3) The name of the third prototype is: “a thing, in which a relation takes place”, and the formal notation is: ∗(∗). 4) The name of the fourth prototype is “a relation, which takes place in a thing”, and the formal notation is: ∗((∗). The object-thing symbol is written inside the brackets, the object-property symbol is written to the right of the brackets, and the object-relation symbol is written to the left of the brackets. The association of an object-thing with an object-property or with an object-relation has a direction. If the symbol of an object-thing is enclosed in ordinary (single) parentheses, then this means that the association is directed from the object-thing to the object-property or object-relation. In words, this is expressed as: “a thing, which possesses a property “, or “a thing in which a relation takes place.” An asymmetric (double) parenthesis means that the association is directed from an object-property or an objectrelation to an object-thing. Verbal formulation: “a property, which attributed to a thing “, or “a relation, which takes place in a thing.” An object, depending on the degree of indefiniteness of knowledge about it, exists in one of three forms: 1) a definite object (the asterisk in the prototype substitutes by symbol t), 2) an indefinite object (the asterisk in the prototype substitutes by symbol a), 3) an arbitrary object (the asterisk in the prototype substitutes by symbol A). The categories “thing, property and relation”, as well as “definiteness, indefiniteness, and arbitrariness” are independent and form nine classes of objects: (1) a definite objectthing, (2) an indefinite object-thing, (3) an arbitrary object-thing, (4) a definite objectproperty, (5) an indefinite object-property, (6) an arbitrary object-property, (7) a definite object-relation, (8) an indefinite object-relation, (9) an arbitrary object-relation. Substituting into prototypes for representing entities in LTD instead of the asterisk one of the symbols t, a or A, we get a set of possible models of knowledge-that represent in the ontological basis of LTD. Since there are four prototypes, in each of which we can substitute two symbols of the object, and the total number of object symbols is three, there are 24 possible models of knowledge-that. 4.2 Representation of Declarative Knowledge in Question and Answer The question-answer transaction, in the context of the LTD-representation of declarative knowledge, will be considered as a development of the idea of the interrogative formula, shown in Fig. 3. The subject of the question is the key element of the question-answer transaction. The reactive agent, when constructing an answer, is essentially engaged in transforming the
42
I. Chimir
indefinite knowledge pointed to by the subject of the question into definite knowledgethat of the answer. When modeling the knowledge pointed to by the subject of the question, we will restrict ourselves to only the following four alternative models. Ksubj = (t)a
(1)
Ksubj = (a)t
(2)
Ksubj = (t))a
(3)
Ksubj = (a))t
(4)
There are several reasons for choosing models (1)–(4) to represent the knowledge of the subject of the question. First, the subjects of most of the examples of questions described in works devoted to the logic of questions and answers can be represented by one of the models (1)–(4) [4, 10, 11]. Secondly, these are exactly the models that correspond to the idea that a reactive agent, when constructing an answer, transforms the indefinite knowledge of the subject of the question into definite knowledge-that of the answer. One of the objects in models (1)–(4) is indefinite, and the active agent expects to receive more specific knowledge about it from the reactive agent. The model Ksubj = (t)a represents knowledge about a specific definite thing, which possesses indefinite property. A question with such a subject is formed by an active agent in the case when he wants to know which properties a given thing possesses. The model Ksubj = (a)t represents knowledge about an indefinite thing, which possesses specific definite property. A question with such a subject is formed by an active agent in the case when he wants to know which things possess a given property. The model Ksubj = (t))a represents knowledge about an indefinite property, which is attributed to a specific definite thing. A question with such a subject is formed by an active agent in the case when he wants to know which specific property is attributed to a given thing. The model Ksubj = (a))t represents knowledge about a definite property, which is attributed to some indefinite thing. A question with such a subject is formed by an active agent in the case when he wants to know to which things a given property is attributed. The disadvantage of models (1)–(4), from the point of view of software engineering, is their poor adaptability for mapping into relevant data structures. These models could be useful for the development of software systems if we could find a way to transform them into types or data structures of modern programming systems. This is, first of all, about the datalogical interpretation of an indefinite object. One of the possible datalogical interpretations of indefiniteness is multiplicity. An indefinite object can be understood as a set of definite objects, and the cardinality of this set as a measure of indefiniteness. Then a decrease in the degree of indefiniteness of an object is equivalent to a decrease in the cardinality of the corresponding set. An indefinite object turns into a definite one when the cardinality of a set becomes equal to one, or when the set is represented by one specific definite object.
Knowledge Associated with a Question-Answer Pair
43
Taking into account such interpretation of indefiniteness, we can replace indefinite objects with lists of specific objects and describe models (1)–(4) with the following patterns. Ksubj = (thing) → list of properties
(5)
Ksubj = (list of things) → propery
(6)
Ksubj = (thing)) ← list of properties
(7)
Ksubj = (list of things)) ← property
(8)
Patterns (5)–(8) are datalogic analogs of models (1)–(4), which describe declarative knowledge that can be transferred to a reactive agent by means of the subject of the question. Since the subject of the question is, in fact, an indefinite answer, and the reactive agent, when constructing the answer, reduces the degree of indefiniteness of the subject of the question to a level acceptable for the answer, then the knowledge-that patterns of the answer should be similar to patterns (5)–(8). The patterns of knowledge-that of the answer differ from the patterns of knowledge of the subject of the question by the cardinality of the set of objects-things or objects-properties. We may describe the knowledge-that of the answer with the following patterns. Kans = thing POSSESSES PROPERTIES list of properties
(9)
Kans = list of things POSSESSES PROPERTY property
(10)
Kans = list of properties ATTRIBUTED TO thing
(11)
Kans = property ATTRIBUTED TO list of things
(12)
5 Conclusion Chatbots are capable of maintaining long-term dialogue with a human. The dialogue protocol between a chatbot and a human can be represented as a sequence of transactions. The human partner forms the interrogative part of the transaction in order to receive from the chatbot a portion of the declarative knowledge necessary to continue the dialogue. The portion of declarative knowledge that is requested from the chatbot belongs to the class of knowledge-that that is a proposition. When forming an answer in the form of a proposition, the chatbot uses procedural knowledge, which depends on how the question is formulated. Thus, a chatbot is a knowledge-based system, and during its development, it is necessary to rely on one or another model of knowledge representation associated with a dialogue transaction. The logical connection between question and answer in a single transaction is modeled by Belnap and Steel’s interrogative formula. The key component of the interrogative formula is the subject of the question, which for the question of “search instruction” type can be considered as knowledge-that of the answer, but with a significant degree of indefiniteness.
44
I. Chimir
It is advisable to build models of knowledge on an ontological basis, which is not associated with linguistic categories but represents the semantic essence of knowledge. In this case, there is a potential opportunity to develop multilingual chatbots, which is especially important for chatbots operating on the Internet. The article illustrates the applicability of Uyemov’s Ternary Description Language for representing portions of declarative knowledge associated with a question-answer transaction. The patterns for representing the knowledge of the subject of the question and the corresponding answer, represented by formulas (5)–(12), explain the internal logical essence of the questionanswer transaction in the context of knowledge representation and are not associated with a specific natural language. Further development of this work is supposed to be carried out in the direction of synthesizing a model of a dialogue knowledge base, which includes not only declarative knowledge associated with dialogue transactions but also procedural knowledge necessary for conducting a question-answer dialogue focused on solving problems.
References 1. Serban, I.V., Lowe, R., Henderson, P., Charlin, L., Pineau, J.: A survey of available corpora for building data-driven dialogue systems. Dialogue Discourse 9(1), 1–49 (2018) 2. Rashkin, H., Smith, E.M., Li, M., Boureau, Y.-L.: Towards empathetic open-domain conversation models: a new benchmark and dataset. ACL (2019) 3. Chimir, I., Ghazi, A., Abu-Dawwas, W.: Modeling human dialogue transactions in the context of declarative knowledge. Int. Arab J. Inf. Technol. 10(3), 305–315 (2013) 4. Belnap, N.B., Steel, T.: The Logic of Questions and Answers. Yale University Press, New Haven and London (1976) 5. Uyemov, A.: The ternary description language as a formalism for the parametric general systems theory. Part I. Int. J. General Syst. 28(4–5), 351–366 (1999) 6. Uyemov, A.I.: The ternary description language as a formalism for the parametric general systems theory. Part II. Int. J. General Syst. 31(2), 131–151 (2010) 7. Uyemov, A.I.: The ternary description language as a formalism for the parametric general systems theory. Part III. Int. J. General Syst. 32(6), 583–623 (2003) 8. Leonenko, L.: The language of ternary description and its founder. Modern Logic 8(3,4), 31-52 (2000-2001) 9. Armstrong, D.M.: Belief, Truth, and Knowledge. Cambridge University Press, Cambridge (1973) 10. Schaffer, J.: Knowing the answer. Philos. Phenomenol. Res. 75(2), 383–403 (2007) 11. Wisniewski, A.: The Posing of Questions. Logical Foundations of Erotetic Inferences. Kluwer Academic Publishers, Dordrecht (1995)
Research on Low Complexity Differential Space Modulation Detection Algorithm Shuiping Xiong1(B) and Xia Wu2 1 School of Artificial Intelligence and Manufacturing, Hechi University, Yizhou 546300,
Guangxi, China [email protected] 2 Lenovo Information Products (Shenzhen) Co. Ltd., Shenzhen, China
Abstract. SM (Spatial Modulation) technology is a key research direction of MIMO (Multiple Input Multiple Output) technology in recent years. Because SM uses a single RF module, only one transmitting antenna is selected for activation at any time slot, Thus, SM avoids the inter-channel interference (ICI), but SM technology requires channel state information (CSI) for signal detection at the receiver, so channel estimation must be performed. The channel estimation process is very complex and needs a lot of resources, and the error of channel estimation will cause serious loss of detection performance. In this case, differential spatial modulation (DSM) technology is proposed. As a new multi-antenna transmission technology, Differential Space Modulation (DSM) can improve the transmission rate of wireless communication to a certain extent, but DSM system has high complexity of signal detection algorithm. In order to reduce the complexity of DSM signal detection, this paper proposes a better low-complexity detection algorithm, called time-division combination (TDC) detection algorithm. Theoretical analysis and simulation results based on MATLAB show that the improved time-sharing merging algorithm not only has the best detection performance, but also greatly reduces the computational complexity, thus improving the theoretical and practical significance. Keywords: Differential space modulation · TDC · Detection algorithm · ML
1 Introduction SISO (Single In Single Out) technology has no inter channel interference (ICI), but its spectral efficiency is low. The traditional MIMO (Multiple Input Multiple Output) technology solves the problem of low spectral efficiency. But on the other hand, the inter channel interference is severe, the decoding complexity for the receiver is high, while some technologies are not suitable for asymmetric antenna systems. A new MIMO technology, according to Spatial Modulation (SM) technology, has emerged and gradually became a research hotspot in recent years [1]. However, SM technology has some disadvantages. First, the traditional SM system hides part of the bit information in the serial number of the active antenna, so it requires the transmit antenna N t to be the power © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 45–54, 2023. https://doi.org/10.1007/978-3-031-36115-9_5
46
S. Xiong and X. Wu
of 2.Second, in order to avoid inter channel interference, the modulation channels from the transmitter to the different receivers of SM systems need to be uncorrelated. Finally, since the receiver of SM system assumes that the channel state information is known, that is, channel estimation must be performed before detection, and the complexity of channel estimation process is very high, and there will be corresponding errors in the estimation, which will have a serious impact on the performance of SM system. In order to solve the problems in the SM system, the DSM technology was proposed by Bian Y et al. in 2015. Each time slot of the DSM also only activates one antenna, which eliminates the need for inter antenna synchronization and avoids inter channel interference [2]. DSM technology, on the basis of SM, additionally introduces a time domain differential algorithm to transmit data in the form of space-time blocks. Each spacetime block contains information about multiple symbols. The information carried by the serial number of the DSM antenna maps the bit information to the activation sequence of the transmission antenna. DSM makes difference in time domain, perfectly avoiding the problem of channel estimation. DSM technology mainly includes two algorithms: maximum likelihood detection and time-sharing combining (TDC) detection [3].
2 Maximum Likelihood Detection Algorithm The maximum likelihood detection algorithm ML (Maximum Likehood) is a classical detection algorithm with good bit error rate performance in the current DSM system. 2.1 ML Detection Algorithm Process The maximum likelihood detection algorithm ML detection algorithm of the DSM system. The ML detection process is as follows [4]: (1) Step 1: Search all possible N t antenna activation sequence/symbol sequence pairs (uorder , sNt ) within a time slot to get a set of all candidate solution information block matrices M . According to formula (1), the euclidean distance between the received signal matrix at the receiver and the post-processing matrix of all candidate solution information block matrices can be calculated in turn; X 2 dDSM _ML = Yt −Yt−1 X F , (X ∈ M )
(1)
X X (2) The second step is to compare all Euclidean distances dDSM _ML ; dDSM _ML to get the minimum; (3) The third step is to obtain the optimal information block matrix corresponding to the minimum Euclidean distance, which will be used as the recovery signal X, and finally obtain the corresponding antenna activation sequence and symbol sequence. The above process is simplified as [5]:
Xˆ t = arg minYt − Yt−1 X 2 ∀X∈M
(2)
For DSM system, since only one transmit antenna is active in each time slot, there is only one element not 0 in each column vector of the information block matrix
Research on Low Complexity
47
X. Then the ML detection algorithm can be further deduced from Eq. (2) as follows [6]. t yt (n) − yt−1 (uorder (n)) · sN (n)2 uˆ = arg min , sˆ t order_ML
Nt _ML
t
(uorder ,sNt )∈M 1≤n≤N t
(3) t the estimated N t Indicates the number of transmitting antennas, uˆ order_ML t sequence of antenna activation, sˆNt _ML the estimated sequence of symbols, the nth time slot, yt (n)the nth column vector, that is, the reception vector of the nth time slot, uorder (n) the sequence number of the nth time slot activation antenna, sNt (n) the constellation point symbol transmitted by the nth time slot activation antenna, and. Represents the column of uorder (n). Equation (3) can be further expressed [7]:
t t , sˆN uˆ order_ML = t _ML
arg min
(uorder ,sNt )∈M 1≤n≤N 1≤w≤N t r
yt (w, n) − yt−1 (w, uorder (n)) · sN (n)2 t
(4) Although the DSM system reduces the process of channel estimation and the total complexity of the receiver compared with the SM system, it can be seen from Formula (2) that every detection requires searching all possible antenna activation sequence/symbol sequence pairs, which shows that the complexity of the ML detection algorithm based on the DSM system is still high [8]. 2.2 Complexity Analysis of ML Detection Algorithm According to Eqs. (2) and (4), where yt−1 (w, uorder (n)), sNt (n) and are complex numbers, four real number multiplication operations are required for each 2 yt−1 (w, uorder (n)) · sNt (n) calculation, and yt (w, n)−yt−1 (w, uorder (n)) · sNt (n) . It represents the sum of the squares of the real part and the imaginary part of the element yt (n, w)−yt−1 (uorder (n), w) · sN (n)2 . The square of the element in requires 2 real t 2 number multiplication [9]. Then calculate once yt (w, n)−yt−1 (w, uorder (n)) · sNt (n) , 6real number multiplications are required. For a candidate solution information block 2 matrix X, N r N t times of yt (n, w)−yt−1 (uorder (n), w) · sNt (n) calculation is required, so 6Nr Nt times of Yt − Yt−1 X 2 real number multiplication is required. In one ML detection of the DSM system, for the N r root transmit antenna, there are 2log2 (Nt !) different antenna activation sequence. Assuming that the modulation mode of each time slot is the same, there are 2Nt b different symbol sequences, that is, there are M different symbols X in the set. Therefore, the computational complexity of the ML detection algorithm based on the DSM system can be expressed as [10]: CDSM _ML = 6Nr Nt · 2log2 (Nt !)+Nt b
(5)
The relative complexity reduction of ML detection algorithm is expressed by the following formula: (6) Cre(%) = 100 × CML −Cpropose /CML
48
S. Xiong and X. Wu
In formula (6), Cpose is the computational complexity of other detection algorithms [11].
3 TDC Detection Algorithm 3.1 TDC Detection Algorithm Process Equation (5) describes that the complexity CDSM _ML of ML detection algorithm in DSM system in one detection is 6Nr Nt · 2log2 (Nt !)+Nt b . Table 1. Detection process of TDC detector
In order to make DSM system more advantageous in practical application, its detection complexity needs to be greatly reduced. A TDC detection algorithm is proposed to directly detect the candidate solution information block X. Instead, the candidate solution vectors of each slot of the candidate solution Nr information block are calculated separately, and then the optimal information block is obtained by comprehensively detecting each slot. The detection process is shown in Table 1 [12]. 3.2 Complexity Analysis of TDC Detection Algorithm TDC detection algorithm breaks through the idea that traditional detection algorithms in DSM system directly detect candidate solution information blocks, and calculates the candidate solution vectors xi of each Nt slot of the candidate solution information block. Firstly, the optimal symbol vector is obtained by traversing all modulation symbol vectors
Research on Low Complexity
49
of different active antennas; Then, all the antenna activation sequences are traversed to obtain the optimal information block. The distance corresponding to the possible candidate solution vector xi of each time slot calculated by TDC detection algorithm can be expressed as D(xi ) = yt (i) − Yt−1 xi 2 , xi ∈ i
(7)
Equation (7): The set (u, s) ∈ i of candidate solution vectors xi is a complex matrix of size Nr × Nt , and the vector is a complex vector of size Nt × 1, and there is only one non-zero complex element, so Yt−1 xi can be simplified as formula (8) [13]. Yt−1 xi = sm yt−1 (ui ), xi ∈ i
(8)
Therefore, the number of 4Nr times of real number multiplication to be calculated once is time Yt−1 xi . Since a complex Nr × 1 vector of size is obtained by Yt−1 xi and is also a complex vector of size, the number of times of real number multiplication to be calculated once is times. Therefore, the total number of times of real number multiplication to be performed for a candidate solution vector is times. In addition, there is a possible candidate solution vector in a time slot, so in one detection of TDC detection algorithm, the distance corresponding to the possible candidate solution vector in any time slot is calculated, and the computational complexity required can be expressed as: CTDC_i = 6Nr Nt · 2b
(9)
One detection of TDC detection algorithm includes calculating the distance corresponding to the possible candidate solution vectors xi of Nt time slots. When all time slots of the DSM system adopt the same modulation method, the computational complexity at this time can be expressed as: CTDC = 6Nr Nt2 · 2b
(10)
When all Nt time slots adopt the same modulation method, the computational complexity of the TDC algorithm is shown in Formula (10), that is, the TDC detection algorithm proposed in this paper theoretically reduces the computational complexity of the DSM system so that the DSM system can be used in more practical scenarios. The relative complexity reduction and relative complexity of TDC detection algorithm proposed by the system are as follows:
Nt (11) CreTDC (%) = 100 × 1 − 2log2 (Nt !)+(Nt −1)b CreTDC =
Nt log (N !)+(Nt −1)b t 2 2
(12)
It can be seen from the above formula that in the DSM system, the relative reduction Nt of the complexity of TDC detection algorithm compared with the ML detection algorithm shows a strong upward trend with the increase of the modulation order b. In the case of large Nt and b, the reduction of complexity is very significant [14].
50
S. Xiong and X. Wu
4 Simulation Analysis The bit error rate performance and computational complexity of ML and TDC detection algorithms are simulated and verified through MATLAB 2019. 4.1 Simulation Analysis of Bit Error Rate Performance The SNR value of the simulation is 0–30 dB. At the same SNR, the TDC detection algorithm is simulated. The length of the data block included in each simulation is 10. The performance curves of the ML and TDC algorithms are given. Figure 1 shows the bit error rate performance curve of TDC algorithm under different Nt modulation combinations when the system spectral efficiency is the same. The modulation methods of, N t and N r are indicated in the figure. Under different conditions, the BER performance curve of TDC detection algorithm and ML algorithm always coincide [15]. When Nt and the modulation mode are the same as the modulation mode, the bit error rate performance of the system becomes better with the increase of Nr . As shown in the figure, when the signal to noise ratio is low, the bit error rate performance curves of Nr .Nt = 3, 8816 modulation mode and Nt = 5, 8888Q modulation mode show that the former has better bit error rate performance than the latter [16]. With the increase of the signal to noise ratio, the bit error rate performance of the two will gradually approach or even coincide, and then with the further increase of the signal to noise ratio, the latter has better bit error rate performance, and the performance difference will gradually increase, And with the increase of the number of receiving antennas, the signal-to-noise ratio corresponding to the intersection of the two curves gradually decreases. The small figure at the bottom left can clearly see the current Nr = 2 trend. The two curves cross and coincide before the signal noise ratio is 8. When the spectral efficiency is the same, the BER performance curve of the system is mainly affected in the case of low SNR. In the case of high SNR, the BER performance curve is mainly affected by the modulation method Nt , and the smaller the number of receiving antennas, the greater the impact of the number of transmitting antennas [17]. It can be seen from the above simulation that in any case, the TDC detection algorithm proposed by the system always maintains the same performance as the ML detection algorithm, which is consistent with the theoretical analysis. In general, under the Nr same conditions, the bit error rate performance of DSM systems decreases with the increase of spectral efficiency; In the case of the same spectral efficiency, the bit error rate performance of the system is determined by Nr and modulation methods; The bit error rate performance of the DSM system will improve with the Nr increase of under the same conditions as the modulation mode [18]. 4.2 Complexity Simulation Analysis Figure 2 shows the complexity curve of TDC algorithm. Figures 2 and 3 respectively show the relative complexity reduction curve and its relative complexity curve of TDC algorithm when N t and modulation modes change. It can be clearly seen from the two figures that, under the same modulation mode, the relative complexity of TDC algorithm decreases with the increase of N t , and decreases faster and faster. Similarly, at the same
Research on Low Complexity
51
Fig. 1. BER performance of TDC in same spectral efficiency
N t time, when the modulation order continues to rise, the relative complexity of the TDC algorithm also continues to decrease, and the greater the number, the faster the reduction. Figure. When N t = 4, the modulation mode of and N t is QPSK, the relative complexity of TDC algorithm has been reduced to 10−2 . When Nt = 8, the modulation mode of and Nt is 16PSK, the relative complexity of TDC algorithm has been reduced to 10−12 . It can be seen that when the modulation order and Nt are large and the modulation order is high,.the complexity of TDC algorithm is significantly reduced and the advantages are obvious, which is consistent with the theoretical analysis [19].
Fig. 2. Relevant complexity reduction of TDC in different Nt and modulation
52
S. Xiong and X. Wu
Fig. 3. Relevant complexity of TDC in different Nt and modulation
Figure 4 shows the relative complexity curve of the improved detection algorithm TDC algorithm with the change of Nt and modulation methods when the number of receiving antennas Nr is different As shown in the figure, with the change of Nt and modulation modes, the curves are completely coincident at different times, which shows that the relative complexity of TDC detection algorithm is independent of the size of Nr , which verifies its theoretical derivation [20].
Fig. 4. Relevant complexity of TDC in different Nr and modulation
Through the experimental simulation of the complexity, the proposed TDC detection algorithm significantly reduces the detection complexity of the DSM system, and with the increase of the modulation order and Nt , the relative complexity decreases exponentially [21], which is consistent with the theoretical analysis. TDC algorithm enables DSM
Research on Low Complexity
53
system to be applied in higher and larger modulation order scenarios and Nt and broadens the application scenarios of DSM system.
5 Conclusion This paper analyzes the idea and detection process of signal detection algorithm in DSM system, and proposes a low complexity detection algorithm. TDC algorithm with optimal performance, aiming at the reason of high complexity of ML detection algorithm in DSM system. The complexity of TDC algorithm is theoretically analyzed, and the formula of the relative reduction of TDC detection algorithm complexity is derived. Then through the simulation and comparison of TDC algorithm and ML algorithm, it is found that in any case, TDC algorithm can guarantee the same bit error rate performance as ML algorithm and reduce the computational complexity. And the relative complexity of the algorithm decreases exponentially with the increase of Nt and modulation order, so TDC algorithm has more advantages when Nt and modulation order are large. Acknowledgment. This project is supported by the School-Level Scientific Research Project “Research on Signal Detection Algorithm of Differential Space Modulation System (2019XJZD004).
References 1. Mugen, P., et al.: Introduction to Wireless Communication, pp. 1–27. Beijing University of Posts and Telecommunications Press, Beijing (2011). (in Chinese) 2. You, X., Cao, S., Li, J.: Development status and prospect of the third generation mobile communication system. Acta Electronica Sinicas 27(Z1), 3–8 (1999) 3. Xie, X.: Long term evolution plan and comparison of the third generation mobile communication. Telecommun. Sci. 22(2), 1–4 (2006). (in Chinese) 4. Hu, D., et al.: Application status and prospect of 4G technology. Pract. Electron. (1), 30–30 (2016). (in Chinese) 5. Zhu, H.: Development prospect of the fourth generation mobile communication. China Sci. Technol. Inform. 17, 111–112 (2010). (in Chinese) 6. Yao, Y.: The fourth generation mobile communication system and its key technologies. Xi’an Univ. Posts Telecommun. J. Sci. 12(5), 25–29 (2007). (in Chinese) 7. Hu, H., Dong, S., Jiang, Y., et al.: Analysis of the fourth generation mobile communication technology. Comput. Eng. Des. 32(5), 1563–1567 (2011). (in Chinese) 8. You, X., Pan, Z., Gao, X., et al.: 5G mobile communication development trend and several issues key technology. Sci. China: Inform. Sci. (5): 551–563 (2014). (in Chinese) 9. Bian, Y., Cheng, X., Wen, M., et al.: Differential spatial modulation. IEEE Trans. Veh. Technol. 64(7), 3262–3268 (2015) 10. Afif, O.: Mobile and wireless communications system for 2020 and beyond (5G) (2014) 11. Osseiran, A., Braun, V., Hidekazu, T., et al.: The foundation of the mobile and wireless communications system for 2020 and beyond: challenges, enablers and technology solutions. IEEE Veh. Technol. Conf. 14(2382), 1–5 (2014) 12. Wang, J.: Research and Simulation of Multi-user and Multi carrier Spatial Modulation Technology, pp. 25–40. University of Electronic Science and Technology, Chengdu (2013). (in Chinese)
54
S. Xiong and X. Wu
13. Zhi, L., Cheng, X., et al.: A low-complexity optimal sphere decoder for differential spatial modulation. In: IEEE Global Communications Conference, pp. 1–6 (2015) 14. Jun, L., Wen, M., et al.: Differential spatial modulation with gray coded antenna activation order. IEEE Commun. Lett. 20(6), 1100–1103 (2016) 15. Wang, Y.: Research on Signal Detection Algorithm of Spatial Modulation System Based on Spherical Decoding. Chongqing University, Chongqing (2016). (in Chinese) 16. Men, H., Jin, M.: A low-complexity ml detection algorithm for spatial modulation systems With PSK constellation. IEEE Commun. Lett. 18(8), 1375–1378 (2014) 17. Mengna, L.: Enhanced Differential Space Modulation. South China University of Technology, Guangzhou (2016). (in Chinese) 18. Kumar, P., Singh, A.K.: Mapreduce algorithm for single source shortest path problem. Int. J. Comput. Netw. Inform. Secur. 12(3), 11–21 (2020) 19. Das, N.R., Rai, S.C., Nayak, A.: Intelligent scheduling of demand side energy usage in smart grid using a metaheuristic approach. Int. J. Intell. Syst. Appl. (IJISA) 6(10), 30–39 (2018) 20. Polatgil, M.: Outlier detection algorithm based on fuzzy c-means and self-organizing maps clustering methods. Int. J. Math. Sci. Comput. (IJMSC) 8(8), 21–29 (2022) 21. Abu Samra, A.A., Qunoo, H.N., Al Salehi, A.M.: Distributed malware detection algorithm (DMDA). Int. J. Comput. Netw. Inform. Secur. 9(8), 48–53 (2017). https://doi.org/10.5815/ ijcnis.2017.08.07
Mathematical Model of the Process of Production of Mineral Fertilizers in a Fluidized Bed Granulator Bogdan Korniyenko(B) and Andrii Nesteruk National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, Kyiv 03056, Ukraine [email protected]
Abstract. An analysis of approaches to mathematical modeling of the process of manufacturing mineral fertilizers in a fluidized bed granulator was carried out. A mathematical model of the granulation process in a fluidized bed has been developed, which considers the process as heterogeneous and three-phase, during which three separate phases interact with each other: particles – granulation centers, starting material – ammonium sulfate in the form of drops, and coolant – air. The mathematical model takes into account the hydrodynamics of the fluidized bed, the transfer of kinetic energy, the dissipation of energy, the compression of droplets with particles, their adhesion to the surface, the kinetics of drying the solution on the surface of the particles. The proposed mathematical model can be used to build information technology for managing the granulation process in a fluidized bed. Keywords: Mathematical Model · Mineral Fertilizers · Fluidized Bed · Granulation
1 Introduction The industrial application of the fluidization method is caused by a significant list of its advantages. Active mixing of the solid phase takes place in the fluidized bed, the quality of the processing of which directly affects the quality of the finished product. Also, fluidized bed granulators have a relatively simple design, and are quite amenable to mechanization and automation. With modern world trends towards the growth of consumption of products of various industries and the environmental situation, the problem of rational use of energy and raw materials in industrial production in order to obtain the maximum amount of a finished product of a given quality is acute. An important step in solving this problem is the creation of an adequate mathematical model of the process. The mathematical model should correctly reflect the technological process, its characteristic features, but also should not be overcomplicated with details that have an insignificant effect on the solution of the given task. Obtaining an adequate model of © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 55–64, 2023. https://doi.org/10.1007/978-3-031-36115-9_6
56
B. Korniyenko and A. Nesteruk
the device allows you to correctly develop a real strategy for the implementation of the control of the technological process. In devices with a fluidized bed, granulation is carried out by spraying pulps, solutions or melts on the surface of liquefied particles. At the same time, thin films of granular matter are deposited on the particles, where they are dried and crystallized, thereby increasing the size of the granules to the required size. Another (insignificant) part, which is introduced into the solution layer, dries in the spray zone, forming small dry particles, some of which are carried by the gas flow from the apparatus (carrying out), others remain in the apparatus as an internal return (new granulation centers). The main advantages of this method include: small dimensions and high productivity of the installation; spherical shape of granules of the finished product; obtaining a product of the required chemical and granulometric composition; intensification of heat and mass exchange processes due to the maximum degree of contact between the solid particles of the suspended layer and the gas coolant; automated control of the installation. Disadvantages of the method include: dust removal and thorough cleaning of the gas leaving the device; different residence time of particles in the apparatus. In connection with the constant growth of the scale of production and, accordingly, devices with a fluidized bed, significant difficulties arise taking into account the phenomena of gas-particle interaction (resistance force) and particle-particle interaction (collision force). Solving these problems through long-term studies on pilot plants is quite expensive. To facilitate the design process of devices with a fluidized bed, computer modeling can be a useful tool. The main difficulties in modeling devices with a fluidized bed in natural size are associated with a large difference in scales: the largest flow structures can be of the order of several meters, some structures can directly depend on particle-particle collisions and particle-gas interactions that occur at the millimeter level. When granulating solutions by dehydration, the requirement to obtain a product with a certain granulometric composition comes to the fore. In general, the mechanism of granule growth depends on the properties of the substances used, the process regime and other factors that determine the nature of the interaction between the dispersed liquid and solid phases. In our case, numerous studies have confirmed that the condition of the finished product, the main quality indicator of which is the equivalent diameter of the particles, is most affected by the temperature of the fluidized bed in which they are formed. That is why it was chosen as the main controlled value in the device under consideration. Consider the construction of an apparatus for granulating substances in a fluidized bed (Fig. 1). The solution is fed into the fluidized bed granulator 1 with the help of the executive device 2, and the granulation centers are fed with the executive device 3. The heated coolant – air – is supplied from the bottom to the top. The finished product – granules are unloaded using the executive device 4.
Mathematical Model of the Process of Production of Mineral Fertilizers
57
Fig. 1. Apparatus for granulating thermolabile substances in a fluidized bed: 1 – granulator, 2 – device for introducing the initial solution, 3 – device for introducing particles; 4 – a device for unloading the finished product.
2 Review of Mathematical Modelling Methods of Processes of Dehydration and Granulation in the Fluidized Bed The balance model is used to describe the change in particle size distribution of the granulation process in a fluidized bed [1, 2]. A general packet-mode length-based balance equation that describes the rate of change of the particle number density function n(t,L) is given as follows [3]: ∂n(t, L) = B(t, L) ∂t −
∂ (G(t, L)n(t, L)) ∂L
(1)
(2)
1
+
∞ 1 L2 L β(t, (L3 − λ3 ) 3 ,λ ) ∫ n(t, (L3 − λ3 ) 3 )n(t, λ)d λ − n(t, L) ∫ β(t, L, λ)n(t, λ)d λ 2 2 0 0 (L3 − λ3 ) 3 (3) ∞
+ ∫ S(t, λ)b(t, L|λ)n(t, λ)d λ − S(t, L)n(t, L) L
(4)
58
B. Korniyenko and A. Nesteruk
The growth of granules in the apparatus with a fluidized bed occurs mainly due to layering. At the same time, it is necessary to take into account the aggregation of particles. Only one kernel of granulation growth in a fluidized bed is used to model the balance [2]: G=
2(1 − b)Φ ∞
(5)
π ∫ nL2 dL 0
In the discrete method, the balance model equation is solved at each size interval. Therefore, the advantage of this method is that as a result we will get a distribution of particles by size. The moment method is a variant of solving the balance model under moment conditions, which is defined as: +∞
mk (t) = ∫ n(L; t)Lk dL
(6)
0
The moments are closely related to important integral quantities of the particle population, such as the average or total surface area and volume of the particles. Compared to the balance model, the numerical solution requires less computing power. The development of computer technology has made it possible to use the hydrodynamics model, which describes the dynamics of the interaction of gas and solid particles. For modeling in a fluidized bed granulator, there are two different categories of hydrodynamics models: the Euler model and the Lagrangian model [4]. The Euler model allows the inclusion of several secondary phase solids. Conservation of mass and momentum are performed respectively for each phase. Thus, the Euler model solves a set of n continuity and momentum equations, making this approach one of the most complex multiphase models. The Euler-Euler model, known as the continuum model or the two-fluid model, describes the evolution of solid-gas phase interactions. The interaction between the two phases depends on the hydraulic resistance between the phases, that is, the local relative velocities of the phases and the local volume fractions of the phases. As a result, the simulation of the method of computational hydrodynamics based on the Euler-Euler framework is accepted for the study of the multi-phase flow of gas-solid bodies in a fluidized bed granulator [5–7]. The Lagrangian model solves the equations of motion for each pellet, taking into account particle collisions and forces acting on the particle from the gas side. Therefore, when the number of particles is large, it is better to use Euler-Euler models [8–10]. The Lagrange-Euler model describes gas bubbles as discrete particles that can collide, coalesce, stop, contract, and grow. The Euler model is not suitable for the solid phase, but it is suitable for describing the emulsion of the gas phase and particles [11–15].
3 Mathematical Model of the Process of Production of Mineral Fertilizers in a Fluidized Bed Granulator During the creation of the model, the process of dehydration and granulation in a fluidized bed was considered as a three-phase heterogeneous process in which three components interact: particles, initial solution and coolant [16–18]. During modeling, it was assumed
Mathematical Model of the Process of Production of Mineral Fertilizers
59
that the fluidized bed parameters change over time without taking into account the change in height and the radial component, the heat exchange between all process components is convective, the particles are monodisperse, there is no porosity and agglomeration, the droplets have a narrow size distribution, between drops do not collide, stick together and stick to the walls of the case [19, 20]. The following system of equations was used to describe the heat exchange process in the granulator: – Particles temperature change: d p = ap Sp (a − p ) − Mp vdry Qdry + Gd xd Qcryst dt + Mp Cp rp p0 + Rad Mp Cd (d − p ) − γ − φps
Mp Cp
(7)
where ap – heat transfer coefficient of particles, W/(m2 *K); S p – particles surface area, m2 ; Θ a – heat carrier temperature, K; Θ p – particles temperature, K; M p – mass of particles, kg; vdry – the specific speed of drying moisture on particles or in drops, kg/(kg*s); Qdry – specific heat of drops drying, J/kg; Gd – mass flow of drops, kg/s; x d – concentration of solution drops, Qcryst – specific heat of drops crystallization, J/kg; M p – mass of particles, kg; C p – specific heat capacity of particles, J/(kg*K); r p – coefficient of axial dispersion of particles, 1/s; Θ p0 – initial temperature of particles, K; Rad – specific speed flowing of drops with particles, kg/(kg*s); C d – specific heat capacity of drops, J/(kg*K); Θ d – drops temperature, K; γ Θ – energy dissipation during collision, J/s; ϕ ps – transfer of kinetic energy, J/s. – Drops temperature change: Md Cd
d d = Md Cd d 0 − Rad Mp Cd (d − p )+ dt +Mp vdry Qdry + ad Sd (a − d )
(8)
where M d – mass of drops, kg; C d – specific heat capacity of drops, J/(kg*K); Θ d0 – initial temperature of drops, K; Rad – specific speed flowing of drops with particles, kg/(kg*s); M p – mass of particles, kg; Θ d – drops temperature, K; Θ p – particles temperature, K; vdry – the specific speed of drying moisture on particles or in drops, kg/(kg*s); Qdry – specific heat of drops drying, J/kg; ad – heat transfer coefficient of drops, W/(m2 *K); S d – drops surface area, m2 ; Θ a – heat carrier temperature, K. – Heat carrier temperature change: d a = Ga (Ca a0 − Ca a ) − ap Sp (a − p ) − ad Sd (a − d ) (9) dt where Ga – mass flow of heat carrier, kg/s; C a – specific heat capacity of heat carrier, J/(kg*K); Θ a0 – initial temperature of heat carrier, K; Θ a – heat carrier temperature, K; ap – heat transfer coefficient of particles, W/(m2 *K); S p – particles surface area, m2 ; Θ p – particles temperature, K; ad – heat transfer coefficient of drops, W/(m2 *K); S d – drops surface area, m2 ; Θ d – drops temperature, K; M a – mass of heat carrier, kg. Ma Ca
The developed mathematical model takes into account fluidized bed hydrodynamics, kinetic energy transfer, energy dissipation, compression of droplets with particles, their
60
B. Korniyenko and A. Nesteruk
adhesion to the surface, kinetics of solution drying on the surface of particles. The system of equations was developed taking into account empirical ratios for calculating the specific rate of drying, the specific rate of deposition of droplets on particles as a result of adhesion, the coefficient of axial dispersion of particles, heat transfer coefficients, the ratio for calculating the loss of material and the thickness of the coating layer, as well as initial conditions. The growth of granules is more likely, the greater the adhesion forces of liquid droplets with solid particles and its speed. The adhesive properties of the drop, in turn, depend on the hardness of the surface of the granules and the properties of the sprayed substance. Specific speed flowing of drops with particles can be calculated by the following formula: Gd Gd Std xd = ( ) (10) Rad = Mp0 Mp0 Std + 0.35 where M p0 – initial mass of particles, kg; Gd – mass flow of drops, kg/s; St d – Stokes number for drops, which we can calculate by the following formula: Std =
ρd νa dd2 μa dp
(11)
where ρ d – drops density, kg/m3 ; va – heat carrier speed of movement, m/s; d d – drops diameter, m; μa – dynamic viscosity of heat carrier, Pa*s; d p – particles diameter, m. Collision energy dissipation is the rate of energy dissipation within the solid phase due to collisions between particles. This phenomenon is modeled using the Lun correlation: γ =
12(1 − ep2 )g0 ρp ap2 1.5 √ p dp π Sp
(12)
where ep – coefficient of restoration of collision of particles with other particles, g0 – radial distribution function, d p – particles diameter, m; S p – particles surface area, m2 ; ρ p – particles density, kg/m3 ; ap – heat transfer coefficient of particles, W/(m2 *K); Θ p – particles temperature, K. The transfer of kinetic energy of random particles from the solid phase to the liquid phase can be calculated by the formula: φps =
−3Kps p Sp
(13)
where K ps – coefficient of the force of interaction between a liquid and a solid body, S p – particles surface area, m2 ; Θ p – particles temperature, K. The amount of heat released when moisture is removed from the surface of the droplets can be rewritten as follows: Ma vdry Qdry = β
Mh2o Sp PQdry Rp
(14)
where β – mass transfer coefficient, m/s; M h2o – molecular weight of water, g/mol; R – universal gas constant, (m2 *kg)/(s2 *K*mol); ΔP – partial pressure difference, Pa.
Mathematical Model of the Process of Production of Mineral Fertilizers
61
4 The Numerical Results Analysis of the Mathematical Model To obtain dynamic results of the developed mathematical model, a program was created to calculate this mathematical model using the Runge-Kutt method of the 4th order in the Python programming language. With the help of built-in mathematical libraries, calculations were carried out and the temperature behavior of particles, drops and coolant was visualized from the moment the installation was turned on until the end of the process. The following values given in Table 1 were used for modeling. Table 1 Values of the main process parameters. Name
Marking
Value
Mass of particles
Mp
1.5
Mass of drops
Md
2
Mass of heat carrier
Ma
3.5
Initial mass of particles
M p0
1.53
Specific heat capacity of drops
Cd
1590
Specific heat capacity of particles
Cp
1420
Specific heat capacity of heat carrier
Ca
1011
Initial heat capacity of heat carrier
C a0
1015
Initial temperature of particles
Θ p0
300
Initial temperature of drops
Θ d0
293
Initial temperature of heat carrier
Θ a0
393
Coefficient of axial dispersion of particles
rp
1*10–5
The specific speed of drying moisture on particles or in drops
vdry
0.47
Specific heat of drops crystallization
Qcryst
82300
Specific heat of drops drying
Qdry
5000
Heat transfer coefficient of particles
ap
5.57
Heat transfer coefficient of drops
ad
5.535
Particles surface area
Sp
7.07*10–7
Drops surface area
Sd
6.07*10–6
Mass flow of drops
Gd
0.7
Mass flow of heat carrier
Ga
1
Concentration of solution drops
xd
0.37
After the calculations, we get a graph of the dependence of the temperatures of drops, drops, and time on time, which is shown in Fig. 2. In order to maintain the stable operation of the apparatus with a fluidized bed in the necessary hydrodynamic mode inside them, it is necessary to develop an effective control system for the process of dehydration and granulation. The quality of this control system is closely related to the accuracy of the mathematical model of the object for which this control system is being developed. The obtained results from the calculation
62
B. Korniyenko and A. Nesteruk
Fig. 2. Graph of dependence of particle, droplet and air temperatures on time (1 – temperature of particles, 2 – temperature of drops).
of the mathematical model prove that it can serve as a basis for the development of an effective control system.
5 Conclusion A mathematical model of the mineral fertilizer production process in a fluidized bed granulator is proposed, which can serve as a basis for creating a control system for the granulation and dehydration process in a pseudo-fluidized bed. The presented mathematical model expresses the heat exchange between drops, particles and the coolant taking into account such parameters as fluidized bed hydrodynamics, kinetic energy transfer, energy dissipation, compression of drops with particles, their adhesion to the surface, kinetics of solution drying on the surface of particles. The mathematical model of the mineral fertilizer production process in the fluidized bed granulator was calculated by numerically and it was established that the system needs 110 s to stabilize the temperature, the temperature of the particles stabilizes at 367 K, and the temperature of the particles at 384 K.
References 1. Adetayo, A.A., et al.: Population balance modelling of drum granulation of materials with wide size distribution. Powder Technol. 82(1), 37–49 (1995) 2. Vreman, A.W., van Lare, C.E., Hounslow, M.J.: A basic population balance model for fluid bed spray granulation. Chem. Eng. Sci. 64(21), 4389–4398 (2009) 3. Syamlal, M., Rogers, W., O.Brien, T.J.: MFIX Documentation: Volume1, Theory Guide. National Technical Information Service, Springfield, VA (1993). DOE/METC9411004, NTIS/DE94000871993
Mathematical Model of the Process of Production of Mineral Fertilizers
63
4. Gidaspow, D., Bezburuah, R., Ding, J.: Hydrodynamics of circulating fluidized beds, kinetic theory approach. in fluidization VII. In: Proceeding of the 7th Engineering Foundation Conference on Fluidization, pp. 75–82 (1992) 5. Lun, C.K.K., et al.: Kinetic theories for granular flow: inelastic particles in couette flow and slightly inelastic particles in a general flow field. J. Fluid Mech. 140, 223–256 (1984) 6. Kornienko, Y.M., Liubeka, A.M., Sachok, R.V., Korniyenko, B.Y.: Modeling of heat exchangement in fluidized bed with mechanical liquid distribution. ARPN J. Eng. Appl. Sci. 14(12), 2203–2210 (2019) 7. Kornienko, Y.M., Haidai, S.S., Sachok, R.V., Liubeka, A.M., Korniyenko, B.Y.: Increasing of the heat and mass transfer processes efficiency with the application of non-uniform fluidization. ARPN J. Eng. Appl. Sci. 15(7), 890–900 (2020) 8. Korniyenko B., Galata L., Ladieva L.: Mathematical model of threats resistance in the critical information resources protection system. Paper presented at the CEUR Workshop Proceedings, vol. 2577, pp. 281–291 (2019) 9. Korniyenko, B., Kornienko, Y., Haidai, S., Liubeka, A., Huliienko, S.: Conditions of nonuniform fluidization in an auto-oscillating mode. In: Hu, Z., Petoukhov, S., Yanovsky, F., He, M. (eds.) ISEM 2021. LNNS, vol. 463, pp. 14–27. Springer, Cham (2022). https://doi.org/10. 1007/978-3-031-03877-8_2 10. Korniyenko, B., Kornienko, Y., Haidai, S., Liubeka, A.: The heat exchange in the process of granulation with non-uniform fluidization. In: Hu, Z., Petoukhov, S., Yanovsky, F., He, M. (eds.) ISEM 2021. LNNS, vol. 463, pp. 28–37. Springer, Cham (2022). https://doi.org/10. 1007/978-3-031-03877-8_3 11. Korniyenko, B., Ladieva, L.: Mathematical modeling dynamics of the process dehydration and granulation in the fluidized bed. In: Hu, Z., Petoukhov, S., Dychka, I., He, M. (eds.) ICCSEEA 2020. AISC, vol. 1247, pp. 18–30. Springer, Cham (2021). https://doi.org/10. 1007/978-3-030-55506-1_2 12. Korniyenko, B., Ladieva, L., Galata, L.: Control system for the production of mineral fertilizers in a granulator with a fluidized bed. In: ATIT 2020 – Proceedings: 2020 2nd IEEE International Conference on Advanced Trends in Information Theory, № 9349344, pp. 307–310 (2020). https://doi.org/10.1109/ATIT50783.2020.9349344 13. Korniyenko, B., Galata, L., Ladieva, L.: Research of information protection system of corporate network based on GNS3. In: 2019 IEEE International Conference on Advanced Trends in Information Theory, ATIT 2019 – Proceedings, № 9030472, pp. 244–248 (2019). https:// doi.org/10.1109/ATIT49449.2019.9030472 14. Korniyenko, B.Y., Ladieva, L.R., Galata, L.P.: Mathematical model of heat transfer process of production of granulated fertilizers in fluidized bed. ARPN J. Eng. Appl. Sci. 16(20), 2126–2131 (2021) 15. Korniyenko, B., Zabolotnyi, V., Galata, L.: The optimization of the critical resource protection system of a mineral fertilizers manufacturing facility. In: Proceedings of the 11th IEEE International Conference on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications, IDAACS 2021, vol. 1, pp. 172–178 (2021) 16. Babak, V.P., Babak, S.V., Myslovych, M.V., Zaporozhets, A.O., Zvaritch, V.M.: Methods and models for information data analysis. In: Diagnostic Systems For Energy Equipments. SSDC, vol. 281, pp. 23–70. Springer, Cham (2020). https://doi.org/10.1007/978-3-030-44443-3_2 17. Polishchuk, M., Tkach, M., Parkhomey, I., Boiko, J., Eromenko, O.: Experimental studies on the reactive thrust of the mobile robot of arbitrary orientation. Indonesian J. Electr. Eng. Inform. 8(2), 340–352 (2020) 18. Karthika, B.S., Deka, P.C.: Modeling of air temperature using ANFIS by wavelet refined parameters. Int. J. Intell. Syst. Appl. 8(1), 25–34 (2016). https://doi.org/10.5815/ijisa.2016. 01.04
64
B. Korniyenko and A. Nesteruk
19. Ghiasi-Freez, J., Hatampour, A., Parvasi, P.: Application of optimized neural network models for prediction of nuclear magnetic resonance parameters in carbonate reservoir rocks. Int. J. Intell. Syst. Appl. (IJISA) 7(6), 21–32 (2015). https://doi.org/10.5815/ijisa.2015.06.02 20. Malekzadeh, M., Khosravi, A., Noei, A.R., Ghaderi, R.: Application of adaptive neural network observer in chaotic systems. Int. J. Intell. Syst. Appl. 6(2), 37–43 (2014). https://doi. org/10.5815/ijisa.2014.02.05
Simulation Study on Optimization of Passenger Flow Transfer Organization at Nanning Jinhu Square Metro Station Based on Anylogic Yan Chen1 , Chenyu Zhang1(B) , Xiaoling Xie1 , Zhicheng Huang1 , and Jinshan Dai2,3 1 College of Traffic and Transportation, Nanning University, Nanning 530200, China
[email protected]
2 Center for Port and Logistics, School of Transportation and Logistics Engineering,
Wuhan University of Technology, Wuhan 430072, China 3 Department of Industrial Systems Engineering and Management, National University
of Singapore, Singapore 117576, Singapore
Abstract. Nanning Jinhu square metro station is the transfer station of Nanning metro Line 1 and Line 3. Its passenger flow transfer organization efficiency greatly affects the operation efficiency of the urban rail transit system. This paper collects the station operation data and analyzes the transfer evaluation parameters, builds the station simulation model using the Anylogic software, sets the relevant scenarios to conduct the operation simulation analysis of the station, looks for the flow bottleneck of the passenger flow in the transfer organization, and takes measures such as iron fence diversion to optimize the passenger flow transfer organization, so as to improve the passenger flow transfer efficiency on the basis of reducing the passenger flow congestion. The paper further determines the effectiveness of the optimization measures by comparing the operation results of the simulation model before and after the optimization, and provides support for the design of transfer facilities and the formulation of operation strategies at the transfer station. Keywords: Anylogic simulation research · Passenger flow transfer at metro station · Organizational optimization
1 Introduction Nanning’s rail transit is in a rapid development stage. In 2019, Nanning’s subway travel accounted for 41% of the city’s total bus travel. The subway transfer station is the hub part of the urban rail transit system and an important distribution point of passenger flow. It undertakes the transfer and connection of passenger flow between lines. The efficiency of passenger flow transfer organization of the subway transfer station affects the operation efficiency of multiple lines. Nanning Jinhu Square Metro Station, an important passenger transfer hub, Nanning Rail Transit Line 3 and Line 1 transfer here. The passenger flow at Jinhu Square Station is relatively large and complex, and the transfer channel will be congested at the morning and evening peak hours, which gradually cannot meet the growing passenger flow transportation demand. Therefore, the optimization © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 65–78, 2023. https://doi.org/10.1007/978-3-031-36115-9_7
66
Y. Chen et al.
of passenger flow transfer organization at Jinhu Square Station is extremely urgent. Due to the scale, uninterrupted operation and passenger flow volatility of passenger transport hub, simulation modeling can be used to explore the bottleneck of passenger flow and provide support for formulating optimization measures. Among many simulation software, Anylogic software is a widely used simulation modeling tool that is applicable to discrete, system dynamics, multi-agent and hybrid systems [1]. It is mostly used for traffic simulation, pedestrian evacuation, production simulation, etc. Simulation modeling is conducted through Anylogic software to analyze and optimize the passenger organization, passenger flow dispersion, passenger flow conflict, etc. of metro stations, which is a current application hotspot. For example, V M. Antonova; N. A. Grechishkina; N. A. Kuznetsov and others used AnyLogic software to analyze the inbound passenger flow of typical subway stations to check the traffic congestion points [2]. Another scholar, Zhao Min published his master’s thesis “Research on Passenger Flow Control Method and Simulation of Urban Rail Transit Transfer Station” in April 2020, and verified the simulation model and method with Chengdu New South Gate Station as an example. According to the passenger flow data and the bottleneck type discrimination table, the identification results of the primary bottleneck are obtained, and then the bottleneck is relieved by adjusting the service time of the gate, setting the one-way passage of stairs, and arranging the fence for batch release. The rationality of passenger flow control measures is verified by simulation results [3]. The scholars’ research has laid a good foundation for the simulation study of the transfer passenger flow analysis and evacuation of the Jinhu Square metro station in Nanning City, and provided a good reference case.
2 Analysis of Transfer Facilities in Nanning Jinhu Square Metro Station 2.1 General Layout Jinhu Square Station is the transfer station between Nanning Metro Line 1 and Line 3. The main station of Line 1 is located at Minzu Avenue, which is arranged along the east-west direction of the road. The main station of Line 3 is located at Jinhu South Road, which is arranged along the north-south direction of the road. In terms of layout location, Nanning Department Store Wuxiang Square Shopping Center is set underground in Wuxiang Square to the west; In the south, there are mainly cultural centers such as Guangxi Press and Publication General Office and Nanning Book City; To the east, there are business office areas such as Modern International Building and East Manhattan Building, and to the north, there are Diwang Building and Agricultural Bank of China Guangxi Branch. It is adjacent to Nanhu Station in the west and Convention and Exhibition Center Station in the east. According to the field observation, the passenger flow characteristics of Jinhu Square Station have very obvious characteristics of the transfer station: the majority of passengers are transfer passenger flow, and the transfer facilities in the station become the main facilities to bear the passenger flow through, with only a small amount of inbound and outbound passenger flow, so the inbound and outbound facilities only bear a small
Simulation Study on Optimization of Passenger Flow Transfer Organization
67
amount of passenger flow; Such characteristics of passenger flow make it rare that the bottleneck of subway passenger flow such as entrance security check and gate is fully loaded, and most passenger flow is concentrated in the transfer channel. In terms of transfer mode, Jinhu Square Station belongs to the “L” shape configuration and uses channel transfer to meet the requirements of line interchange. Line 1 is relatively independent from the station located in Line 3. The probability of cross interference of passenger flow between the two stations is small, and the impact of inbound passengers outside the station on station efficiency is small. However, the passenger flow is all transferred through the passage, and the pressure of passenger flow inside and at both ends of the passage is high. The escalator near the passage will bear most of the passenger flow. 2.2 Analysis of Advantages and Disadvantages of Channel Transfer Channel transfer is generally applied to the relatively independent station structure at the intersection of the two lines. When the station hall cannot be directly used for transfer due to the distance between the platforms or the topographic conditions, separate transfer channels and stairs are set between the two stations for passengers to transfer. The layout of the channel transfer mode is flexible, which has great adaptability to the intersection of the two lines and the location of the station layout, and is conducive to the phased implementation of the two line projects. Generally, the transfer channel should be set in the middle of the station as far as possible to avoid the cross interference between the passenger flow of both sides and the passenger flow of the entrance and exit of the station, to a certain extent, reducing the concentration of passenger flow in the station hall. But at the same time, the disadvantages of channel transfer are also obvious. The transfer distance between two stations will increase the length of the passage, and the transfer distance is longer. All passenger flows are transferred through one or two channels. The channel bears greater pressure on passenger flow, and bears less impact of short-term large passenger flow. The transfer stations with height difference between two stations often arrange the vertically moving facilities in the transfer channel in the form of transfer escalator to realize the transfer between the two stations. The speed of passenger flow drops sharply at the transfer escalator, which is easy to cause the aggregation of the escalator ports. 2.3 Analysis of Transfer Facilities in Jinhu Square Station The station is divided into five parts according to the height setting: Line 1 station hall layer, Line 1 platform layer, transfer channel layer, Line 3 station hall layer, and Line 3 platform layer. A transfer channel is set in the west of the station of Line 1, which is connected with the transfer channel of Line 3 by escalator. There is a transfer node intersecting the transfer channel of Line 1 in the south of the station of Line 3. Therefore, the facilities at the transfer passage of Jinhu Square Station are arranged as follows: two ascending escalators, one descending escalator and one two-way staircase. Among them, the ascending escalator bears the passenger flow from Line 3 to Line 1, the
68
Y. Chen et al.
descending escalator bears the passenger flow from Line 1 to Line 3, and the staircase is used as a diversion measure during peak hours. It can be seen from the quantity configuration that more passengers transfer from Line 3 to Line 1 than from Line 1 to Line 3. Two ascending escalators and one descending escalator are arranged here. It can be seen from the actual observation that there are different levels of queuing at the entrance of escalators on both sides during the peak period, but passengers at the ascending escalator tend to wait for the escalator. Passengers at the descending escalator tend to change stairs. Therefore, in fact, the queue length at the entrance of the ascending escalator is longer than that of the descending escalator. There are two load-bearing columns at the exit of the descending escalator, which are easy to cause the passenger flow to slow down and small-scale aggregation. According to the field observation, at the transfer escalator, the passenger flow from Line 3 to Line 1 escalator gathered at the entrance. The passenger flow of the escalator from Line 1 to Line 3 gathers at the exit.
3 Results and Analysis of Passenger Flow Transfer at Nanning Jinhu Square Metro Station 3.1 Passenger Flow Data Analysis
Table 1. PassengerFlow Statistics of Jinhu Square Station Statistical Items
Flat Peak Period (person/hour)
Peak Period (person/hour)
Get off at Line 1
3396
3828
Transfer to Line 3
2630
2804
Get off at Line 3
3126
3556
Transfer to Line 1
2772
3482
Entrance/Exit G (Line 3)
90
125
Entrance/Exit F (Line 3)
48
64
Entrance/Exit D (Line 3)
18
22
Entrance/Exit A (Line 1)
162
201
Entrance/Exit B (Line 1)
258
331
Entrance/Exit C (Line 1)
300
402
In order to have a more comprehensive understanding of the transfer characteristics of passenger flow in Jinhu Square Metro Station and provide enough experimental data for the simulation model, a field survey of Jinhu Square Metro Station was conducted on weekends and holidays. The survey data obtained are shown in Table 1.
Simulation Study on Optimization of Passenger Flow Transfer Organization
69
3.2 Station Simulation Modeling The basic simulation of the station model needs to determine the parameters first.
Fig. 1. Plan Structure of Station
Fig. 2. Three Dimensional Structure of Station Line 1
The basic simulation parameters of the station model include the station plane structure diagram (as shown in Fig. 1), the station three-dimensional structure diagram (as shown in Figs. 2, 3, and 4), and the pedestrian logical flow line. The plane structure of the station shows the horizontal movement ability of the pedestrian, the three-dimensional
70
Y. Chen et al.
structure of the station shows the vertical movement ability of the pedestrian, and the logical flow line of the pedestrian constructs the pedestrian action plan.
Fig. 3. Three Dimensional Structure of Station Line 3
Fig. 4. Three Dimensional Structure Diagram of Station Transfer Channel
Compile train arrival code such as. “pedSource6.inject(300); pedWait.freeAll(); pedWait6.freeAll();” Set the target line for passengers to enter and leave the station in a fixed time and quantity, and complete the simulation of train entering and leaving the station.
Simulation Study on Optimization of Passenger Flow Transfer Organization
71
3.3 Station Simulation Modeling 3.3.1 Station Passenger Flow Density It refers to the density of passengers gathered in the effective service area of the station. The passenger aggregation density is inversely proportional to the station service level [4]. The dynamic value of passenger flow density at the platform with time can be observed through simulation. The passenger distribution density diagram of different time periods is also presented. According to the density map obtained, the passenger flow density of each region can be roughly obtained according to the passenger flow density comparison bar of the thermal map, and the bottleneck and congestion situation can be analyzed according to this data. According to the simulation of Anylogic, the thermal diagram is shown in Fig. 5.
Fig. 5. Station Passenger Density
3.3.2 Passenger Flow Density at the End of the Facility It refers to the density of passengers gathered at the upper or lower ports of the escalator. Due to the unbalanced characteristics of passenger arrival [5]. The escalator, gate and security check machine are easy to form passenger flow bottlenecks, resulting in congestion and reducing the safety and operation efficiency of the station. It has a great impact on the station service level. Therefore, whether the passenger density gathered at the end of the facility matches the arrival level of passenger flow is selected as the service level evaluation index. According to the Anylogic simulation, the passenger flow density of the facility section is shown in Fig. 6.
72
Y. Chen et al.
By the 180th second of operation, the data had begun to be distorted, and the flow at each port was still fluctuating normally, but the flow of the upper escalator of Line 1 was kept at about 75 people/s, which did not change with the change of passenger flow. It was obvious that there had been irresolvable congestion, leading to the inability of pedestrians to use the escalator normally.
Fig. 6. Flow Diagram of Each Port
3.3.3 Average Transfer Time of Passengers Average transfer time of passengers refers to the average time it takes for passengers to get off from one line to another. It directly reflects the convenience of the transfer facilities in the subway station. According to the simulation results of Anylogic, the average transfer time of passengers is 11.7 min. 3.4 Bottleneck Analysis of Passenger Flow Transfer at Stations Based on Simulation Model 3.4.1 Analysis of Simulation Operation Data When the train runs for about 180 s, the model is distorted and there are large congestion points that affect the operation of the station. 1) Through traffic statistics and analysis Except for the upper escalator of Line 1, all escalators show fluctuations in passenger flow as the train enters and leaves the station, indicating that the passenger flow through the escalator is still within its capacity and can be used normally. The escalator at the
Simulation Study on Optimization of Passenger Flow Transfer Organization
73
upper part of Line 1 fluctuates around 75 people per second, with no downward trend, and is not affected by the train entering and leaving the station, indicating that there has been a large congestion point near it, which is beyond its capacity, and cannot be cleared, and the channel function is lost. 2) Through passenger boarding/departure time analysis According to statistics, the time for passengers to get into the station from this station is much shorter than the time for passengers to get off and transfer, indicating that there are fewer inbound passengers outside the station compared with transfer passengers, each escalator is within the capacity range, and can enter the platform smoothly, and the passenger flow density at the platform is moderate, which does not affect the passengers to get on the train. Transfer passengers are the main passenger flow of the station. Most passengers enter the station and choose to transfer. Since each station does not affect the passengers entering the station, it indicates that the channel of each platform itself is smooth, and congestion may occur near the transfer channel. 3.4.2 Station Passenger Flow Density/Bottleneck Analysis As can be seen from Fig. 5, the escalator on the left side of Line 3, the transfer channel and the escalator on the upper side of Line 1 bear most of the passenger flow, with the density of about 1.5 people per square meter. Among them, although the escalator and transfer channel on the left side of Line 3 bear most of the passenger flow, and there is a certain degree of congestion in the simulation, it does not exceed the bearing range of the traffic capacity, and there will be bottlenecks affecting the efficiency, but it will not affect the overall operation of the station [6, 7]. On the contrary, the escalator at the upper part of Line 1 has a large passenger flow, which may be blocked in the transfer channel, blocking the outbound passenger flow and affecting the station function. Therefore, it can be determined that the key optimization point is located at the upper escalator of Line 1, and the secondary optimization point is the left escalator and transfer channel of Line 3. 3.4.3 Analysis of Passenger Flow Line Conflict Points According to the layout of the facilities in the station, passengers will present different action logic, and streamline is the intuitive expression of action logic. The passenger flow line conflict caused by the passenger facilities arrangement may become one of the reasons for congestion [9, 10]. The paper will draw the passenger flow lines and the conflict points generated by the intersection of different flow lines, so as to observe and determine the improvement plan.
74
Y. Chen et al.
4 Optimization and Simulation of Passenger Flow Transfer Organization at Jinhu Square Metro Station 4.1 Optimization Scheme After considering the implementation cost and practical operability, combined with the operation situation of the station which mainly focuses on passenger flow, the paper determines to use the combination of iron barrier and manual guidance in the area near the transfer channel, and try to avoid the conflict point by changing the passenger flow line, so as to achieve the purpose of optimizing the transfer organization. The specific optimization measures are as follows. 1) Partial optimization analysis of Line 1 The iron barrier shall be arranged at the upper escalator of Line 1, and the transfer/exit passenger flow line of Line 1 shall be moved down to avoid the passenger flow at the lower platform, and more passenger flow shall be diverted to the upper escalator and the middle staircase to avoid the conflict point. The escalator in the middle of Line 1 is equipped with two layers of iron horses, one is to separate the passenger flow on and off the platform, and the other is to separate the passenger flow from the lower escalator, so that the passenger flow confluence is no longer located at the stair port, and moves to a relatively open area. The right passage of the transfer passage is equipped with iron horses to divert passengers in advance. 2) Partial optimization analysis of Line 3 The iron barrier is set at the entrance of the transfer passage of Line 3 to separate the passenger flow, widen the passage for the passengers who transfer from Line 1 to Line 3, guide the passengers who transfer from Line 3 to Line 1, and avoid passenger flow conflicts. 4.2 Operation Results of Optimized Simulation Model 4.2.1 Data Analysis of Average Transfer Time The transfer time between different lines is analyzed as follows. 1) Comparison of average transfer time between Line 3 and Line 1 According to the simulation operation statistics, the average transfer time between Line 3 and Line 1 remains unchanged. Compared with that before optimization, the number of passengers with the longest transfer time decreases by 3%, while the proportion of other passengers remains unchanged. For specific analysis: After the use of the iron barrier to separate the inbound/outbound passenger flow from the transfer passenger flow, the conflict point disappeared, and the large congestion that caused the escalator facilities to lose their service function disappeared during the peak hours. However, the transfer passenger flow from Line 3 to Line 1 accounted for a high proportion of the passenger flow at the station, and the goal was to ensure the
Simulation Study on Optimization of Passenger Flow Transfer Organization
75
passenger flow under the peak conditions, so the overall transfer time did not change significantly, but the maximum transfer time decreased. The optimization results retained the basic capacity for the peak hours. 2) Comparison of average transfer time between Line 1 and Line 3 The average transfer time between Line 1 and Line 3 is reduced by 5%, the proportion of passengers with medium transfer time is reduced by 7%, and the proportion of passengers with short transfer time is increased by 7%. Compared with the other direction, the optimization effect is better. For specific analysis: Before optimization, the transfer of passenger flow from Line 1 to Line 3 is mainly through the narrow passage between the column and the exit gate partition. Although the space above is large, according to the social force model, people like to take short cuts, and often walk by sticking to the column or wall, which is easy to waste space, squeezing the space of the opposite passenger flow, and the insufficient space of the opposite passenger flow will encroach on the passage, and the collision between the two waves of passenger flow will affect the efficiency. The paper uses the iron barrier to separate a new travel channel for the passenger flow in this direction. The passenger flow in this direction can add a new passenger flow channel on the basis of the previous channel. The proportion of passenger flow in this direction is less than that of the target passenger flow, and the passenger flow in the opposite direction is less affected because of the large space. The optimization effect achieved on this basis is better. 4.2.2 Port Traffic Data Analysis Comparison with previous, the escalator at the upper part of Line 1 has a significant fluctuation of passenger flow caused by the train entering and leaving the station. The passenger flow of the middle staircase of Line 1 increased significantly. The passenger flow of the escalator on the left side of Line 3 has increased significantly. The overall passenger flow of escalators has increased. For specific analysis: 1) The passenger flow in and out of the station accounts for a very small proportion of the total passenger flow. The escalator that undertakes the passenger flow in and out of the station is always idle, and the optimization effect has no obvious change. 2) The escalator presents normal passenger flow fluctuations, indicating that after the passenger flow is separated from the passenger flow at the iron barrier and the ascending passenger flow is combined with the inbound passenger flow, several nearby congestion points/conflict points are relieved. The upper escalator of Line 1 is within the normal capacity range. The escalator is the key gathering place of passenger flow, with important points, large passenger flow, many conflict points, easy congestion, difficult to optimize, and difficult to reflect the effect. Therefore, the optimization goal is to retain the basic traffic capacity at peak hours and achieve the optimization goal. 3) In order to dissipate the passenger flow diversion results of the upper escalator congestion point measures and relieve the pressure of the upper escalator passenger
76
Y. Chen et al.
flow, there is normal passenger flow fluctuation after the new passenger flow, which is within the range of traffic capacity, and the diversion measures are successful. 4) This point is similar to the escalator point on the upper part of Line 1, and both are important distribution points for transfer passenger flow. However, when the passenger flow is divided from the opposite passenger flow under the condition of deviation to ensure the transfer of Line 1 to Line 3, the passage increases and the arrival efficiency improves significantly. 5) The passenger flow of escalator transfer has increased to a certain extent, indicating that after the use of iron barrier to divide the passenger flow and delay the impact time of large passenger flow, several conflict points/congestion points near some main escalators/stairs of Line 1 have been alleviated, and the efficiency of passenger flow arrival has been improved. 4.2.3 Thermodynamic Diagram Analysis Comparison with previous, although there are still some conflict points and congestion points after the diversion, there is no large congestion after the optimization of the diversion. The operation of the station during peak hours is guaranteed. For specific analysis: The main objectives of the paper are two key points, namely, the upper escalator of Line 1 and the left escalator of Line 3. The specific objectives are to ensure the basic traffic capacity of the upper escalator of Line 1 and improve the traffic efficiency of the left escalator of Line 3 as a whole. According to the thermal diagram, the passenger flow aggregation form of the upper escalator of Line 1 is no longer a block aggregation centered on a conflict point, but a strip aggregation with a certain queuing form. As a diversion measure, the passenger flow to the escalator in the middle of Line 1 also increases orderly. The optimization effect of this point is achieved. In the passage between the two columns in front of the escalator on the left of Line 3, the blocky passenger flow gathered and disappeared, which was expressed as a strip queue. There were also passengers traveling in the free space at the upper part of the station, no longer wasting space, and the optimization effect was achieved.
5 Conclusion According to the analysis of simulation model operation data, it can draw the following conclusions. 1) This paper establishes an agent-based simulation model of Jinhu Square transfer station. Integrate the physical configuration of the station, passenger flow, social force performance of passenger flow, train inbound operation and other factors into one system, and conduct unified parameter adjustment. The simulation model can be quickly modified under different passenger flow conditions to adapt to different scheme comparison and selection, and facilitate the adjustment of optimization measures
Simulation Study on Optimization of Passenger Flow Transfer Organization
77
2) The paper uses the actual collected passenger flow data, through the use of agent-based Anylogic software for simulation, uses the thermal diagram and port flow monitoring to determine the location of passenger flow congestion points and conflict points. The simulation results show that under the operation of the social force model, the passenger flow performance selects the shortest path, which will lead to a large number of conflict points under this condition. This logical choice also causes congestion at some corners. Therefore, certain measures are needed to force the passenger flow to follow the established route to avoid unnecessary conflict points 3) The passenger flow transfer optimization mainly uses measures such as passenger flow organization and facility separation to control the passenger flow density at the bottleneck at a low level. However, through simulation, it is found that measures such as separation of organization and facilities can only disperse passengers to various points and then merge them in a staggered or delayed manner to reduce the impact on facilities in a short time, so as to reduce the density of passenger flow at or within the congestion points. The number of congestion points is reduced, but the main conflict points still exist. The passenger flow at the transfer channel is still saturated at peak hours, and the transfer time of passengers is only slightly increased. When transferring large passenger flow, other auxiliary means should also be used to help improve 4) The simulation model is built by using the Anylogic software, and the passenger flow parameters are substituted and sampled. The fuzzy station operation can be converted into specific visual data, and these data can be used as the station service level evaluation, which is conducive to the improvement of station facilities and management Acknowledgment. This project is supported by the second batch of school-level teaching team in 2019 “Applied Effective Teaching Design Teaching Team” (2019XJJXTD10) and Nanning University’s second batch of teaching reform project of specialized innovation integration curriculum (2020XJZC04).
References 1. Zhu, W.: Analysis of high-speed railway passengers’ queuing at the station based on anylogic simulation. Urban Rail Transit Res. 7, 133–137 (2020). (in Chinese) 2. Liu, W., Wang, F., Zhang, C., et al.: A simulation study of urban public transport transfer station based on anylogic. KSII Trans. Internet Inform. Syst. (TIIS) 15(4), 1216–1231 (2021) 3. Zhao, M.: Research on Passenger Flow Control Method and Simulation of Urban Rail Transit Transfer Station, pp. 5–6. Southwest Jiaotong University, Chengdu (2020). (in Chinese) 4. He, Y.: Research on the Design of Evacuation Stairs in Underground Commercial Buildings Based on Personnel Evacuation Simulation, pp. 10–12. Chongqing University, Chongqing (2018). (in Chinese) 5. Zhan, Y.: Real Time Task Evaluation of General Aviation Emergency Medical Transfer Based on Hybrid Simulation, p. 8. Nanjing University of Aeronautics and Astronautics, Nanjing (2019). (in Chinese) 6. He, Y.: Control of Transfer Passenger Flow in High-Speed Rail Metro Hub, p. 10. Southwest Jiaotong University, Chengdu (2018). (in Chinese)
78
Y. Chen et al.
7. Shi, H, Yang, Z.: Research on streamline organization optimization of urban rail transit transfer station based on anylogic. Comp. Transport. 42(06): 69–75+81 (2020). (in Chinese) 8. Zhao, J., Tyler, D.C.: Quantifying the impact of classification track length constraints on railway gravity hump marshalling yard performance with anylogic simulation. Int. J. Comput. Methods Exp. Meas. 10(4), 345–358 (2022) 9. Liu, Y., Song, Y.: Research on simulation and optimization of road traffic flow based on anylogic. E3S Web Conf. 360, 01070 (2022) 10. Yang, Y., Chen, J., Du, Z.: Analysis of the passenger flow transfer capacity of a bus-subway transfer hub in an urban multi-mode transportation network. Sustainability 12(6), 2435 (2020) 11. Omri, A., Omri, M.N.: Towards an efficient big data indexing approach under an uncertain environment. Int. J. Intell. Syst. Appl. (IJISA) 14(2), 1–13 (2022) 12. Aliyev, A.G.: Technologies ensuring the sustainability of information security of the formation of the digital economy and their perspective development directions. Int. J. Inform. Eng. Electr. Bus. (IJIEEB) 14(5), 1–14 (2022) 13. Saida, A., Yadav, R.K.: Review on: analysis of an IoT based blockchain technology. Int. J. Educ. Manag. Eng. (IJEME) 12(2), 30–37 (2022) 14. Samiul Islam, M., Islam, F., Ahsan Habib, M.: Feasibility analysis and simulation of the solar photovoltaic rooftop system using PVsyst software. Int. J. Educ. Manag. Eng. (IJEME) 2(6), 21–32 (2022)
Engine Speed Measurement and Control System Design Based on LabVIEW Chengwei Ju1 , Geng E. Zhang2,3(B) , and Rengshang Su1 1 Guangxi Special Equipment Inspection and Research Institute, Nanning 530219, China 2 College of Intelligent Manufacturing, Nanning University, Nanning 530000, China
[email protected] 3 Faculty of Engineering, University Malaysia Sabah, 88400 Kota Kinabalu, Sabah, Malaysia
Abstract. The engine speed signal is the main control signal of the engine electronic control fuel injection system, which plays a crucial role in the engine power, fuel economy and other performance. In this paper, the principle of engine speed measurement is analyzed, and a new engine speed measurement and control system is designed. The system is based on LabVIEW virtual instrument, using NI USB-6216 data acquisition card as the hardware foundation, collecting the speed signal output by magnetoelectric speed sensor, and using LabVIEW software to calculate and process the measured signal, then the actual speed is obtained. The experimental results show that the system has the characteristics of stable performance, clear data display, high real-time performance, high sensitivity, good precision, simple operation, simple program and so on. It plays a good role in monitoring the engine speed, and is conducive to improving the engine power and fuel economy. This system also provides a reference for other vehicles to design speed measurement and control system. In addition, man-machine interface operation is obvious humanized. Keywords: LabVIEW · Speed sensor · Measurement and control system
1 Instruction With the development of automobile industry, automobile performance has attracted more and more people’s attention, among which engine speed is an important indicator reflecting automobile power and economy [1]. RPM measurement technology has always been the basis of engine measurement technology, with the continuous efforts of predecessors, a variety of measurement methods have appeared [2]. However, there are some defects in the practical application of these methods. For example, most of the methods require the measurement sensor to be installed on the crankshaft, these methods are direct and effective, but they also bring a lot of inconvenience to engine manufacturing and maintenance [1]. In order to monitor the engine speed signal, many researches have been carried out on the monitoring of engine speed signal in the domestic and overseas. The sensors usually used to measure the rotate speed include Hall type, magnetoelectric type, capacitive type, photoelectric type and photoelectric encoder [3, 4], © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 79–89, 2023. https://doi.org/10.1007/978-3-031-36115-9_8
80
C. Ju et al.
and the corresponding measurement and control systems are also various. Measurement and control systems has the following kinds: C8051F340 MCU as the control center, the use of Fourier transform to realize the system signal analysis and processing, in order to measure the real-time rotate speed; With DSPF2812 as the control core, LabVIEW software realizes signal monitoring and data storage; With the embedded system as the core, the rotate speed value of the starter is calculated indirectly by analyzing the frequency information of vibration [2, 5, 6]. The design results of each system have achieved very good results in practical tests, with relatively high measurement accuracy [7–9]. For the above problems, according to the characteristics of the commonly used car magnetoelectric speed sensor design engine speed measurement and control system based on virtual instrument technology, in the case of without changing the hardware can add new features to the equipment, such not only can save expensive hardware, but also can further improve and perfect the automation of testing instruments. This design takes the engine speed measurement as the goal, LabVIEW as the software development platform, modularity as the design idea, chooses the Baojun 510 vehicle to do the experiment, the magnetoelectric crankshaft position sensor and PFI-6216 data acquisition card as the hardware basis, and finally realizes the real-time data acquisition, display and processing. The engine speed measurement and control system based on LabVIEW mainly take ordinary computers as the technology foundation. Software and hardware have adopted the modular design concept and ideas, data collection, display, storage and computing information processing are all realized through the computer. Such a design greatly improves efficiency, makes data processing more flexible, and has more powerful application functions. It not only has friendly interface but also has simple user operation, takes into account convenience in security and maintenance, has easy expansion in requirements, and significantly reduces manufacturing costs [2, 10].
2 Hardware System Design In order to complete the data acquisition and calculation of engine speed measurement and control system, a complete hardware system is needed. The hardware of the measurement and control system mainly consists of the following three parts: 1) Magnetoelectric speed sensor, which is mainly used to convert various physical signals into electronic pulse signals that can be received by the data acquisition card [11]. 2) Data set card, which is mainly used to connect the computer with external equipment and working environment, automatically collect non-power or power signal from the analog and digital unit of the sensor and other equipment to be tested, and send it to the host computer for analysis and processing. 3) Auxiliary equipment, including computers, keyboards, monitors, etc. They are used to realize online monitoring and online modification of parameters of measurement and control system, data storage, etc. [12].
Engine Speed Measurement and Control System Design
81
2.1 Rotate Speed Sensor The current rotate speed sensor types include Hall rotate speed sensor, photoelectric rotate speed sensor and magnetoelectric rotate speed sensor. At present, magnetoelectric rotate speed sensor is the most used rotate speed sensor, and its structure is shown in Fig. 1. It is composed of permanent magnet, coil, disk and so on.
Fig. 1. Measurement principle of magnetoelectric rotate speed sensor
When the rotating shaft rotates, the induction coil with magnetic circuit passing through will induce a pulse potential of a certain amplitude when the magnetic flux changes abruptly. Its frequency is: f=
z∗n 60
(1)
In the formula, z is the number of teeth of the magnetic wheel, n is the number of revolutions of the magnetic wheel in r/min. The crankshaft position sensor used in the experiment is electromagnetic pulse signal sensor, which is installed on the flywheel. The pulse signal is used to sense the position of the missing two teeth of the 58 teeth of the crankshaft timing disk to determine the speed of the engine and the relative position of the piston when the crankshaft rotates. The engine control unit uses the information from the crankshaft position sensor to generate timing ignition signals and injection pulses, which are then sent to the ignition coil and fuel injector, respectively. The actual rotate speed sensor of the experimental vehicle is shown in Fig. 2. The sensor is three-wire type, including one signal anti-interference shielding wire, one signal ground wire and one sensor signal wire. The connection between the sensor of the experimental vehicle and the car computer is shown in Fig. 3.
82
C. Ju et al.
Fig. 2. Magnetoelectric rotate speed sensor
Fig. 3. Connection diagram of magnetoelectric speed sensor and ECU
Fig. 4. NI USB 6216 Data acquisition card
2.2 Data Acquisition Card Rotate speed sensor measurement and control system requirements: more measurement parameters, high sampling rate, strong compatibility requirements, digital input and output [13]. According to the design requirements, the USB-6216 model data acquisition card is selected, and the communication interface of the data acquisition card is selected,
Engine Speed Measurement and Control System Design
83
that is, the interface mode. The communication mode between data acquisition card and computer mainly includes serial port (485 module, RS232), parallel port (PCI interface), USB and Ethernet. USB data acquisition card is characterized by plug and play, support hot swap, easy to carry. It is transmitted via USB bus data pass. USB is an external bus standard, conveniently used to regulate the connection and communication between computers and external devices. The appearance of the data acquisition card is shown in Fig. 4, and the main performance indicators of the card are shown in Table 1. Table 1. Data acquisition card parameters Sampling frequency
400 k/s
Digital input/output
24 digital I/O lines (5V TTL signal)
Input range
±0.2V, ±10 V
Analog output
Two-way
Input accuracy
16-bit
Output accuracy
16-bit
Analog input
16
Counter/timer
2
3 Software System Design The software design is mainly composed of LabVIEW block diagram, front panel design, data acquisition card and communication module. Data acquisition is the core part, its principle is to first sense the physical signal to be collected by the sensor, and then transmit the sensed signal to the converter, which converts the physical signal into the voltage signal that can be collected by the acquisition card, and then regulate and transmit it to the acquisition card. After the acquisition card goes through the process of amplification, sampling preservation, A/D conversion, etc., LabVIEW program reads the data through the DAQ acquisition channel, filters and calculates through function controls, and the computer displays the collected signals on the front panel of the system after software programming of the virtual instrument [14]. 3.1 Program Block Diagram Design The program of the detection system includes three modules: data acquisition, signal filtering, data calculation and display [15]. 1) Data acquisition module 2) It is composed of creating array virtual channels (functions). The sensor pulse signal is collected by DAQ data acquisition channel, and the collected pulse frequency is processed by array size (functions). The corresponding block diagram program design is shown in Fig. 5. 3) Signal filtering module The high and low frequency signals of the sensor are sorted according to the value, and the average value is calculated according to the highest and lowest frequency continuously. The sensor with interference signals is effectively filtered. The corresponding block diagram programming is shown in Fig. 6.
84
C. Ju et al.
Fig. 5. Data acquisition module
Fig. 6. Signal filtering module
4) Parameter setting module The sampling period can be set in this module. The quotient of input is calculated, and then the total frequency collected is divided by the number of pulses generated during a rotation of the crankshaft. Parameters can be flexibly set according to the different pulse numbers of a rotation of different sensors for convenient measurement, with good compatibility, as shown in Fig. 7. 5) Measurement and display module By creating the pulse frequency signal collected in the virtual channel, and manually configuring the function to output the specified type of data, so that when wiring an array to the function, this function can be automatically resized to display the index input of each dimension in an N-dimensional array, or by adjusting the node
Engine Speed Measurement and Control System Design
85
Fig. 7. Parameter setting module
Fig. 8. Measurement and display module
size, adding elements or subarrays, the wire board displays the default data type of the polymorphic function [16]. Each time data is entered from the wire, LabVIEW feeds the data from the wire into a buffer and draws a graph with all the data in the buffer. When the second data input, the input data will clear the data left in the buffer last time, and then send the data into the buffer for plotting. The data will be displayed in real time through the shape chart and the value display control, as shown in Fig. 8.
86
C. Ju et al.
3.2 Front Panel Design The front panel of the system is mainly composed of waveform display window, realtime data display window and function control button [17]. The background beautifying is filled with green, which can reduce the visual fatigue of operators, as shown in Fig. 9.
Fig. 9. Front panel
The data display window is arranged on the upper right side of the user interface, so that users can observe the changes of relevant data in real time at a glance. The parameter setting window is placed in the middle position to facilitate parameter setting. The waveform display window is arranged below, which can correspond to the real-time data and observe the changes of the data waveform in real time. The control switch is placed at the top of the data display window, and the eye-catching red button is adopted to facilitate data collection while observing data. In order to make the system easy to operate and observe, every parameter setting window, waveform display window, function button and other functions are marked with Chinese characters.
4 Experiment of Measurement and Control System 4.1 Experimental Conditions In the test, a Baojun 510 car was selected for the test, and it was found that the number of teeth of the reluctance ring of the crankshaft position sensor was 58 teeth, which was converted into a pulse number of 58 during one turn of the crankshaft rotation. Before the test of the measurement and control system, it was necessary to know the pulse number of one turn of the crankshaft rotation, and the pulse signal of the engine speed was measured by the data acquisition system during the test. The accurate real-time speed can be obtained by calculating the pulse times of one rotation. The system can adapt to the speed measurement of multiple different frequencies.
Engine Speed Measurement and Control System Design
87
4.2 Experimental Procedure 1) Start the vehicle and check whether the experimental vehicle runs normally; 2) Open the computer, start the test program, data acquisition card USB interface connected to the computer; 3) Connect the crankshaft position sensor signal wire and earth wire to the USB - 6216 type of data acquisition card; 4) Select the appropriate measurement channel in the software program; 5) Slowly increase the speed from the idle state, gradually increase the speed to 5000 RPM from 500 to one unit, record the measured speed value, stop the test, and the engine stops running. Click the “Stop” button. The operation display of the measurement and control system in the experiment is shown in Fig. 10.
Fig. 10. Display of running pulse signals in the experiment
4.3 Analysis of Experimental Results It can be seen from the experimental operation that the output pulse signal of the engine revolution sensor is stable. After comparing and analyzing the data of the engine instrument speed record and the measurement and control system, the measured real-time data is basically consistent with the actual speed, which accords with the working principle of the system.
5 Conclusion This paper analyzes the principle of engine speed test, selected the sensor, data acquisition card, calibrate the sensors, designed the front panel and program speed test system with LabVIEW software, and completed the measurement and control system real vehicle experiment. Through the engine speed test results show that the system is stable
88
C. Ju et al.
performance, high sensitivity, good accuracy, simple operation, simple, program, to achieve the requirements of the design target. In this paper, there are still many deficiencies in the further research and design of the speed measurement and control system to be improved, such as the multi-functional software measurement and control system, beautiful and simplified operation interface, etc. The experiment in this paper is done with the whole vehicle, the observed accuracy of the automobile instrument speed is 100r/min, this is in order to better reflect the accuracy of the measurement and control system, which can be tested on an engine bench with a digital tachometer. Acknowledgment. This paper is supported by: (1) Sub-project of Construction of ChinaASEAN International Joint Laboratory for Comprehensive Transportation (Phase I), No. GuiKeAA21077011–7(2) Automobile Electrical Appliances, The second batch of Nanning University “Curriculum Ideological and Political Education” demonstration course construction Project, No. 2020SZSFK19. (3) Diversified “Collaborative Ideological and Political Education” teaching team, No. 2022SZJXTD03.
References 1. Yang, X.: Development of Portable Automobile Engine Speed Measuring Instrument. Hefei University of Technology, Hefei (2014). (in Chinese) 2. Li, Y., Zhou, L.: Research on automobile engine speed measurement system. Indus. Control Comput. 23(04), 34–35 (2010). (in Chinese) 3. Chen, Y., Zhao, Y., Dai, J.: Motor vehicle engine tachometer calibration device. Shanghai Meas. Test 46(04), 11–14 (2019). (in Chinese) 4. Yang, C., Yang, L., Yao, Q.: A test system of vehicle operating condition parameters based on labVIEW. Tractor Agric. Transporter 37(05), 51–54 (2010). (in Chinese) 5. Xu, S., Wei, M.: Measurement method and experimental study of engine instantaneous speed. Automobile Technol. 10, 49–52 (2011). (in Chinese) 6. Song, G.: Design of speed measurement system based on 89C51. J. Weifang Univ. 6, 32–34 (2008). (in Chinese) 7. Huang, X.: Research and Development of engine Data Acquisition and Display System Based on LabVIEW. Tianjin Vocational and Technical Normal University (2017). (in Chinese) 8. Zhang, X.: Research and analysis of instantaneous speed measurement of vibration and noise engine. Sci. Technol. Innov. Appl. (01), 16 (2017). (in Chinese) 9. Cui, H., Zhu, L.: Research on automobile engine speed measurement and single cylinder power balance test. Sci. Technol. Innov. (07), 9–10 (2019). (in Chinese) 10. Wang, X.: Research on virtual instantaneous speed test system of engine based on LabVIEW. Agr. Equipment Veh. Eng. 51(11), 64–67 (2013). (in Chinese) 11. Xie, D., Chu, H.: Application analysis of magneto electric and hall type engine speed sensors. Intern. Combust. Engine Accessories (24), 60–61 (2020). (in Chinese) 12. Yue, Y., Zhao, Z.: Design of portable automobile engine speed measuring instrument. Instrum. Tech. Sensor (02), 39–42 (2019). (in Chinese) 13. Cheng, D., Jiang, W., Huang, Z., Zhang, K., Wang, W.: Design of measurement and control platform for engine speed sensor. J. Hubei Univ. Automotive Technol. 27(04), 14–17+23 (2013). (in Chinese) 14. Li, H., Guo, L., Fu, H.: Design of engine speed measurement system based on single chip microcomputer. J. Yunnan Agric. Univ. (Nat. Sci.) 30(02), 294–297 (2015). (in Chinese)
Engine Speed Measurement and Control System Design
89
15. Patel, D.M., Shah, A.K.: LabView based control system design for water tank heater system. Trends Electr. Eng. 7(3), 31–40 (2017) 16. Yang, H., Zhou, Q.: Pressure data acquisition and processing system based on LabVIEW and AVR microcontroller. Control Instrum. Chem. Ind. 37(11), 92–94 (2010) 17. Bo, L., Liu, X., He, X.: Measurement system for wind turbines noises assessment based on LabVIEW. Measurement 44(2), 445–453 (2011)
Research and Design of Personalized Learning Resources Precise Recommendation System Based on User Profile Tingting Liang, Zhaomin Liang(B) , and Suzhen Qiu College of Artificial Intelligence, Nanning University, Nanning 541699, China [email protected]
Abstract. Online education is developing rapidly, and the scale of digital learning resources is increasing sharply. In the process of learning, learners are faced with relatively redundant resources and information drift, which easily leads to blindness and ineffectiveness in learning. For this reason, from the perspective of students’ individualized learning needs, using artificial intelligence and data analysis and mining technology, the corresponding portraits are constructed according to students’ basic information, online learning behavior, resource information, domain knowledge, etc. This paper puts forward a mixed recommendation mechanism for different learning stages, which takes the learner as the center, combines four methods of recommendation based on learning style, knowledge map, resource preference and behavior sequence of learning partners, establishes an online precise recommendation system for learning resources, matches the different learning needs of learners, provides precise recommendation-related services for learners and managers, and facilitates the personalized learning and growth of learners. Keywords: User portrait · Learning resources · Recommended algorithm · Recommendation system · Personalized Recommendation
1 Introduction In the large data environment, artificial intelligence, cloud computing technology and other rapid development, the field of education has gradually entered the intelligent stage. This puts forward new requirements for meeting the personalization of user learning, guaranteeing collaborative learning, and personalization of recommendation content [1]. Digital resources and platforms are the products of the deep integration of “Internet + Education”. At the present stage, more emphasis is placed on the informationization and networking of learning resources, often ignoring the construction of the logical connection of related learning resources. A large number of learning resources are relatively scattered in different system platforms, and the resources of related learning content cannot form a synergy. This often results in the fragmentation and fragmentation of the resources that learners acquire. As a result, learners are prone to overload information or get lost in the process of learning. The learners’ systematic learning and growth are facing many challenges. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 90–100, 2023. https://doi.org/10.1007/978-3-031-36115-9_9
Research and Design of Personalized Learning Resources
91
The way learners acquire learning resources has gradually changed from active retrieval to automatic recommendation by the learning system. In the past, digital learning resources were mainly pushed in three ways: Top-N recommendation, keyword query and the latest resource recommendation, which improved the hit rate of knowledge entity search and the efficiency of learning resources, but the accuracy of content-based recommendation is still insufficient. In addition, after the completion of the current stage of learning, there will be difficulties in the correct orientation of learning and systematic learning of courses. In actual teaching, there are three types of online education in colleges and universities: MOOC platform, practical teaching platform and teaching process management platform. Table 1 is a simple survey of the platforms widely used in the three types of courses. Table 1. Broadly used platforms and surveys Type
Name
Description
Search results with “program design” as the keyword
MOOC Platform
China University MOOC
Undertake the task of national high-quality open courses of the Ministry of Education, and provide MOOC courses of well-known universities in China to the public
1517 related courses
MOOC Platform
Superstar Learning
It is the teaching software that many universities choose to cooperate with for online teaching
90 courses related to the demonstration teaching package
Practice teaching platform
EduCoder
It is the official cooperation platform of MOOC Alliance Practical Teaching Working Committee of China’s University Computer Education
61 related practical courses
Teaching process management
Rain class
The teaching tools are Other course resources skillfully integrated into cannot be retrieved PowerPoint and WeChat, and the hybrid teaching is simple and convenient
92
T. Liang et al.
Learners can retrieve multiple courses on the first two types of platforms. Each course contains videos, documents, pictures and other types of resources. Each course contains dozens or even hundreds of resources. The platform is rich in resources. Learners need to click into the curriculum to further search for resources. It is difficult to directly transfer big data in the teaching process between different courses. The sharing and utilization rate of teaching resources is low, and there is a lack of personalized learning resource recommendation services. How to integrate learners’ personalized characteristics, degree of knowledge construction, learning needs and scientific rationality of technology, incorporate some scientific, applicable and innovative technologies to improve the accuracy and efficiency of recommendation, realize learners’ individualized acquisition of demand resources and professional systematic learning, in order to reduce learning blindness and improve the logic and relevance of learning process, is still a problem worth further discussion.
2 Research on Recommendation of Educational Resources 2.1 User Portrait User portraits are an important tool for accurate service recommendation. User portraits are descriptions of the principal features and needs. They are authentic, dynamic, interactive and clustered, which can personalize services and achieve the purpose of successful recommendation. In order to better achieve personalized recommendations for learners, user portraits are constructed from two aspects: user data collection and behavior data mining, combining learner characteristics. Wang Shu based on the “Chinese Bridge” learner data proves that using user data to build user portraits can improve the learner’s experience in learning recommendation, which has practical significance [2]; L Qin By analyzing the user behavior of the resource platform and mining rules from it, combining with the user personality characteristics, dynamic and static data, it can improve the understanding of user behavior and provide personalized service quality [3]. S Shrestha uses the learner’s interactive data to determine the learning style and then constructs a picture of the learner to support personalized teaching [4]. T Xian builds user portraits by analyzing logs to get users’ acceptance and preferences for content [5]. Teaching designers design content based on portraits to improve the efficiency and effectiveness of the course. 2.2 Research on Personalized Recommendation of Educational Resources By sorting out the relevant research on Learning Resource Recommendation in the field of education, it can be found that it is mainly divided into three levels: recommendation algorithm, recommendation model and recommendation system. These three are inseparable. Recommendation algorithm is the basis of the latter two, and its recommendation effect and evaluation are directly or indirectly affected by the latter two. The three commonly used recommendation algorithms for learning resources in the field of education are content-based, collaborative filtering and knowledge-based recommendation algorithms. A single recommendation algorithm often fails to achieve a
Research and Design of Personalized Learning Resources
93
comprehensive and accurate recommendation for learning content, so a hybrid push of multiple recommendation algorithms is often used in both business and education fields. Multidimensional integration of technologies and algorithms based on tag recommendation, social network, clustering, neural network, in-depth learning, knowledge mapping, hybrid recommendation is a hot spot for scholars [6]. Combining the three technologies of association rule-based recommendation, content filtering recommendation and collaborative filtering recommendation, we can actively recommend resources and information suitable for users’learning needs according to their personalized characteristics, such as learning interests, preferences and learning needs. Whether the construction of the recommendation model meets the requirements requires selecting the algorithm according to the recommendation mechanism. The construction of Learning Resource Recommendation Model and the design of recommendation strategy focus on the aspects of user attributes and user preference mining, learner characteristics analysis, intelligent recommendation and multivariate model building [8– 10]. The recommendation of learning resources, learning partners and learning paths are merged from several dimensions such as recommendation algorithm, learning scenarios and recommendation content, which helps to promote learning resources through multiple channels. The recommendation system can realize and test the rationality of the recommendation model and the accuracy of the recommendation algorithm. At present, most of the applications of learning resource recommendation system are mainly concentrated in the vertical subdivisions of Library materials, an online learning platform, a course, etc. Few studies are based on cross-domain and multi-source data [11]. It mainly considers the combination of multiple dimensions, such as algorithm recommendation strategy, recommendation content and learning real situation, to better improve the recommendation effect and the level of multiple learning services. From the perspective of students’ individualized learning needs, this paper collects resources construction and electronic imprint of students’ learning, uses artificial intelligence and big data technology to analyze and portray learning style and preference data and form a picture of students, designs a recommendation system, realizes the orientation of learning direction and methods of personalized recommendation, and accurately recommends learning resources for them, which has a strong epochal and development value.
3 Key Links to Building a Learner Portrait Model Student portrait technology is built on the basis of large educational data. Due to the multidimensional and cross-fusion of data information, in order to achieve the efficiency of resource utilization, it is necessary to overcome the problems of data redundancy, heterogeneity and missing [12]. Filter information according to different needs, and use cluster coupling analysis, deep mining and other related technologies to complete the analysis and processing of learner data information. The main steps of building a learner data model include data collection, analysis, presentation and application (evaluation, prediction, intervention). In data fusion, the learner portrait technology needs to obtain the learner’s dynamic feature information at multiple levels, and use the multidimensional
94
T. Liang et al.
fusion technology to quantify all aspects of the information of the students, analyze and identify the information with objective data, ensure that all kinds of data complement each other, and achieve the fusion of heterogeneous data from multiple sources [13], so as to ensure the comprehensiveness and integrity of the learner’s information. In terms of technical implementation, the natural language processing (NLP) algorithm is generally used to synthesize the various characteristics of the learner to form a fullcycle, full-process portrait data of the learner. Among them, learning behavior feature information is usually extracted through behavior labels, and the establishment of labels mainly through three methods: meta-analysis-based methods, statistics-based methods, and correlation-based analysis methods.
4 Precise Recommendation of Personalized Learning Resources Based on User Portraits 4.1 Technical Framework The main way to realize the personalized learning resource recommendation service is to collect the specific behavior data generated by the learner using the learning interaction system during the improvement of the learner portrait model, compare the similar learner attribute characteristics, recommend the learning resources that may be needed, and use the Apriori association algorithm to find the frequent itemset and association rules [14]. The online teaching platform covers both learning resources and students’ data information. By digging and analyzing these information in depth, it can effectively compare the learning and resource needs of the learners, and then realize the labeling and unit processing of learning resources. At the same time, it systematically compares the learner’s characteristic data, extracts the learner’s portrait data, and establishes the learner’s tag to mark the specific individual attributes of the learner. Then, it digs deeply into the learning resources of interest to the learners and their groups, matches the data collection of the learners and the learning resources, establishes the association rules between the learner label system and the learning resources, improves the recommendation accuracy of the course resources by continuously optimizing the existing learning resources, and combines the collaborative filtering algorithm of the Association rules, so as to help the students obtain the learning resources better, thus improving the learning efficiency of the students. In the system, learners can further improve their own information, take the initiative to assess learning style, pre-test course knowledge, and so on. They can view their own level of development, and compare the level of professional needs, and achieve their own growth based on the personalized learning resources recommended by the system. 4.2 Recommendation Policies The main basis for the accurate recommendation of personalized learning resources is the learner’s portrait. Therefore, when constructing the learner’s portrait, the learner’s personal characteristics, interests, and dynamic information of learning activities should be fully acquired, the points of interest and learning direction should be analyzed and
Research and Design of Personalized Learning Resources
95
identified, combined with their existing knowledge level, the learner should be scientifically recommended for learning resources, and even the learning path planning. There are four main strategies for precise recommendation of personalized learning resources [15]. Recommendation based on learning style: When the learner completes the task of assessing learning style with good reliability, the four dimensions of Felder-Silverman model are used to divide the learning style of the learner, which are information processing, information perception, information input and information understanding, and then the system makes corresponding recommendation for learning resources according to the learner’s learning style. Recommendation based on knowledge map: first, build a learning resource classifier to classify resources. When new learning resources enter the resource pool, the classifier classifies them and puts them into the corresponding learning resource pool [16]. When making a recommendation, search the knowledge labels of the students’ professional information or currently viewed resources in the feature library of the knowledge map, use the similarity algorithm to extract the knowledge points related to the resource knowledge labels, and then find the learning resources matching the recommended knowledge points in the learning resource database. Recommendation based on resource preference: After the learner has a certain learning record, the weight of each kind of label value under the learner can be calculated by using TF = IDF algorithm, thus the learner’s resource preference can be obtained, and the system can recommend the corresponding type of resources according to the learner’s resource preference [17]. Recommendation based on learning partner behavior sequence: record the sequence of learners’ behaviors in the set and form a status sequence, use clustering method to calculate similar user (learning partner) groups, and then calculate the similarity of behavior sequence in the cluster space, including state value similarity, state transition similarity and state order similarity [18]. It then filters the learning resources used by the learning partners in their subsequent States and recommends them to the current users. Because each recommendation method has its advantages and disadvantages and is suitable for a specific scenario, and the process of online learning is also a dynamic and changing process, it is not appropriate to consider only one recommendation method for learning resource recommendation. Therefore, a mixed recommendation mechanism is necessary.
5 Precision Recommendation System Architecture for Online Learning Resources Aiming at precise recommendation of learning resources, the system is divided into four layers from bottom to top: data layer, data analysis layer, recommendation calculation layer, and human-computer interaction layer. The overall system architecture design is shown in Fig. 1.
96
T. Liang et al. Human Computer Interaction Layer Resource operation learner
Resource details and statistics controller
Knowledge map navigation Discussion Posting
Details and statistics of learners Recommended model weight settings
Recommended Computing Layer Mixed recomme ndation
Learning style based recommendations
Recommendation based on knowledge map
Recommendation based on resource preference
Recommendation based on learning partner behavior sequence
Data Analysis Layer Learner analysis module
Learner Portrait Resource preference analysis
Resource analysis module
Analysis of learner similarity
Resource portrait Resource Similarity Analysis
Knowledge map construction
Domain knowledge Knowledge analysis and organization
Data Layer Learner Basic information of learners feature Evaluation information library
Learner behavior information
Learning resource feature library
Basic resource information Resource tag: category, knowledge, etc
Knowledge points Knowledge feature Relationship between base knowledge points
Fig. 1. Architecture of personalized learning resources precise recommendation system
5.1 Data Layer The data layer is used for portrait modeling and portrays the characteristics of learners, resources and knowledge. It consists of three modules: the learner feature library, the learning resource feature library and the knowledge feature library. 5.1.1 Learner Feature Library Stores the learner’s characteristic information, including the basic information of the learner, the assessment information, and the learner’s behavior information. Basic information is inherent to each learner and does not change over time, such as student number, gender, specialty, etc. The assessment information does not change much in a short time, but there are also situations in which the learners do not complete it carefully, resulting in large differences between the data before and after, such as the assessment of learning style, the pre-test of curriculum knowledge, etc. The characteristics of learners’ behavior change significantly over time, such as browsing, collecting, commenting, downloading and posting resources.
Research and Design of Personalized Learning Resources
97
5.1.2 Learning Resource Feature Base It consists of the basic information of the resource and the label information of the resource. The basic information of the resource includes subject, resource type (such as video, audio, document, PPT, picture, etc.), storage location (local path, web address), reading amount, favorites, creator, source, language (Chinese, English), etc. Resource label information is a general description of the content and form of learning resources, including categories (such as theory, practice, scenario cases, etc.), owning knowledge (knowledge blocks or points), and other (such as formula derivation, programming language, commentary vocabulary). 5.1.3 Knowledge Feature Base Is the representation of the knowledge map stored in a computer-understandable way, usually with a head, relation, tail tuple representing the data of the knowledge map. Includes knowledge points (head entities, tail entities) and the relationship between knowledge points (e.g., contained, contained, homologous). 5.2 Data Analysis Layer This layer is mainly used to process the data of the data layer. It quantifies, statistics and models the three feature libraries of the data layer, corresponding to three modules: learner analysis, resource analysis and knowledge map building. 5.3 Learner Analysis Module This includes student portraits, resource preference analysis and similarity analysis between learners. The portrait of the learner describes the learning characteristics of the learner from various perspectives, and constructs the personalized portrait of the learner quantitatively and qualitatively [19]. It includes learning style, learning blocks, points of knowledge already learned, difficulties in knowledge, sequence of commonly used modules in learning, grade of achievement, etc. The learner can understand and master his own learning situation through portraits so as to adjust learning strategies. Resource preference analysis, through quantitative and statistical learning style test data and learner behavior data, tends to choose the content and form of learning resources preference, preferences and other preferences. For example, users A like learning resources with video as the carrier, while users B prefer learning resources with theoretical type. The calculation formula of resource preference similarity is: N(X) ∩ N(Y) Sim_Area(a, b) = con(X, Y) = √ N(X) × N(Y)
(1)
where, N (X) and N (Y) are the number of keywords in the resource type vector browsed by users a and b. The similarity analysis between learners is the basis for subsequent modeling of learning partner recommendation. The degree of correlation between learners is calculated based on the learner’s characteristic information to determine the similarity between
98
T. Liang et al.
learners. Similar learners are referred to as “learning partners” [20]. Finally, the learning resources selected by “learning partners” can be recommended to the current user. The calculation formula of learner similarity is: n 1 (Ai × Bi ) (2) Sim_Info(a, b) = con(A, B) = n n 2 2 1 (Ai ) × 1 (Bi ) where, A and B are the user attribute vectors of users a and b, and Ai and Bi are the attribute values of each dimension of the two users. 5.4 Resource Analysis Module Includes Resource Portrait and Resource Similarity Analysis A resource portrait, which describes the content and form of a resource in a hierarchical and refined vocabulary (tags). The tags come from keywords added by the resource creator, keywords extracted from comments, and tags extracted from the tag library by applying semantic calculations. Resource similarity analysis is the basis for subsequent learning resource recommendation modeling. Compute the degree of correlation of resources based on the tag characteristics of resources to determine the similarity between resources and recommend similar resources to current users. The construction of knowledge map mainly aims at domain knowledge. The main steps are: defining domain category, extracting terminology, extracting entity and relationship, classifying entity and relationship types, organizing knowledge framework, and completing knowledge attributes. 5.4.1 Recommended Computing Layer At the beginning of platform learning, learning resources are recommended using knowledge map-based or learning style-based methods, and the weights of these two methods on the reverse order of proposed resources are adjusted according to whether the learner has completed a reliable learning style test or not. With the increase of the learning time and interaction of the learners, the mining analysis of the learners’ learning behavior is carried out. The collaborative filtering recommendation is mainly based on the sequence of learning partner behaviors and supplemented by the recommendation based on resource preferences.The recommended system calculation formula is: ∀ ∈ C, S* = arg maxs∈S u(c, s)
(3)
where, C is the collection of all users, S * is the final collection of recommended items, S is the collection of items to be recommended, and u (c, s) is the recommendation of item s to user c. 5.4.2 Human-Computer Interaction Layer The application layer is the outermost layer of the platform and the interactive interface between the learner and the platform. In the application layer, the learner can browse
Research and Design of Personalized Learning Resources
99
the recommended resources, collect, download and browse the resources, select the corresponding resource classifications of knowledge elements according to the navigation of the knowledge map, and exchange discussions with teachers and peers to express their own views. Resource managers can view resource details and learner details, make intelligent assessments and scientific management based on statistical analysis results, and adjust the weight of objects in the recommended model.
6 Conclusion Accurate recommendation system for personalized learning resources based on user portraits can help students’ personalized learning and growth. This paper combs the characteristics of learners, resources and knowledge from the perspective of learners’ personalized learning needs, and constructs a portrait. On this basis, in light of the different learning stages of learners, this paper combines four personalized recommendation strategies to design a mixed recommendation model and a precise recommendation system for personalized learning resources. By implementing the “personalized” and “accurate” recommendation of learning resources, we can serve the orientation and personalized learning of learners. With the rapid development of personalized recommendation system and user portrait theory, the recommendation of learning resources will receive more and more attention in the future online learning. In terms of the theoretical model, recommendation mechanism and practical application of the personalized recommendation system for learning resources, there are still many areas that need to be improved continuously and need to be improved and further explored. In the follow-up work, the recommendation system will also be further improved and optimized to make it have better recommendation effect. Acknowledgment. This project is supported by the Basic Ability Improvement Project for Young and Middle-Aged Teachers in Guangxi Colleges and Universities (2021KY1804, 2021KY1800).
References 1. Gu, X., Li, S., Li, R.: International vision of artificial intelligence innovation application – Prospective progress and future education prospects of NSF Institute of Artificial Intelligence. China Distance Educ. (12), 1–9+76 (2021) 2. Wang, S.: Study on Recommendation of Personalized Learning Path for Portraits of Users of Chinese Bridge. Yunnan Normal University (2021) 3. Qiu, L., Leung, K.Y., Jun Hao, H.O., et al.: Understanding the psychological motives behind microblogging. Stud. Health Technol. Inform. 154, 140–144 (2010) 4. Shrestha, S., Pokharel, M.: Determining learning style preferences of learners. J. Comput. Sci. Res. 3(1), 33–43 (2021) 5. Xian, T.: Practice of sandbox game in higher education based on graphic and game programming environment. In: Stephanidis, C., Antona, M. (eds.) HCI International 2020 - Posters: 22nd International Conference, HCII 2020, Copenhagen, Denmark, July 19–24, 2020, Proceedings, Part II, pp. 356–364. Springer International Publishing, Cham (2020). https://doi. org/10.1007/978-3-030-50729-9_51
100
T. Liang et al.
6. Yinli, S., Sun, Y.: Hot spots, trends and Inspirations of domestic digital learning resource recommendation algorithms. J. Yunnan Normal Univ. (Nat. Sci. Ed.) 42(03), 60–66 (2022) 7. Zhu, H., Liu, Y., Tian, F., et al.: Across-curriculum video recommendation algorithm based on a video-associated knowledge map. IEEE Access 6, 57562–5757 (2018) 8. Lifeng, H., Li, C.: User cold start recommendation model combining user attributes with project popularity. Comput. Sci. 48(2), 114–120 (2021) 9. Mou, Z., Wufati: Study on recommendation of personalized learning resources based on Learner Model in e-book bag. Audio-visual Educ. Res. 36(1), 69–76 (2015) 10. Lina, Y., Yonghong, W.: Contextualized intelligent recommendation for universal learning resources. Res. Audio-visual Educ. 35(10), 103–109 (2014) 11. Peng, R.: Intelligent portrait construction system and application for college students based on multi-source data. Compu. Programm. Skills Maintenance (09): 165–168 (2022) 12. Wang, Y., Yang, L., Wu, J., et al.: Mining multi-source campus data: an empirical analysis of student portrait using clustering method. In: 2022 5th International Conference on Data Science and Information Technology (DSIT), pp. 01–06. IEEE (2022) 13. Zhang, L., Xie, Y., Xidao, L., et al.: Multi-source heterogeneous data fusion. In: 2018 International conference on artificial intelligence and big data (ICAIBD), pp. 47–51. IEEE (2018) 14. Wang, H., Fu, W.: Personalized learning resource recommendation method based on dynamic collaborative filtering. Mobile Netw. Appl. 26, 473–487 (2021) 15. Almu, A., Ahmad, A., Roko, A., et al.: Incorporating preference Changes through users’ input in collaborative filtering movie recommender system. Int. J. Inform. Technol. Comput. Sci. (IJITCS) 14(4), 48–56 (2022) 16. Geng, X., Deng, T.: Research on intelligent recommendation model based on knowledge map. J. Phys.: Conf. Ser. IOP Publishing 1915(3), 032006 (2021) 17. Qaiser, S., Ali, R.: Text mining: use of TF-IDF to examine the relevance of words to documents. Int. J. Comput. Appl. 181(1), 25–29 (2018) 18. Chen, M., Yang, X.P., Liu, T.: A research on user behavior sequence analysis based on social networking service use-case model. Int. J. u-and e-Serv., Sci. Technol. 7(2), 1–14 (2014) 19. Vandewaetere, M., Desmet, P., Clarebout, G.: The contribution of learner characteristics in the development of computer-based adaptive learning environments. Comput. Hum. Behav. 27(1), 118–130 (2011) 20. Erkens, M., Bodemer, D.: Improving collaborative learning: guiding knowledge exchange through the provision of information about learning partners and learning contents. Comput. Educ. 128, 452–472 (2019)
Mixed Parametric and Auto-oscillations at Nonlinear Parametric Excitation Alishir A. Alifov(B) Mechanical Engineering Research Institute of the Russian Academy of Sciences, Moscow 101990, Russia [email protected]
Abstract. The development of the theory of oscillatory processes, taking into account the properties of the energy source, contributes to improving the accuracy of calculating real objects for various purposes, the optimal choice of the power of the energy source to save it. Mixed parametric and self-oscillations under nonlinear parametric excitation in the interaction of an oscillatory system with an energy source are considered. The solution of the equations of motion is constructed using the method of direct linearization of nonlinearity. The conditions of stability of stationary modes of motion are derived. To study the effect of nonlinear parametric excitation on the properties of vibrations, calculations were carried out, the results of which were compared with the results of linear excitation. The results obtained show the difference between the dynamics of systems with linear and nonlinear parametric excitations, which has a quantitative and qualitative character. At the same time, quantitative differences are more significant. Keywords: nonlinearity · parametric oscillations · self-oscillations · direct linearization · limited excitation · energy source
1 Introduction Human civilization is inexorably moving towards an environmental crisis on a global scale, which, along with other factors, is promoted by the growth in the amount of energy consumed, and its reduction can make a certain contribution to solving environmental problems. In this context, the systematic theory of oscillatory systems with energy sources, created by V.O. Kononenko [1] and further developed by his followers [2–7, etc.], has become relevant again. It served as the basis for the emergence of a new direction in the theory of oscillations. In the work [8], the connection of environmental problems with metrology, etc. is noted. During the operation of a number of objects for various purposes, parametrically excited oscillations occur. Among them, one can indicate, for example, pendulums, rods, helical springs, rotating shafts, measuring instruments, cardan transmission, drive systems, gears, railway bridges, ship masts, etc. Therefore, the study of parametric oscillations is important for solving applied problems of solid mechanics, theory of mechanisms and machines, structural mechanics, electrodynamics, optics (in particular, © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 101–108, 2023. https://doi.org/10.1007/978-3-031-36115-9_10
102
A. A. Alifov
lasers), etc. In this context, we note the monograph [9], in which parametric oscillations with linear and nonlinear parametric excitation and an ideal energy source are considered. Oscillations under nonlinear parametric excitation, taking into account the properties of the energy source, are considered only in [10], at least, other works are unknown to the author. According to the classification described in [2], in [10] the parametric type of oscillations was studied, and below the class of mixed (interacting types) parametric and self-oscillations is considered - a self-oscillating system with nonlinear parametric excitation and an energy source of limited power. The purpose of this work is to develop the theory of mixed oscillatory processes taking into account the properties of an energy source for calculating real objects for various purposes. As is known, a number of approximate methods (averaging, energy balance, harmonic linearization, etc.) of nonlinear mechanics are used to solve nonlinear equations [11–17, etc.]. They are characterized by labor intensity, increasing with increasing degree of nonlinearity, and the presence of this problem for the analysis of coupled oscillatory networks is indicated in [18]. Among them, the asymptotic method of averaging has received wide application [12]. The direct linearization method (DLM) differs significantly from them, which is characterized by rather small labor and time costs and ease of use. Such features are very important in the design and calculation of real technical devices. And also, according to the DLM, one can obtain the final calculated ratios, regardless of the specific type of nonlinear characteristic [20–26, etc.]. For example, if we take yn , then according to known methods it is impossible to obtain any final calculated ratios for an unknown value of n, a specific value of n = 2,3, is necessary… This drawback is absent in the DLM. Comparison of DLM and known methods is given in [20, 24, etc.]. Therefore, using it, solutions of the nonlinear equations considered in this paper are constructed.
2 System Model and Equations of Motion Let us consider a model of a mechanical self-oscillatory system, which is widely used to analyze self-oscillations in a mechanical system [1–3, 11, etc.]. It well describes selfoscillations under the action of friction, which can occur during the operation of various engineering objects (textile equipment, guides for metal-cutting machines, brakes, etc. [27, 28, etc.]). Taking into account the nonlinear parametric excitation bx3 cos vt, b = const, ν = const, the equations of motion of the system are: m x¨ + k0 x˙ + c0 x = T (U ) − F(x) − b x3 cos vt J ϕ¨ = M (ϕ) ˙ − r0 T (U )
(1)
where c0 = const and k0 = const are, respectively, the spring stiffness and damper ˙ resistance coefficients, T (U) is the self-oscillating friction force, U = V − x˙ , V = r0 ϕ, r0 = const, F(x) is the nonlinear part of elasticity, J = const is the total moment of inertia of the rotating parts, M (ϕ) ˙ is the torque of the energy source, ϕ˙ is the rotation speed. In practice, the friction force T (U) is widely distributed in a non-linear form T (U ) = T0 (sgnU − α1 U + α3 U 3 )
(2)
Mixed Parametric and Auto-oscillations
103
which was also observed under space conditions [29]. Here α1 , α3 are positive constants, T0 is the normal reaction force, sgnU = 1 for U > 0 and sgnU = −1 for U < 0. In the case of relative rest U = 0, −T0 ≤ T (0) ≤ T0 takes place. Let’s nonlinear elasticity in the form of a polynomial F(x) = γs xs , s = 2,3,4,…. s
Using DLM, we replace this polynomial with a linear function F∗ (x) = BF + kF x
(3)
where BF and kF are the linearization coefficients defined by the expressions Ns γs as , s = 2, 4, 6, . . . (s is even number) BF = s
kF =
N s γs as−1 ,
s = 3, 5, 7, . . . (s is odd number)
s
Here a = max |x|, Ns = (2i + 1)/(2i + 1 + s), N s = (2i + 3)/(2i + 2 + s), i is a linearization accuracy parameter. We also use DLM for the friction force T (U), part of which we represent in the form T∗ (˙x) = BT + kT x˙
(4)
where BT = −α1 V + α3 V 3 + 3α3 N2 V υ 2 , kT = α1 − 3α3 V 2 − α3 N 3 υ 2 , υ = max |˙x| N2 = (2i + 1)/(2i + 3), N 3 = (2i + 3)/(2i + 5) Equations (1) taking into account (3) and (4) take the form m x¨ + k x˙ + cx = B + T0 sgnU − b x3 cos vt J ϕ¨ = M (ϕ) ˙ − r0 T0 (sgnU + BT + kT x˙ )
(5)
where c = c0 + kF , k = k0 − T0 kT , B = T0 BT − BF
3 Construction of Solutions of Equations The solution of the linearized DLM equation can be constructed by two methods [20], one of which is the method of replacing variables with averaging. Of these, the method of replacing variables with averaging makes it possible to study stationary and nonstationary processes. We use the standard form relations obtained for a linearized nonlinear equation of a fairly general form. One of the special cases of such a general type of equation is the first (5) and the standard form allows you to write out its solution. And to solve the second (5), we apply the averaging procedure [22], in which the velocity ϕ˙ is replaced by its averaged value . The sign of the relative velocity U in the characteristic of the friction force T (U ), which can be either positive or negative, determines the nature of the oscillations [2]. Therefore, we will consider these two cases separately.
104
A. A. Alifov
As a rule, the main interest in the analysis of oscillation dynamics is the main resonance, which is called the main parametric resonance under parametric influence. The solutions in this case have the form x = a cos ψ, x˙ = −ap sin ψ, ψ = pt + ξ, p = ν/2, ϕ˙ =
(6)
and in this regard, expressions (4) will have υ = ap, and instead of V, the expression u = r0 . Using the noted standard form of the DLM relations, we have the following equations for (5) in the case of u ≥ ap: ak ba3 da =− + sin 2ξ dt 2m 8pm dξ ω 2 − p2 ba2 = + cos 2ξ dt 2p 4pm r0 u du = M ( ) − r0 T0 (1 + BT ) dt J r0
(7’a)
At speeds of u < ap, the second equation (phases) in (7’a) remains unchanged, and the other two have the form ba3 4T0 a da 2 2 2 + k+ a p − u =− sin 2ξ (7’b) dt 2m πa2 p2 8pm r0 u du r0 T0 = M ( ) − r0 T0 (1 + BT ) − (3π − 2ψ∗ ) dt J r0 π where ω2 = ω02 + (kf /m), ω02 = c0 /m, ψ∗ = 2π − arcsin(u/ap). The frequency difference ω0 − p in the resonance region is quite small, which is why it is possible to accept (ω02 − p2 )/2p ≈ ω0 − p. The conditions a˙ = 0, ξ˙ = 0 and u˙ = 0 give relations for determining the stationary values of the amplitude, phase of oscillations and velocity of the energy source. At u ≥ ap, these ratios are as follows: 16p2 k 2 + 4m2 (ω2 − p2 )2 = b2 a4 tg2ξ = −
4pk 2m (ω2 − p2 )
(8)
M (u/r0 ) − S(u) = 0, S(u) = r0 T0 (1 + BT ) where S(u) represents the load on the energy source, taking into account that in the expression BT in (4), instead of V, there will be u = r0 . In the case of u < ap and small external forces, an approximate equality ap ≈ u takes place for the magnitude of the stationary oscillation amplitude, and the load S(u) on its basis can be determined by the expression (9) S(u) ≈ r0 T0 1 − α1 u + α3 (1 + 3N2 )u3 The stationary values of the velocity u are determined by the intersection points of the curves M (u/r0 ) and S(u).
Mixed Parametric and Auto-oscillations
105
4 Stability of Stationary Movements Having composed the equations in variations for (7’a) and (7’b) and using the RouthHurwitz criteria, we have the following conditions for the stability of stationary motions D1 > 0, D3 > 0, D1 D2 − D3 > 0,
(10)
where D1 = −(b11 + b22 + b33 ), D2 = b11 b33 + b11 b22 + b22 b33 − b12 b21 − b23 b32 − b13 b31 D3 = b11 b23 b32 + b12 b21 b33 − b11 b22 b33 − b12 b23 b31 − b13 b21 b32
For u ≥ ap speeds we have r 2 T0 ∂BT ∂BT r0 (Q − r0 T0 ), b12 = − 0 , b13 = 0 J ∂u J ∂a 3ba2 ∂kT T0 a ∂kT 1 , b22 = − (k − aT0 )+ sin 2ξ = 2m ∂u 2m ∂a 8pm ba3 ba b23 = cos 2ξ, b31 = 0, b32 = cos 2ξ 4pm 2pm ba2 sin 2ξ b33 = − 2pm
b11 = b21
and with u < ap undergo changes only T0 r02 ∂BT 2r0 T0 2u r0 ∂BT
−
+ b11 = , b12 = − Q − r0 T0 I ∂u J ∂a π a 2 p2 − u 2 π a a 2 p2 − u 2 4uT0 a ∂kT
b21 = T0 + 2m ∂u π a 2 p2 a 2 p2 − u 2 b22 = −
where Q =
d u du M ( r0 )
4T0 u2 ∂kT 1 3ba2
(k − aT0 + sin 2ξ )+ 2m ∂a 8pm π a 2 p2 a 2 p2 − u 2
represents the steepness of the energy source characteristic.
5 Calculations The calculation parameters are: ω0 = 1 s−1 , m = 1 kgf · s2 · cm−1 , k0 = 0.02 kgf · s · cm−1 , b = 0.07 kgf · cm−1 , T0 = 0.5 kgf, α1 = 0.84 s · cm−1 , α3 = 0.18 s3 · cm−3 , r0 = 1 cm, J = 1 kgf · s · cm2 . Depending on the linearization accuracy parameter i, the values of the coefficients N2 and N 3 are taken as follows: N2 = 0.6 (i = 1), N 3 = 0.75 (i = 1.5). The nonlinear elasticity is chosen in the form f (x) = γ3 x3 and, in accordance with (3), is replaced by kf = 0.75 γ3 a2 (N 3 = 0.75 at i = 1.5), where γ3 = ±0.2 kgf · cm−3 . Note that the number 0.75 also occurs if using, for example, the widespread averaging method is used [12].
106
A. A. Alifov
Fig. 1. Amplitude-frequency curves
Figure 1 shows the amplitude-frequency curves for the velocity values u = 1.14 cm· s−1 , u = 1.2 cm · s−1 . They completely coincide with the results of the well-known
Mixed Parametric and Auto-oscillations
107
Bogolyubov-Mitropolskii averaging method. Solid line 1 corresponds to the linear parametric excitation x cos vt in eq. (1), is shown for comparison, and aa denotes the selfoscillation amplitudes. The upper branch of curve 1, shown by the thick line, corresponds to stable oscillations, while the lower branch (thin line) corresponds to unstable ones. The dashed (u = 1.14) and dotted (u = 1.2) curves correspond to nonlinear parametric excitation. Figure 1,a takes place for a linear characteristic of elasticity (γ3 = 0), Fig. 1,b is soft (γ3 < 0), and Fig. 1,e is rigid (γ3 > 0). In the case of a nonlinear elastic force, there are no amplitude-frequency curves at u = 1.2, i.e. there are no real solutions to Eq. (8). The horizontal line in Fig. 1,b corresponds to ap ≈ u = 1.14. As can be seen from the graphs, in the case of γ3 = 0, the nonlinear parametric excitation reduces the amplitude, at u = 1.2 it significantly narrows the frequency range and changes the shape of the amplitude curve. Within the shaded sector, for the steepness of the characteristics of the energy source, fluctuations with amplitude are stable. Note that these sectors should actually be indicated on the load curve S(u). Sectors with black color reflect rather weak stability. In these sectors, the stability criteria (criterion) (10) are met with 0.000Y > 0 values, where Y ≤ 9.
6 Discussion and Conclusion The above calculation results show that nonlinear parametric excitation narrows the range of amplitude curves and can change their shape. The dynamics of the system under nonlinear parametric excitation differs from linear excitation, the differences are both quantitative and qualitative. At the same time, quantitative differences are more significant. However, the possibility is not ruled out that for other parameters of the system, these differences will change in the other direction.
References 1. Kononenko, V.O.: Vibrating Systems with Limited Power-Supply. Iliffe, London (1969) 2. Alifov, A.A., Frolov, K.V.: Interaction of Nonlinear Oscillatory Systems with Energy Sources. Hemisphere Publishing Corporation, New York, Washington, Philadelphia, London (1990) 3. Frolov, K.V.: Selected Works: in 2 vol. Nauka, Moscow (2007). (in Russian) 4. Krasnopolskaya, T.S., Shvets, A.Y.: Regular and Chaotic Dynamics of Systems with Limited Excitation. Regular and chaotic dynamics. M.-Izhevsk (2008) 5. Samantaray, A.K.: On the non-linear phenomena due to source loading in rotor-motor systems. Proc. Instit. Mech. Eng. J. Mech. Eng. Sci. 223(4), 809–818 (2008) 6. Cveticanin, L., Zukovic, M., Cveticanin, D.: Non-ideal source and energy harvesting. Acta Mech. 228(10), 3369–3379 (2017). https://doi.org/10.1007/s00707-017-1878-4 7. Pust, L.: Electro-mechanical impact system excited by a source of limited power. Eng. Mech. 15(6), 391–400 (2008) 8. Alifov, A.A.: About calculation of self-oscillatory system delayed and limited excitation. In: Proceedings of the International Conference on “Measurement and quality: problems, perspectives”, pp. 289–293. AzTU, Baku (2018) 9. Schmidt, G.: Parametric Oscillations. Mir, Moscow (1978). (in Russian)
108
A. A. Alifov
10. Alifov, A.A.: Oscillations at nonlinear parametric and limited excitation. Dyn. Syst., Mech. Mach. 10(1), 2–6 (2022) 11. Vibrations in the Technique, vol. 2. Mashinostroyeniye, Moscow (1979). (in Russian) 12. Bogolyubov, N.N., Mitropolskii, Y.: Asymptotic Methods in Theory of Nonlinear Oscillations. Nauka, Moscow (1974). (in Russian) 13. Moiseev, N.N.: Asymptotic Methods of Nonlinear Mechanics. Nauka, Moscow (1981). (in Russian) 14. Butenin, N.V., Neymark, Y., Fufaev, N.A.: Introduction to the theory of Nonlinear Oscillations. Nauka, Moscow (1976). (in Russian) 15. Hayashi, C.: Nonlinear Oscillations in Physical Systems. Princeton University Press, New Jersey (2014) 16. Wang, Q., Fu, F.: Variational iteration method for solving differential equations with piecewise constant arguments. Int. J. Eng. Manuf. 2(2), 36–43 (2012) 17. Karabutov, N.: Structural identification of nonlinear dynamic systems. Int. J. Intell. Syst. Appl. 09, 1–11 (2015) 18. Gourary, M.M., Rusakov, S.G.: Analysis of oscillator ensemble with dynamic couplings. In: AIMEE 2018. The Second International Conference of Artificial Intelligence, Medical Engineering, Education, pp. 150–160 (2018) 19. Ziabari, M.T., Sahab, A.R., Fakha-ri, S.N.S.: Synchronization new 3D chaotic system using brain emotional learning based intelligent controller. Int. J. Inform. Technol. Comput. Sci. 7(2), 80–87 (2015) 20. Alifov, A.A.: Methods of Direct Linearization for Calculation of Nonlinear Systems. Regular and chaotic dynamics. M.-Izhevsk 2015. (in Russian) 21. Alifov, A.A.: Method of the direct linearization of mixed nonlinearities. J. Mach. Manuf. Reliab. 46(2), 128–131 (2017) 22. Alifov, A.A.: About calculation of oscillatory systems with limited excitement by methods of direct linearization. Eng. Autom. Problems 4, 92–97 (2017) 23. Alifov, A.A.: About some methods of calculation nonlinear oscillations in machines. In: Proceedings of the International Symposium of Mechanism and Machine Science, pp. 378– 381. Izmir (2010) 24. Alifov, A.A.: About direct linearization methods for nonlinearity. In: Advances in Artificial Systems for Medicine and Education III. Advances in Intelligent Systems and Computing, 1126, pp. 105–114. Springer, Cham (2020) 25. Alifov, A.A.: On the calculation by the method of direct linearization of mixed oscillations in a system with limited power-supply. In: Advances in Computer Science for Engineering and Education II. Advances in Intelligent Systems and Computing, 938, pp. 23–31. Springer, Cham (2020) 26. Alifov, A.A.: On Mixed Forced and Self-oscillations with Delays in Elasticity and Friction. In: Zhengbing, H., Petoukhov, S., He, M. (eds.) CSDEIS 2020. AISC, vol. 1402, pp. 1–9. Springer, Cham (2021). https://doi.org/10.1007/978-3-030-80478-7_1 27. Murashkin, L.S., Murashkin, S.L.: Applied nonlinear mechanics of machine tools. Mashinostroenie, Leningrad (1977). (in Russian) 28. Ponomarev, A.S., et al.: Transverse self-oscillations of power tables caused by friction forces. Bull. Kharkov Polytech. Inst., Mashinostroenie 130(8), 67–69 (1977) 29. Bronovec, M.A., Zhuravljov, V.F.: On self-oscillations in systems for measuring friction forces. Izv. RAN, Mekh. Tverd. Tela 3, 3–11 (2012). (in Russian)
Spectrum Analysis on Electricity Consumption Periods by Industry in Fujian Province Huawei Hong, Lingling Zhu, Gang Tong, Peng Lv, Xiangpeng Zhan(B) , Xiaorui Qian, and Kai Xiao State Grid Fujian Electric Power Co., Ltd., Fuzhou 350003, China [email protected]
Abstract. Electricity is the basic energy for urban economic development and people’s daily life. Exploring periods of electricity consumption has a great significance in improving quality of life, promoting industry development and completing electric power planning. This paper applies the single spectral analysis method to explore the periods of electricity consumption in nine cities at Fujian Province, i.e. Fuzhou, Putian, Quanzhou, Xiamen, Zhangzhou, Longyan, Sanming, Nanping and Ningde, for six industries, i.e. big industry, on-general industry, residential living, non-residential lighting, commercial electricity and agricultural electricity. The data used is the daily electricity consumption data of Fujian Province from January 1, 2019 to April 17, 2022 for each pair of city and industry. It is found that the electricity consumption of most industries has an obvious annual period, except for residential living. In addition, the electricity consumption of some industries in some cities also have week period, quarter period, four-month period or semi-annual period, respectively. Keywords: Spectrum analysis · Electricity consumption · Periods · Different cities · Industry classified
1 Introduction Electricity is an indispensable energy source in today’s production and life, and electricity consumption is a direct manifestation of electricity consumption. As the “barometer” [1] and “vane” [2] of the national economy, electricity consumption is an important indicator of the city’s economic activities. Generally speaking, increasement of electricity consumption in industry means increasement of output value [3, 4]. Therefore, it is of great economic and social significance to study the electricity consumption periods in Fujian Province by industry, which leads to a more deeply understanding of electricity consumption period pattern. There have been some researches focusing on electricity consumption. [5] made a medium and long-term forecast of China’s electricity demand by using gray system theory and analyzed the fluctuation period by using autoregressive model; [6] predicted the annual household electricity consumption by calculation experiment; [7] used classical time series ARIMA method, ETS model and neural network autoregressive model to © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 109–119, 2023. https://doi.org/10.1007/978-3-031-36115-9_11
110
H. Hong et al.
forecast electricity consumption in Jiangsu Province, and finally found that the combined model was better than the non-combined model for electricity consumption; [8] based on feature selection, clustering and Markov process techniques to structure model and predict consumption. The simulation results show the proposed model over other similar algorithms such as LASSO-QRNN and HyFIS. Although electricity consumption has been studied by many scholars, there is no detailed and in-depth study on electricity consumption periods by industry so far. Time domain has always been the focus of various fields, but time domain is to study the time series as a whole. Spectral analysis can make up for the shortcomings of time domain, and many foreign scholars have introduced Spectral analysis into the research field and achieved important research results [9–11]. Spectral analysis is one of the main methods used in Statistics to find hidden periods in time series data and complete period analysis in various industries [12–15]. Besides, [16] used singular spectrum analysis (SSA) to detect the hidden periodicity information in the data and compared the forecasting performance with method combining linear recurrent formulas and artificial neural networks (ANNs), getting the conclusion that SSA-ANN model is more accurate; [17] extracted periodic component by SSA, and concluded that the performance of SSA was best, compared with the classical methods; [18] extracted a financial period by using multi-channel SSA to obtain more robust and conclusive results of the global financial period based on time-series data from 1970 to 2018; [19] employed a wavelet spectrum analysis to study globalization and business periods in China and G7 countries, found that the overall co-movements and trade tended not to be significantly related. In summary, most of the studies on electricity consumption stayed at the level of constructing models to forecast electricity consumption [20, 21], while the studies on electricity consumption periods are insufficient, and the studies on electricity consumption periods by industries have not been reported. Spectral analysis, as one of the most excellent methods for period theory research, has been widely used in period research. In this paper, we will explore the period pattern of electricity consumption by industry in different cities of Fujian Province through single-spectrum analysis, and provide new ideas for the in-depth study of electricity consumption.
2 Basic Idea of Spectral Analysis According to the Fourier transform theory, a smooth time series can be regarded as a superposition of several regular sine or cosine waves with appropriate amplitude, frequency and phase. Spectral analysis is to convert the sequence from the time domain to the frequency domain with the help of Fourier transform, so as to decompose the time series into the superposition of different harmonics, and analyze the periodic characteristics of the time series by comparing the characteristics of different frequency harmonics. The basic principle of single-spectrum analysis is to find out the main frequency components according to the spectral density function. There will be obvious spikes in the frequency-spectral density diagram, and the periods corresponding to the spikes are the main periods of the time series. First, the expression for the spectral density function [22, 23] is given.
Spectrum Analysis on Electricity Consumption Periods
111
For the time series {Yt }, there is a self-covariance function {γ (h)}, a monotonically nondecreasing right continuous function F(ω) at the frequency ω ∈ (−1/2, 1/2) satisfying F(−1/2) = 0, such that −1/2 e2π iωh dF(ω), h = 0, ±1, ±2, · · · . (1) γ (h) = 1/2
F(ω) is said to be the spectral distribution function of {Yt }, and i i is imaginary unit root. If there is a non-negative function f (ω) with ω ∈ (−1/2, 1/2) such that −1/2 e2π iωh f (ω)d ω, h = 0, ±1, ±2, · · · , (2) γ (h) = 1/2
then f (ω) is said to be the spectral density function of {Yt }, referred to as the spectral density. Reversing Eq. (2) gives the spectral density as ∞
f (ω) =
γ (h)e−2π iωh .
(3)
h=−∞
Define d (ωj ) as the Discrete Fourier Transformation (DFT) of Yt with the following expression: d (ωj ) = n−1/2
n
Yt e−2π iωj t , j = 0, 1, · · · n − 1,
(4)
t=1
where n is the number of samples. Due to
n
e−2π iωj t = 0, j = 0, the DFT can be
t=1
rewritten as d (ωj ) = n−1/2
n
(Yt − Y )e−2π iωj t .
(5)
t=1
It can be shown that f (ωj ) has the following relationship with d (ωj ) n n 2 ωj = 4n−2 Yt − Y Ys − Y e−2π iωj (t−s)
n 4 d
= 4n−2 = 4n−2
t=1 s=1 n−1
Yt+|h| − Y Yt − Y e−2π iωj
h=−(n−1) t=1 n−1 h=−(n−1)
where j = 0, γˆ (h) = n−1
n−|h|
n−|h|
−2π iωj h γˆ (h)ej
(6)
= 4n−1 fˆ (ωj ),
(Yt+|h| − Y )(Yt − Y ), h = t − s is an estimator of γ (h)
t=1
and fˆ (ωj ) is an estimator of f (ω).
112
H. Hong et al.
Second, prove the relationship between periodogram and the spectral density. Time series Yt , t = 1, · · · n can be expressed as a linear combination of sine and cosine components of different frequencies according to Fourier transform. For m = (n − 1)/2, Yt at ωj = j/n, the expression is as follows: m
Yt = A0 +
[Aj cos(2π ωj t) + Bj sin(2π ωj t)]
(7)
j=1
where A0 = Y
(8)
2 Yt cos(2π tωj ) n n
Aj =
(9)
t=1
Bj =
n 2
n
Yt sin(2π tωj )
(10)
t=1
Periodic graph I is defined as I (ωj ) = A2j + Bj2
(11)
Next, prove that I (ωj ) = A2j + Bj2 =
4 |d (ωj )|2 n
(12)
By Euler’s formula, we have d (ω) = n−1/2 = n−1/2
n
Yt e−2π iωj t
t=1 n
t=1
Yt cos(2π ωj t) − i
n
Yt sin(2π ωj t) .
(13)
t=1
Thereby, |d (ωj )|2 = n−1 (
n
Yt cos(2π iωj t))2 + n−1 (
n
Yt sin(2π iωj t))2
(14)
I (ωj ) = A2j + Bj2 n n 2 2 = 4n−2 Yt cos 2π ωj t + 4n−2 Yt sin 2π ωj t t=1 t=1 2 = 4n−1 d ωj
(15)
t=1
t=1
Combining Eqs. (9), (10), (11) and (14) yields that
and Eq. (12) is proved.
Spectrum Analysis on Electricity Consumption Periods
113
That is, we can analyze the sample spectral density f (ωj ) of Yt by study the periodogram I (ωj ). Hence, we often use the periodogram method instead of estimating the spectral density in practice. According to Eq. (11), we can quickly calculate the periodogram and obtain the spectral density estimator.
3 Data Preprocessing The daily electricity consumption data of six types of industries in nine cities at Fujian Province from January 1, 2019 to April 17, 2022 were collected as raw data. In total, the data used are 54 time series data for six types of industries in nine cities. Daily trend will introduce very low frequency components in the periodogram, which often masks the higher frequency. Preprocessing is completed by fitting the raw data to a linear regression over time, i.e. yt = α + βt, and the residuals, i.e. t = yt − yˆ t , are used as data for the subsequent spectral analysis. The F-test is used to test whether the linear regression is significant. Under null SSR/1 follows F(1, n − 2) distribution, hypothesis H0 : β = 0, the test statistic F = SSE/(n−2) where: SSR =
n t=1
(ˆyt − y)2 , SSE =
n
(yt − yˆ t )2 .
t=1
Then P-value is calculated as P(F(1, n − 2) > F). Finally, all the P-values of 54 linear regressions are smaller than 0.05, which means the null hypothesis should be rejected and the regressions are significant. Hence, the linear trend of raw data should be removed by using the residuals instead.
4 Study on Periods by Single-Spectrum Analysis In the following, single-spectrum analysis will be applied on the time series data. The periods corresponding to the spikes on the frequency-spectral density plot are the main period of the time series. To facilitate the conclusion, the horizontal coordinates of the frequency-spectral density plot are transformed to T = 1/ω, so that the single-spectrum analysis plot with the horizontal coordinate as period and the vertical coordinate as spectrum is obtained. 4.1 Big Industry The spectrum-period curves of big industry are shown in Fig. 1.
114
H. Hong et al.
Fig. 1. Spectrum analysis of big industry daily electricity consumption in 9 cities.
From Fig. 1, the periods of electricity consumption of big industry in nine cities are concluded as follows: 1) All nine cities have annual period, but Sanming has a weaker one; 2) All nine cities have semi-annual period, four-month period, and quarter period, but the periods are more obvious in northwestern cities of Longyan, Sanming, Nanping and Ningde than that in southeastern cities; 3) Only Xiamen has a weak week period. 4.2 Non-general Industry The spectrum-period curves of non-general industry are shown in Fig. 2.
Fig. 2. Spectrum analysis of non-general industry daily electricity consumption in 9 cities
Spectrum Analysis on Electricity Consumption Periods
115
From Fig. 2, the periods of electricity consumption of non-general industry in nine cities are concluded as follows: 1) There is an obvious annual period in all nine cities; 2) Longyan, Sanming and Nanping have an obvious semi-annual period, while Putian, Quanzhou, Zhangzhou and Ningde have a weak semi-annual period; 3) Except Xiamen, other cities have a weak quarter period. 4.3 Residential Living The spectrum-period curves of residential living are shown in Fig. 3.
Fig. 3. Spectrum analysis of residential living daily electricity consumption in 9 cities
From Fig. 3, the periods of electricity consumption of residential living in nine cities are concluded as follows: 1) Southern cities, including Fuzhou, Putian, Quanzhou, Xiamen and Zhangzhou, have an obvious annual period, and Ningde has a weak annual period; 2) Besides Fuzhou and Putian, northwestern cities, including Longyan, Sanming, Nanping and Ningde, have an obvious semi-annual period, and Quanzhou and Zhangzhou have a weak one, while Xiamen’s semi-annual period is the weakest. According to the above results, we can conclude that the main period of residential living electricity consumption in southeastern Fujian Province is annual period, while the main period in northwestern Fujian Province is semi-annual period. This phenomenon may be derived from climate [24]. Since winter temperature in the northwest is lower than that in the southeast, northwestern cities have a higher electricity consumption in winter. Hence, the electricity consumptions in summer and winter of northwestern cities are approximate, which leads to a semi-annual period. 4.4 Non-residential Lighting The spectrum-period curves of non-residential lighting are shown in Fig. 4.
116
H. Hong et al.
Fig. 4. Spectrum analysis of non-residential lighting daily electricity consumption in 9 cities
From Fig. 4, the periods of electricity consumption of non-residential lighting in nine cities are concluded as follows: 1) There is an obvious annual period in all nine cities; 2) Except Xiamen and Zhangzhou, other seven cities have semi-annual period. 4.5 Commercial Electricity The spectrum-period curves of commercial electricity are shown in Fig. 5.
Fig. 5. Spectrum analysis of commercial electricity daily electricity consumption in 9 cities
From Fig. 5, the periods of electricity consumption of commercial electricity in nine cities are concluded as follows: 1) There is an obvious annual period in all nine cities; 2) Four northwestern cities, including Longyan, Sanming, Nanping and Ningde, have a weak semi-annual period.
Spectrum Analysis on Electricity Consumption Periods
117
4.6 Agricultural Electricity The spectrum-period curves of agricultural electricity are shown in Fig. 6.
Fig. 6. Spectrum analysis of agricultural daily electricity consumption in 9 cities
From Fig. 6, the periods of agricultural electricity consumption in nine cities is concluded as follows: 1) There is an obvious annual period in all nine cities; 2) Quanzhou has an obvious semi-annual period, and Putian, Zhangzhou, Longyan, Sanming and Ningde have a weaker one; 3) Longyan, Sanming and Ningde have a weak four-month period; 4) Only Quanzhou has an obvious quarter period. Quanzhou has high quality vegetables industry, and Longyan, Sanming and Ningde have well developed animal husbandry. This may be the reason that these cities have special agricultural electricity consumption period.
5 Conclusion In this paper, the periods of electricity consumption by industry in Fujian Province are studied by using single-spectrum analysis, based on daily electricity consumption data of six types of industries in nine cities in Fujian Province. In general, first, except residential living, other five types of industries in all nine cities have an obvious annual period; second, the industries with four-month period and quarter period concentrates in big industry and non-general industry; third, northwestern cities, including Longyan, Sanming, Nanping and Ningde, have similar daily electricity consumption periods, and their quarter, four-month and semi-annual periods are more obvious than southeastern cities; finally, Xiamen has the most simple period pattern, where only annual period is obvious. Although this paper has studied the period of electricity consumption by industry in Fujian Province through single-spectrum analysis, further analysis is required to find the
118
H. Hong et al.
reasons of above results. Next, we plan to first extend our research to study influencing factors of electricity consumption periods in different cities and industries, such as geographic location, climate, electricity consumption habits of users, industry development policies, GDP and so on. And then investigate the reasons for formation of different electricity consumption periods, so as to better understand the law of electricity consumption periods by industry in Fujian Province. Acknowledgement. This paper is supported by Research and Application of Electric Power Forecasting Model Library Facing the Development of Electricity Market (No. 52130X230008).
References 1. Guoqiang, J., Jinhui, D., Jiangtao, L., et al.: National economic development barometerinterpretation of the whole society’s electricity consumption in 2020. National Grid 2, 50–53 (2021). (in Chinese) 2. Yang, L.: Stable economic recovery: July electricity consumption up 12.8% year-on-year. China Securities Journal, 2021-08-12 (A02) (in Chinese) 3. Zufei, X., Meng, W., Wen, H., et al.: An empirical study on the relationship between industrial structure, electricity consumption and economic growth in Shanghai. China Business Journal 03, 42–47 (2021). (in Chinese) 4. Limin, J., Lei, X.: Research on the relationship between electricity consumption and economic growth in China. Management Observation 30, 43–46 (2016). (in Chinese) 5. Wenxia, Z.: Economy forecasting and periodicity analysis for power demand. Information on Electric Power 04, 14–16 (2001). (in Chinese) 6. Sidorov, A.I., Tavarov, S.S.: Method for forecasting electric consumption for household users in the conditions of the Republic of Tajikistan. Int. J. Sustain. Dev. Plan. 15(04), 569–574 (2020) 7. Qi, W.: Analysis of residential electricity consumption forecasting in Jiangsu Province based on combined model. Jiangsu Business Theory 01, 11–14 (2022). (in Chinese) 8. Dalkani, H., Mojarad, M., Arfaeinia, H.: Modelling electricity consumption forecasting using the Markov process and hybrid features selection. Int. J. Intell. Sys. Appli. (IJISA) 13(5), 14–23 (2021) 9. Yasir, M., Shah, Z.S., Memon, S.A., et al.: Machine learning based analysis of cellular spectrum. Int. J. Wireless and Microwave Technolog. (IJWMT) 11(2), 24–31 (2021) 10. Zerihun, B.M., Olwal, T.O., Hassen, M.R.: Spectrum sharing technologies for cognitive iot networks: Challenges and future directions. Int. J. Wireless and Microwave Technolog. (IJWMT) 10(01), 17–25 (2020) 11. Joshi, D., Sharma, N., Singh, J.: Spectrum sensing for cognitive radio using hybrid matched filter single cycle cyclostationary feature detector. Int. J. Info. Eng. Elect. Bus. (IJIEEB) 5, 13–19 (2015) 12. Elsner, J.B., Tsonis, A.A.: Singular. Spectral analysis: a new tool in time series analysis. Plenum Press, New York and London (1996) 13. Hassani, H., Zhigljavsky, A.: Singular spectrum analysis: methodology and application to economics data. J. Syst. Sci. Complexity 22, 372–394 (2009) 14. Dabbakuti, J.R.K.K., Gundapaneni, B.L.: Application of singular spectrum analysis using artificial neural networks in TEC predictions for ionospheric space weather. IEEE J. Selec. Top. Appl. Earth Observ. Remo. Sens. 12(12), 5101–5107 (2019)
Spectrum Analysis on Electricity Consumption Periods
119
15. Andi, Z., Lele, H., Huewen, W.: The implicit period of non-stationary time series based on spectral analysis. Math. Practice Theory 46(18), 197–203 (2016). (in Chinese) 16. Sun, M., Li, X., Kim, G.: Precipitation analysis and forecasting using singular spectrum analysis with artificial neural networks. Clust. Comput. 22(S12), 633–640 (2019) 17. Coussin, M.: Singular spectrum analysis for real-time financial cycles measurement. J. Int. Money Financ. 120, 102532 (2022) 18. Škare, M., Porada-Rocho´n, M.: Multi-channel singular-spectrum analysis of financial cycles in ten developed economies for 1970–2018. J. Bus. Res. 112, 567–575 (2020) 19. Pomˇenková, J., Fidrmuc, J., Korhonen, I.: China and the world economy: wavelet spectrum analysis of business cycles. Appl. Econ. Lett. 21(18), 1309–1313 (2014) 20. Lee, Y.W., Tay, K.G., Choy, Y.Y.: Forecasting electricity consumption using time series model. Int. J. Eng. Technol. 07(04), 218–223 (2018) 21. Guofeng, F., Xiao, W., Yating, L., Weichiang, H.: Forecasting electricity consumption using a novel hybrid model. Sustain. Cities Soc. 61, 102320 (2020) 22. Shumway, H., Robert, S.S.D.: Time Series Analysis and Applications. Springer International, USA, pp. 165–203 (2016) 23. Jonathan, C.D., Chan, K.-S.: Time series analysis and applications: R language, 2nd edition. In: Pan, H., et al. (eds.) Translation, pp. 229–238. Machinery Industry Press, Beijing (2011) 24. Chen, Y.: Research on the influence of temperature change on urban residential electricity consumption. Nanjing University of Information Engineering (Nanjing), 42–49 (2022). (in Chinese)
Determination of the Form of Vibrations in Vibratory Plow’s Moldboard with Piezoceramic Actuator for Maximum Vibration Effect Sergey Filimonov, Sergei Yashchenko, Constantine Bazilo(B) , and Nadiia Filimonova Cherkasy State Technological University, Shevchenko blvd., 460, Cherkasy 18006, Ukraine {s.filimonov,n.filimonova}@chdtu.edu.ua, [email protected]
Abstract. Agriculture is the practice of growing plants and livestock. The task of agriculture is to supply raw materials for industry and provide the population with food. One of the main problems of agriculture is the complicacy and efficiency of soil cultivation. The plow is the main instrument for soil’s tillage in agriculture. The analysis of present-day plows is conducted and their characteristics are analyzed. The method of reducing the friction of the cultivation unit with the help of ultrasonic vibrations has been experimentally tested. For the first time, the oscillation forms of the vibratory plow’s moldboard with a piezoceramic actuator at ultrasonic frequencies are obtained. According to the results of the research, it is found that the frequencies of 36 kHz and 37.4 kHz can be chosen as the resonant ones. The influence of negative factors, such as soil sticking during its cultivation, is reduced, as well the resistance created by the soil during its cultivation is reduced. Thus, the time of plow operation is increased. When using a piezoceramic actuator, more flexible possibilities open up in the control of the vibratory plow. Keywords: Vibratory plow · Vibration technologies · Chladni figures · Friction reduction · Prototype · Ultrasonic transducer
1 Introduction One of the most important sectors of the economy is agriculture, because it provides the population with food, other sectors with raw materials. The economic condition and food security of the country depends on the state of agriculture. The International Food Policy Research Institute says that agricultural technologies will affect food production if they are applied jointly. Technological advances are helping to provide farmers with the tools and resources to make agriculture more sustainable [1, 2].
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 120–128, 2023. https://doi.org/10.1007/978-3-031-36115-9_12
Determination of the Form of Vibrations
121
The main tool in agriculture for tillage is the plow. Most of the agricultural enterprises are sure that it is impossible to completely abandon the plow, since it is necessary for loosening the soil, controlling weeds and improving the quality of the soil in general, and is also one of the well-proven methods of tillage [3, 4]. Plowing is a method of tillage, which consists in cutting the treated layer, raising it with loosening and turning by 130–180° and laying a previously open furrow on the bottom [5, 6]. In the technological process of growing crops, tillage is the most energy intensive operation. The main problem of agriculture is the labour intensity and efficiency of tillage. One of the most common tillage methods is loosening the soil, so improving its properties is one of the most effective solutions of the problem of excessive energy consumption [7]. When plowing, the soil layer is deformed, destroyed and displaced [8]. Loosening the ground is created through the use of conventional plows. One of the main disadvantages of this plow is the high resistance (friction) that occurs when plowing the soil, which results in high energy consumption, faster wear of the material from which the plow is made and its failure [9]. The effect of friction during plowing is shown schematically in Fig. 1.
Fig. 1. Friction and its negative factors in tillage
As can be seen from Fig. 1, the effect of friction is of great importance in tillage. Therefore, an important task is to reduce its influence.
2 Literature Review The plow consists of several components, two of which (a plowshare and a moldboard) are the main ones [10, 11]. Plow moldboard’s main task is to cut from the wall of the soil layer, crumble it and turn it over. That is, in many respects, how well the soil will be prepared depends on the quality of the moldboard [12, 13]. The ease of manufacture and relatively low price are plow’s main advantages. The main limitation is the high resistance (friction) that occurs when plowing the soil, while energy costs increase significantly.
122
S. Filimonov et al.
Figure 2 shows a comparative diagram of plows by their type and purpose.
Fig. 2. A comparative diagram of plows according to their type and purpose (the plow under development and research in this article is highlighted with a dotted line)
As can be seen from the diagram, modern plows can be divided into two types: conventional and vibratory ones. In this work we will focus on vibratory plows. The main elements for creating vibration [14–16] in vibratory plow are mechanical elements, hydraulic systems and piezoceramic actuators. It has been experimentally established that when using vibrations in a plow, the sliding friction of the soil, which is the main component in the total traction resistance, is significantly reduced. The sticking of the working bodies is also reduced [17]. It is presented and experimentally proved in [18] the decrease in friction between two surfaces of bodies when exposed to ultrasonic vibrations. Thereby, many constructions of plows with vibrating operating units have recently appeared [19]. Figure 3 shows a diagram of a vibratory plow.
Fig. 3. Functional diagram of a hydraulic vibratory plow: a) general view; b) view A; 1 – plowshare, 2 – shelves, 3 – rack, 4 – shoe, 5 – field board, 6 – rubber-metal absorbers, 7 – hydraulic vibrator, 8 – lever, 9 – piston, 10 – spool valve.
Determination of the Form of Vibrations
123
The traction resistance of the plow is reduced due to the pseudo liquefaction of the soil, which occurs under the action of a vibrating share 1. Rubber-metal absorbers 6 must allow vibrations of the share 1 of the vibrating plow body with an amplitude of 3– 4 mm. Forced oscillations transmitted to the plowshare 1 by the vibrator 7 make changes to the natural vibratory process of soil destruction. Plowshare 1 oscillation frequency is 300 oscillations per minute. They contribute to this destruction, that is, the traction resistance during forced vibrations of the tool decreases in comparison with the traction resistance without vibrations and reduces the overall energy intensity of the process [20]. However, such improvements have a number of limitations: an increase in the weight of the working body and the complexity of maintenance and manufacturing of the product, as well as decrease in the operating time of the vibration unit, etc. In paper [21] the construction of a vibratory plow is discussed. To create vibrations, a mechanical system with an offset eccentric is used. The limitation of such plow is the complexity of the design and the limited vibrations’ power. Another example is a vibratory plow model using a piezoceramic actuator [22], which is given in the work [23] (Fig. 4).
Fig. 4. 3D Model of vibratory plow with piezoelectric actuator: a) front view; b) back view; 1 – moldboard, 2 – plowshare, 3 – body stand, 4 – piezoelectric actuator in the form of a disk, 5 – shoe
In the works cited above, it is not experimentally given how the oscillations will propagate in the plow moldboard when they are created. Determining the shape of the oscillations in a working vibratory plow will allow to select the operating mode more effectively and correctly distribute the load. The aim of the work is reducing the plow’s friction during soil plowing as well as reducing the sticking of the operating units by determining the form of vibration of the vibratory plow based on a piezoceramic actuator.
3 Materials and Methods In order to choose the correct operating mode of the vibratory plow, it is necessary to have an idea about the form of oscillations occurring in it, which directly depends on the frequency. To visually determine the shape of vibrations of a plow with piezoelectric actuator, the method of Chladni figures for the main resonant frequencies was used.
124
S. Filimonov et al.
The principle of the method is as follows. First, the current-frequency characteristic of a vibratory plow with a piezoceramic actuator is measured to determine the main resonant frequencies. To study and determine the Chladni figures, the installation shown in Fig. 5 is used.
Fig. 5. Scheme of the installation for determining the form of oscillations of a vibratory plow with piezoelectric actuator: 1 – alternating voltage generator, 2 – high-voltage amplifier, 3 – ammeter, 4 – voltmeter, 5 – plow with piezoceramic actuator, 6 – video camera
Activated carbon in powder form “Carbolong” was used as an indicator. In the case of standing waves at resonant frequencies, the powder is concentrated at the vibration nodes and thus detects them. The experiments were conducted on an intelligent complex for the development and research of piezoelectric components [24, 25], created within the scientific work (the number of state registration 0117U000936). The installation works as follows. A rectangular voltage of 300 V is supplied from the generator to the piezoceramic actuator. Due to the action of the reverse piezoelectric effect, oscillations occur in the piezoceramic actuator. Since the piezoceramic actuator is rigidly connected to the plow blade, and “Carbolong” is poured on its surface, during oscillations, part of the powder falls off, and part remains in the oscillation nodes, showing what oscillations occur.
4 Experiments and Results A prototype of a vibratory plow with piezoelectric actuator is shown in Fig. 6.
Determination of the Form of Vibrations
125
Fig. 6. Prototype of vibratory plow with a piezoceramic actuator: a) front view; b) back view
The plow “Lux 200” of Ukrainian production was used for research. The piezoceramic actuator was connected to the plow moldboard with epoxy adhesive. The parameters of the ultrasonic piezoelectric actuator are presented in Table 1. Table 1. Main parameters of the ultrasonic piezoelectric actuator Parameter of ultrasonic actuator
Value
Waveguide’s diameter maximum
45 mm
Waveguide’s diameter minimum
38 mm
Waveguide’s length
25 mm
Reflector’s thickness
14 mm
Piezoelectric elements’ diameter
38 mm
Total length
49 mm
Power
60 W
Resonant frequency
40 kHz
Figure 7 shows the initial state of the experimental study (part of the plow moldboard, piezoelectric actuator and “Carbolong” powder).
Fig. 7. The initial state of the experimental study of a vibratory plow with a piezoceramic actuator
126
S. Filimonov et al.
Figure 8 shows some experimental results of the vibration form of the moldboard of a vibratory plow with piezoelectric actuator, obtained using the Chladni figures.
Fig. 8. The experimental results of Chladni figures of a vibratory plow with a piezoceramic actuator: a) 36 kHz; b) 37.1 kHz; c) 39.1 kHz; d) 40 kHz
From Fig. 8 it can be seen that the maximum vibration area of the moldboard of a plow with piezoceramic actuator is observed at a frequency of 36 kHz. Separately, it should be noted the obtained Chladni figures at a frequency of 37.4 kHz (Fig. 9).
Fig. 9. The experimental results of Chladni figures of a vibratory plow with piezoelectric actuator at a frequency of 37.4 kHz
Determination of the Form of Vibrations
127
From Fig. 7 it can be seen that the generated vibrations occupy a large part of the plow moldboard. Thus, a decrease in friction in the moldboard occurs due to a decrease in friction contact, which is indicated by the Chladni figures. Further research by the authors can be aimed at studying the possibility of alternately using a piezoceramic transducer (vibrator) as an actuator and a sensor for adjusting the oscillation amplitude, as well as modeling the developed design of a vibratory plow in the COMSOL Multiphysics software package.
5 Conclusions The advantages and disadvantages of modern models of vibratory plows were revealed. The method of reducing the friction of the cultivation unit with the help of ultrasonic vibrations has been experimentally tested. For the first time, the oscillation forms of the moldboard of a vibratory plow with piezoelectric actuator at ultrasonic frequencies were obtained. According to the results of the research, it was found that the frequencies of 36 kHz and 37.4 kHz can be chosen as the resonant ones. When using a piezoceramic actuator, more flexible possibilities for controlling the vibratory plow open up, and the weight and dimensions of the plow practically do not increase. The use of a vibratory plow based on a piezoelectric actuator will reduce the effect of soil sticking during its cultivation, reduce the resistance created by the soil, and increase the operating time of the plow.
References 1. Kalogiannidis, S., Kalfas, D., Chatzitheodoridis, F., Papaevangelou, O.: Role of CropProtection Technologies in Sustainable Agricultural Productivity and Management. Land 11, 1680 (2022). https://doi.org/10.3390/land11101680 2. Rosegrant, M., Ringler, C., Cenacchi, N., Cindy C.: Food security in a world of natural resource scarcity: The role of agricultural technologies. International Food Policy Research Institute (IFPRI), Washington, D.C. (2014). https://doi.org/10.2499/9780896298477 3. Safety and health in agriculture. International Labour Organization, Report VI (1) (2000). https://www.ilo.org/public/english/standards/relm/ilc/ilc88/rep-vi-1.htm 4. United Nations Environment Programme. Making Peace with Nature: A scientific blueprint to tackle the climate, biodiversity and pollution emergencies. Nairobi (2021). https://www. unep.org/resources/making-peace-nature 5. Weil, R.C., Brady, N.C.: Elements of Nature and Properties of Soil. Pearson (2009) 6. Acquaah, J.: Principles of Crop Production: Theory, Techniques, and Technology. PrenticeHall (2011) 7. Hukov, Ya.S.: Soil cultivation. Technology and technique. Mechanical and technological substantiation of energy-saving means for the mechanization of soil cultivation in the conditions of Ukraine. Kyiv, Nora-print (1999) 8. Lurie, A.B., Grombchevskii, A.A.: Calculation and design of agricultural machinery Leningrad, Mashinostroenie (1977)
128
S. Filimonov et al.
9. Filimonov, S.O., Yashchenko, S.S., Batrachenko, A.V., Filimonova, N.V.: Using smart piezoceramics for soil cultivation in agriculture. Bulletin of Cherkasy State Technological University 2, 30–36 (2019). https://doi.org/10.24025/2306-4412.2.2019.169817 10. Chen, P., Tao, W., Zhu, L., Wu, Q.-M.: Effect of varying remote cylinder speeds on ploughbreast performances in alternative shifting tillage. Computers and Electronics in Agriculture 181, 105963 (2021). https://doi.org/10.1016/j.compag.2020.105963 11. Lim, Y., Zhai, Z., Cong, P., Zhang, Y.: Effect of plough pan thickness on crop growth parameters, nitrogen uptake and greenhouse gas (CO2 and N2 O) emissions in a wheat-maize doublecrop rotation in the Northern China Plain: A one-year study. Agricultural Water Management 213, 100081 (2019). https://doi.org/10.1016/j.agwat.2018.10.044 12. Hrabˇe, P., Müller, M., Hadaˇc, V.: Evaluation of techniques for ploughshare lifetime increase 61, 72–79 (2015). https://doi.org/10.17221/73/2013-RAE 13. Ansorge, D., Godwin, R.J.: The effect of tyres and a rubber track at high axle loads on soil compaction: part 3: comparison of virgin compression line approaches. Biosystems Engineering 104(2), 278–287 (2009). https://doi.org/10.1016/j.biosystemseng.2009.06.024 14. Li, J.: Design of active vibration control system for piezoelectric intelligent structures. Int. J. Edu. Manage. Eng. (IJEME) 2(7), 22–28 (2012). https://doi.org/10.5815/ijeme.2012.07.04 15. Tomaneng, S., Docdoc, J.A.P., Hierl, S.A., Cerna, P.D.: Towards the development a costeffective earthquake monitoring system and vibration detector with SMS notification using IOT. Int. J. Eng. Manuf. (IJEM) 12(6), 22–31 (2022). https://doi.org/10.5815/ijem.2022.06.03 16. Hu, Z., Legeza, V., Dychka, I., Legeza, D.: Mathematical model of the damping process in a one system with a ball vibration absorber. Int. J. Intelli. Sys. Appli. (IJISA) 10(1), 24–33 (2018). https://doi.org/10.5815/ijisa.2018.01.04 17. Bulgakov, V.M., Sviren, M.O., Palamarchuk, I.P., Dryga, V.V., Chernysh, O.M., Yaremenko, V.V.: Agricultural Vibrating Machines. Kirovograd, KOD (2012) 18. Dong, S., Dapino, M.: Experiments on ultrasonic lubrication using a piezoelectrically-assisted tribometer and optical profilometer. J. Vis. Exp. 103, 52931 (2015). https://doi.org/10.3791/ 52931 19. Bulgakov, V.M., Sviren, M.O., Kisiliov, R.V., Oryshchenko, S.B., Lisovyi, I.O.: Research of vibration processes at the main processing of the soil. Scientific Bulletin of the Tavriya State Agrotechnological University. Melitopol 1(5), 3–13 (2015) 20. Loveikin, V.S., Yaroshenko, V.F., Bychenko, L.A.: Vibratory plough body. Patent 19528 UA. Publ. 15.12.06, Bul. 12 21. Doering, D., Campbell, T.: Vibratory plow. Patent US 7,546,883 B1 (2009) 22. Sharapov, V.M., et al.: Improvement of piezoceramic scanners. In: IEEE XXXIII International Scientific Conference Electronics and Nanotechnology (ELNANO), Kiev, pp. 144–146 (2013). https://doi.org/10.1109/ELNANO.2013.6552063 23. Bazilo, C., Filimonov, S., Filimonova, N., Yashchenko, S.: Method of reducing friction in the plow moldboard with soil during cultivation due to the implementation of ultrasonic vibrations. In: Hu, Z., Petoukhov, S., Yanovsky, F., He, M. (eds.) Advances in Computer Science for Engineering and Manufacturing. ISEM 2021. Lecture Notes in Networks and Systems, vol 463, pp. 281–289 (2022). https://doi.org/10.1007/978-3-031-03877-8_25 24. Filimonov, S.A., Bazilo, C.V., Bondarenko, Y., Filimonova, N.V., Batrachenko, A.V.: Creation of a highly effective intellectual complex for the development and research of piezoelectric components in instrument engineering, medicine and robotics. Bulletin of Cherkasy State Technological University 3, 33–43 (2017) 25. Bazilo, C., Filimonov, S., Filimonova, N., Bacherikov, D.: Determination of Geometric Parameters of Piezoceramic Plates of Bimorph Screw Linear Piezo Motor for Liquid Fertilizer Dispenser. In: Hu, Z., Petoukhov, S., Yanovsky, F., He, M. (eds.) ISEM 2021. LNNS, vol. 463, pp. 84–94. Springer, Cham (2022). https://doi.org/10.1007/978-3-031-03877-8_8
Evaluating Usability of E-Learning Applications in Bangladesh: A Semiotic and Heuristic Evaluation Approach Samrat Kumar Dey1 , Khandaker Mohammad Mohi Uddin2(B) , Dola Saha2 , Lubana Akter2 , and Mshura Akhter2 1 School of Science and Technology, Bangladesh Open University, Gazipur 1705, Bangladesh 2 Department of Computer Science and Engineering, Dhaka International University,
Dhaka 1205, Bangladesh [email protected]
Abstract. In this age of technology, the popularity, and necessity of mobile applications in the field of education are increasing day by day. Many students in Bangladesh are not able to develop as qualified due to a lack of a suitable environment and skilled teachers. This problem has been solved to a large extent through various E-Learning applications. But no research has yet been done on exactly how efficient, intuitive, and usable these applications are. Therefore, this study’s objectives are to assess these E-Learning apps’ usability in the context of Bangladesh and to offer potential architectural recommendations for improving their usability and acceptance. To accomplish the objectives, the usability of five E-Learning applications (Bacbon, Prottoy, Shikho, Classroom, and Muktopaath) developed in the context of Bangladesh was evaluated through Heuristic Evaluation and Semiotic Evaluation process. The study showed that all selected applications have several usability problems. The heuristic evaluation is based on the principle of efficiency, consistency, flexibility, conventions, and error management of each application while the semiotic evaluation upheld the intuitiveness of the interface elements of the selected applications. Finally, to enhance the usability of these applications, some sets of recommendations and implications have been presented for practitioners. Keywords: Usability · Heuristic Evaluation · Semiotic Evaluation · E-Learning · Usability Problems · HCI · Mobile Applications
1 Introduction In the field of education, mobile technology has taken a big place for itself. Education is gaining popularity through mobile apps in Bangladesh like other developed countries for the advancement of Information and Communication Technology (ICT). Although the rate of education has increased in Bangladesh, in many cases due to lack of a proper efficacy education system, many children drop out prematurely. The Teacher-Student Ratio acts as a big influencer in this case. According to the Bangladesh Bureau of © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 129–140, 2023. https://doi.org/10.1007/978-3-031-36115-9_13
130
S. K. Dey et al.
Educational Information and Statistics (BANBEIS), the Teacher-Student Ratio (TSR) in primary and secondary schools is 1:37 and 1:45 respectively [1, 2]. Some private education organizations maintain a proper teacher-student ratio. But it is quite hard to maintain the ratio in government schools. In this case, mobile technology can be a ground breaking tool. Through a website or app full of numerous educational resources, everyone in every region of the country may get the proper education. It will be easy for everyone to get education through this. It is possible to teach that subject to all the students of the country through the teaching material of one teacher. According to a UNESCO report [3], more than 5,050 million children and young people, due to the epidemic, could not attend classes in almost half of the world’s student population. In such difficult times, mobile learning apps have smoothed the path of learning for students. The use of mobile learning apps for epidemics in the education sector has increased a lot. But the question is, how useful are these learning apps? Can students really gain accurate knowledge through this app? How much will learn through the app benefit students? Usability refers to the quality of a user’s experience when interacting with the products or systems, including websites, software, devices, or applications [4]. Usability is about effectiveness, efficiency, and the overall satisfaction of the user. Applications should be designed in a way that makes them comprehensible (so that it is easily understood, comfortable by using, required jobs can be done easily and faster) for the benefit of users. High usability is required for these applications in order to increase students’ knowledge bases. These apps have been evaluated through Heuristic Evaluation and Semiotic Evaluation and also give priority to customer experience. Based on their user interface, decoration some app has positive user rating some has negative. The entire app has some similar problem that was lack of proper content. It is quite easy because the maximum app is new and most of these are in developing mode. At last, by following all the problems like user experience and app Efficacy the necessary steps have been highlighted to increase the effectiveness, also for growing the acceptance and usage of these five apps in Bangladesh. Therefore, the aim of this study is to investigate and assess the usability of learning applications created in Bangladesh, to identify the features that make them less userfriendly for students, and to suggest potential design changes that may be made to these apps to increase their usability. In this research paper, to attain these objectives, from Google play store five learning apps namely Bacbon School [5], Classroom [6], Prottoy [7], Shikho [8], and Muktopaath [9] have been selected based on ratings, the number of downloads and features. The rest of the articles are organized as follows: in Sect. 2, related works along with theoretical background are presented, in Sect. 3, the research methodology with an overview. In Sect. 4, the data analysis and findings of this research are presented, and in Sect. 5, this study concluded with the outcomes and the discussion of the highlighted recommendations.
Evaluating Usability of E-Learning Applications in Bangladesh
131
2 Related Works In this pandemic time, the world runs with its way to make our locked life easy. Today’s world is dependent on the internet, and also dependent on different kinds of mobile application systems. People are showing an imperative interest in different kinds of mobile application systems. People are showing an imperative interest in different kinds of mobile based interactive solution. These applications are made up by applying various usability heuristic and semiotic heuristic rules. There has been a lot of research based on a semiotic and heuristic evaluation for many years now. Over the past few years, researchers have been focusing more on evaluation of mobile applications. Muaz et al. [10] have experimented with the truck hiring mobile application in Bangladesh. In the paper, researchers worked on 3 different mobile applications. Researchers were attempting to discover usability and semiotic problems from these three applications. Authors have find out usability heuristic problems and intuitiveness scores of the interface of the selected applications. In Digi truck mobile application authors have identified 20 usability heuristic problems, in truck Lagbe, 21 usability heuristic problems, and in Trux24, 27 usability heuristic problems were discovered. A study based on Pakistan by Bibi et al. [11] have evaluated different Islamic learning mobile applications. Author have evaluated around 50 Islamic mobile learning applications as the cornerstone to effectiveness, learnability, and user satisfaction which was selected based on popularity at the play store. Evaluators have categorized those fifty mobile applications into 3 catalogs, each application containing five applications. These categories were Duas and Kalimas, How to Learn Namaz, and99 Names of Allah. About 75 participants were participated in the study and all were children who are aged between five to twelve. In another study, Mubeen et al. [12] considered the available healthcare based mobile application and those application were chosen from two different digital platform (App Store and Play Store). All the mobile applications were selected based on the five features: the first one was assessment, the second was a questionnaire, the third one was privacy statement, the fourth one was Precautionary measures if test positive for COVID-19, and last one was plans of pandemic and region. In January 2020, some students at a university in Malaysia created their own TARC [13] mobile application to establish their communication and daily resource. TARC mobile applications are quite exoteric applications that helped their daily basis to access necessary resources comfortably. Researchers have attempted to evaluate the usability of the application by balancing with the Concordia mobile application from the aspects of effectiveness, efficiency, and satisfaction. It is observed that completion rate for the TARCApp was 86% and the Concordia application was 96%. The mean completion times for the TARCApp mobile application ranged from 1.88s to 4.66s and the Concordia mobile application ranged from 4.17s to 7.11s.
132
S. K. Dey et al.
In another work, the authors have evaluated the usability of the pregnancy tracker application in Bangladesh [14]. In the article, it applied heuristic rules on the two Bangladeshi mobile applications (MAA and Aponjon). Researchers have analyzed the usability of these two applications and applied them to maternity applications that was developed in Bangladesh and took steps to possible design solicitation for improving the overall usability of such applications. The findings of the study also discovered usability heuristic and semiotic heuristic and compared their number of heuristics. It concluded that the semiotic heuristic rule 17 was mostly violated in MAA apps for 6 times. In MAA application, a total of 23 usability heuristic and 45 semiotic heuristic problems were identified. However, in Aponjon, a total of 105 usability heuristics and 50 semiotic heuristic problems were discovered. The initiative score of Aponjon apps was lower than MAA apps, respective values were 2.7612 and 4.038.
3 Methodology This section explains the study’s methodology. First, we go into great depth on the applicant selection procedure. Next, the heuristic assessment strategy is explained, and after that, the semiotic evaluation method is covered. 3.1 Application Selection We have selected 5 applications from among many renowned learning applications in Bangladesh. These 5 applications are selected for Heuristic Evaluation (HE) and Semiotic Evaluation (SE) based on ratings, the number of installations, and the date of release. These 5 apps are, 1. “Bacbon” (rating: 4.5, number of installations: 10k+ and the release date: 5th August 2019, updated on: 26th July 2021), 2. “Prottoy-Visual Learning App” (rating: 4.7, number of installations: 1k+ and the release date: 21st February 2021, updated on: 12th July 2021), 3. “Shikho Learning App” (rating: 4.5, number of installations: 50k+ and the release date:27th October 2020, updated on: 6th June 2021), 4. “Classroom” (rating: 3.9, number of installations:100k+ and the release date:12th July 2019, updated on: 27 May 2021), 5. “Muktopaath” (rating: 3.3, number of installations: 100k+ and the release date: 17th June 2019, updated on: 22nd June 2021). The first one the “Bacbon School” is developed for those students who are making preparations for every public examination such as JSC, SSC, HSC, and admission tests. This app provides the opportunity for students for testing themselves after every video and provides a descriptive file about the video content so it becomes easier to learn. The second one is “Prottoy- Visual Learning App” which is a one-stop learning app, where learners would find complete course materials as per the NCTB curriculum. The third one is the “Shikho Learning App” for the Bangladeshi National Curriculum; SSC exam. Though this app is specifically designed for the NCTB and eventually covers a wide range of subjects for the JSC, SSC, and HSC curriculum and exams, the initial release of the app features a full exam preparation course for the SSC General Math Syllabus. The fourth one is “Classroom” which provides a comprehensive E-Learning platform for the students of SSC, HSC, and University Admission. Another related app to “Classroom” is “Classroom Parent” for parents which is the monitoring app to check the
Evaluating Usability of E-Learning Applications in Bangladesh
133
progress of a student’s education. Parents can check their children’s educational progress and can easily track the activity related to their children’s learning. And the last one is “Muktopaath” which is an e-learning platform for education, skills, and professional development. This app also offers courses and tutorials for the unemployed and underemployed youths in Bangladesh intending to encourage them in self-employment. And also this app provides digital certificates after completing each course. The authors of this article selected these 5 apps based on the given information. 3.2 Heuristic Evaluation Firstly, these five selected applications have been evaluated using heuristic evaluation based on [15, 16] heuristic and the followings are: “Visibility of system status and find ability of the mobile device (heuristic 1/ H1)”, “Match between system and the real world (H2)”, “User control and freedom (H3)”, “Consistency and standards (H4)”, “Flexibility and efficiency use (H5)”, “Recognition rather than recall (H6)”, “Aesthetic and minimalist design (H7)”, “Error prevention (H8)”. After that these problems were categorized based on severity ratings (0-4) proposed by Neilson [17]. In this concept where 0 indicates not a usability problem at all, 1 indicates a cosmetic problem, 2 indicates a minor usability problem, 3 indicates a major usability problem, and 4 indicates a catastrophic usability problem. The findings from the independent analysis of the 3 authors of this article have been aggregated. 3.3 Semiotic Evaluation Subsequently the obtained problems were analyzed through Semiotic Evaluation following the semiotic heuristic based on the proposal of Islam et al. [18, 19] (Table 1). The evaluation is based on the interface signs of the selected applications including signs, buttons, navigation bars, symbols, and other visual directives [20]. These signs are common to all selected applications for semiotic evaluation. Again, M. Speroni [21] proposed the W-SIDE (Web Semiotic Interface Design Evaluation) framework that pays explicit attention to the intuitiveness of user interface signs (even if it is the tiniest element of the UI) to design and assess web interfaces. Authors of this article have brought these interface elements under intuitiveness score (1 to 9) where 1-3 is recognized as low intuitiveness, 4-6 is recognized as middle intuitiveness and 7-9 is recognized as high intuitiveness score (1 is less and 9 is highly intuitive). And after that, again Neilson’s [17] severity rating was adopted.
4 Results Table 2 and Fig. 2 showed the result of the semiotic evaluation. Backbone has high intuitiveness (7-9) in all selected interface signs. Prottoy and Shikho both have seven high intuitiveness (7-9) and low intuitiveness for the remaining 1 icon. Classroom displays 4 high intuitiveness (7-9) for the selected interfaces, 1 medium intuitiveness (4-6), and 2 low intuitiveness for the remaining interfaces. Similarly, Muktopaath has 5 high intuitiveness (7-9) except 3 low intuitiveness (1-3) for the selected icons. The av-erages of the five apps are, Bacbon is 7.625, Prottoy is 6.875, Shikho is 7.125, Class-room is 5.625 and Muktopaath is 5.00.
134
S. K. Dey et al. Table 1. Semiotic Heuristics of the Side Framework
Level Syntactic
Pragmatic
Social
Environment Semantic
Semiotic Heuristics SH1. Clearly present the purpose of interactivity SH2. Make the effective use of color to design an interface sign SH3. Make the representamen readable and clearly noticeable SH4. Make a sign presentation clear and concise SH5. Create the representamen context appropriately SH6. Follow a consistent interface sign design strategy SH7. Place the interface sign in the proper position in a UI SH8. Make effective use of amplification features SH9. Create good relations among the interface signs of a UI SH10. Retain logical coherence in interface sign design SH11. Design interface signs to be culturally sensitive or reactive SH12. Match with the reality, conventions, or real-world objects SH13. Make effective use of organizational features SH14. Map with metaphorical and attributing properties SH15. Model the profiles of the focused end-users SH16. Make effective use of ontological guidelines SH17. Realize a match between a designer's encoded and a user’s decoded meaning.
Table 2. Intuitiveness score of the selected interface sign of 5 apps Interface Sign Logo Notifications Log out Query Buying course Learning progress Downloads Recent activity Average
Bacbon
Prottoy
Shikho
Classroom
Muktopath
8 6 8 8 8 7 8 8 7.625
8 8 8 7 7 8 1 8 6.875
8 8 8 1 8 8 8 8 7.125
8 1 8 8 6 6 1 7 5.625
8 7 7 7 1 1 8 1 5.00
The result showed that the most intuitive application from a semiotic perspective is Bacbon compared to the other 4 applications. It is clear from Fig. 1, that none of these applications have any catastrophic (severity rating 4) violation in semiotic evaluation. Bacbon and Shikho both have seven cosmetics (severity rating1) violations and one minor (severity rating 2) violation which occurred in Bacbon for notification symbols and Shikho for a query.
Evaluating Usability of E-Learning Applications in Bangladesh
135
Fig. 1. Number of problems to each Severity Rating
Prottoy violates five cosmetic (severity rating 1) problems, two minor (severity rating 2), and one major (severity rating 3) violations that occurred in the download option (no download option in this application). The classroom application has three cosmetic (severity rating 1) and three minor (severity rating 2) violations except for two major (severity rating 3) violations for notifications and downloads icons. The last one Muktopaath has two cosmetic problems, three minor problems, and three major problems. The result shows that maximum major (severity rating 3) violations have occurred in Muktopaath. Moreover, Bacbon and Shikho both have the maximum cosmetic problem (severity rating 1), and Muktopaath and Classroom both have the highest number of minor (severity rating 2) problems from all the selected applications. The Heuristic evaluation has been done on these five selected apps based on Bertiniet al. [15, 16] heuristics. Some examples are given to show how the evaluation was conducted. For “Bacbon School” in the profile section (Not Applicable) shown in Fig. 2(a) is not clear which violates the Heuristic H4. The severity of this violation has been graded as 2 by the evaluators. Another violation is that the app shows only the last action, it is unable to show all history, shown in Fig. 2(b). This violates the heuristic H6 and H3 with severity rating 3. In the second app, “Prottoy-Visual Learning App” it was important to keep the download option because without an internet connection the videos of this app cannot be viewed. Also, the method of subscription is not flexible, there are so many processes to buy a course. Figure 2(c) shows that nothing was said clearly about the “promo code”. Both violations violate The Heuristic H5 with a severity rating of 3. In “Shikho Learning App” there is no way to contact the tutor for any study-related problems or any queries which violate both The Heuristic H1 and H2 with severity rating 2. Another violation is the terms and conditions are not given which again violates The Heuristic H1 and alongside H3, was given a severity score 2 by all evaluators. In “Classroom” there are so many subject options with no videos, basic and concept papers, shown in Fig. 2(d) and Fig. 2(e). This violates both the Heuristics H1 and H2 with a severity score of 3. In another related app “Classroom Parents”, it would be easier for the parents to understand their child’s progress, if it had a brief description
136
S. K. Dey et al.
alongside the graph which violates The Heuristic H5 and the severity score was graded as 2. In the last one “Muktopaath”, three evaluators found that the feedback section shown in Fig. 2(f), represents a blank page, without showing any relevant message. This violates the heuristic H1 with a severity rating of 2. This app is also unable to show any recent activity which violates H6 and the severity was graded as 3.
Fig. 2. Some screenshots of bacbon (a, b), Prottoy (c), Classroom (d, e), and Muktopaath (f)
Table 3 showed that the number of violated heuristics for all eight heuristics are maximum (n = 10) in both Bacbon and Classroom apps on the other hand minimum (n = 4) in Shikho. Besides, the second maximum (n = 8) violations found in Muktopaath app and the second minimum (n = 5) on Prottoy. The table also showed that Bacbon is the app that violates all 8 heuristics, on the other hand just 3 heuristics violations occurred in Shikho app. Table 3. Usability problem to each heuristic Heuristics H1 H2 H3 H4 H5 H6 H7 H8 Total
Bacbon 1 1 2 2 1 1 1 1 10
Prottoy 1 0 1 0 2 0 1 0 5
Shikho 2 1 1 0 0 0 0 0 4
Classroom 2 1 0 2 3 1 0 1 10
Muktopaath 1 1 1 1 2 1 1 0 8
Here are some examples of semiotic evaluation in Fig. 3. The intuitiveness scores of the notification icons of the five apps which are Bacbon (Fig. 3(a)), Prottoy (Fig. 3(b)), Shikho (Fig. 3(c)), Classroom (not available) and Muktopaath (Fig. 3(d)) were 6,8,8,1 and 7 respectively. The notification icon of Bacbon is placed within the list of the
Evaluating Usability of E-Learning Applications in Bangladesh
137
dashboard with other buttons which violate a pragmatic heuristic (SH7) and there is no notification icon in the classroom app, for this reason, its intuitiveness score is very low. On the other side, icons of buying courses of these 5 apps are holding the score 8,7,7,6 and 1 respectively.
Fig. 3. Notification sign (a)Bacbon, (b)Prottoy, (c)Shikho, (d)Classroom (no notification option), and Muktopaath
In semiotic evaluation there have also some rules on the based on graphical interface. Those are “logo, notification, exit, query, buying course, learning progress, download, recently activity”. For the experiment on the semiotic evaluation the interface signs in Table 4 have been considered and selected.
138
S. K. Dey et al. Table 4. Interface sign selected for Semiotic Evaluation
Interfacesign Logo
Bacbon
Prottoy
Shikho
Classroom
Muktopath
Notification Not ble
Availa-
Logout/Exit Query
Not Available
Course Buy
Doesn’t show properly Not Available
Learning progress Downloads Recent activity
Not Available
Not Available Not Available
5 Conclusion This study makes these five programs more accessible to users, mostly students, by giving a comprehensive understanding of their present usability status and prevalent problems. The information obtained from the study will help to further develop and improve application performance significantly so many more users are encouraged to use these applications. The results of the study would be extremely helpful to practitioners in creating study-related applications in Bangladesh with enhanced usability and user experience so that these apps may offer users the most advantages. Students should be provided an easier way for data input, self-error control, the flexibility of buying courses, and the video download process. It needs to be ensured for novice users. The meaning of the interface signs should be more consistent with their convention. The symbol used as an interface sign should be more compatible with their elaborate meaning of having a constant phase relationship with users. The evaluations are made separately by each evaluator without including any common users. Again, this research only focused on ELearning applications in the context of Bangladesh to evaluate by principles of heuristic and semiotic evaluations. Future studies may use different applications to explore deeper usability issues.
Evaluating Usability of E-Learning Applications in Bangladesh
139
References 1. Teacher shortage makes education recovery a stiff job, https://www.tbsnews.net/bangla-desh/ education/teacher-shortage-makes-education-recovery-stiff-job-206482. Accessed 30 May 2022 2. Teacher-student ratio not ideal at many public universities | Daily Sun |, https://www.dailysun.com/printversion/details/331748/Teacherstudent-ratio-not-ideal-at-many-public-uni-ver sities. Accessed 30 May 2022 3. Half of world’s student population not attending school: UNESCO launches global coalition to accelerate deployment of remote learning solutions, https://en.unesco.org/news/half-wor lds-student-population-not-attending-school-unesco-launches-global-coalition-acceler-ate. Accessed 30 May 2022 4. Moumane, K., Idri, A., Abran, A.: Usability evaluation of mobile applications using ISO 9241 and ISO 25062 standards. Springerplus 5(1), 1–15 (2016). https://doi.org/10.1186/s40064016-2171-z 5. BacBon School, https://bacbonschool.com/. Accessed 30 May 2022 6. Classroom: Crunchbase Company Profile & Funding, https://www.crunchbase.com/organization/. Accessed 30 May 2022 7. Prottoy | Click. Learn. Excel., https://www.prottoy.com.bd/. Accessed 30 May 2022 8. Shikho Academic Program, https://shikho.tech/. Accessed 30 May 2022 9. Muktopaath- Bangla eLearning Platform, https://www.muktopaath.gov.bd/. Accessed 30 May 2022 10. Muaz, M.H., Islam, K.A., Islam, M.N.: Assessing the usability of truck hiring mobile applications in Bangladesh using heuristic and semiotic evaluation. In: Advances in Design and Digital Communication: Proceedings of the 4th International Conference on Design and Digital Communication, Digicom 2020, November 5–7, 2020, Barcelos, Portugal, pp. 90–101 (2020) 11. Bibi, S.H., Munaf, R.M., Bawany, N.Z., Shamim, A., Saleem, Z.: Usability evaluation of Islamic learning mobile applications. J. Islamic Sci. Technol. 6(1), 1–12 (2020) 12. Mubeen, M., Iqbal, M.W., Junaid, M., Sajjad, M.H., Naqvi, M.R., Khan, B.A., Tahir, M.U.: Usability evaluation of pandemic health care mobile applications. In: IOP Conference Series: Earth and Environmental Science, Vol. 704, No. 1, p. 012041. IOP Publishing (2021) 13. Tunku Abdul Rahman University College - ICDXA, https://www.tarc.edu.my/icdxa/icdxa2020/. Accessed 30 May 2022 14. Kundu, S., Kabir, A., Islam, M.N.: Evaluating usability of pregnancy tracker applications in Bangladesh: a heuristic and semiotic evaluation. In: 2020 IEEE 8th R10 Humanitarian Tech-nology Conference (R10-HTC), pp. 1–6. IEEE (2020) 15. Bertini, E., Gabrielli, S., Kimani, S., Catarci, T., Santucci, G.: Appropriating and assessing heuristics for mobile computing. In: Proceedings of the Working Conference on Advanced Visual Interfaces, 119–126 (2006) 16. Bertini, E., Catarci, T., Dix, A., Gabrielli, S., Kimani, S., Santucci, G.: Appropriating heuristic evaluation for mobile computing. Int. J. Mob. Hum. Comput. Interact. (IJMHCI) 1(1), 20–41 (2009) 17. Severity Ratings for Usability Problems: Article by Jakob Nielsen, https://www.nngroup. com/articles/how-to-rate-the-severity-of-usability-problems/. Accessed 30 May 2022 18. Islam, M.N., Bouwman, H.: Towards user–intuitive web interface sign design and evaluation: a semiotic framework. Int. J. Hum. Comput. Stud. 86, 121–137 (2016) 19. Islam, M.N., Bouwman, H.: An assessment of a semiotic framework for evaluating userintuitive Web interface signs. Univ. Access Inf. Soc. 14(4), 563–582 (2015). https://doi.org/ 10.1007/s10209-015-0403-6
140
S. K. Dey et al.
20. Islam, M.N.: A systematic literature review of semiotics perception in user interfaces. J. Syst. Inf. Technol. 15(1), 45–77 (2013) 21. Speroni, M., Paolini, P.: Mastering the Semiotics of Information-Intensive Web Interfaces 212
Perceptual Computing Based Framework for Assessing Organizational Performance According to Industry 5.0 Paradigm Danylo Tavrov1(B) , Volodymyr Temnikov2 , Olena Temnikova1 , and Andrii Temnikov2 1 National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”,
37 Peremohy Ave., Kyiv 03056, Ukraine [email protected] 2 National Aviation University, 1 Kosmonavta Komarova Ave., Kyiv 03680, Ukraine
Abstract. The new economic paradigm of Industry 5.0, unlike Industry 4.0, puts greater emphasis on fostering human-centric, sustainable, and resilient industrial practices, and is focused on human well-being and shifting towards new forms of sustainable economics. However, the concept is still at a nascent stage and remains largely ill-defined. Existing organizational performance approaches do not take into account the new goals of Industry 5.0, largely because of their subjective and imprecise nature. In this paper, we propose a formalized framework for evaluating performance of an organization in terms of compliance to subjective and ill-defined Industry 5.0 criteria. Our framework is based on perceptual computing and enables the regulators to express their subjective opinions using words. We illustrate applicability of our framework to a specific task of evaluating the functional state of a person as one of the core aspects of human well-being. Given uncertainties involved in measuring appropriate physiological, psychological, mental, and physical characteristics, we show that perceptual computing can allow us to assess such criteria using words, which makes the whole process easily adoptable by regulators. Keywords: Perceptual computing · Type-2 fuzzy set · Industry 5.0 · Functional state of a person
1 Introduction Extensive literature on evaluating organizational performance focuses on defining various measures and metrics, which can help ensure high overall performance of an organization. It has long been recognized that financial and economic indicators such as return on assets, return on capital, economic value added, and others, when taking alone, lead to [1, p. 689] short-term performance at the expense of long-term performance, by reducing expenditure on new product development, human resources, customer development, and so on. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 141–151, 2023. https://doi.org/10.1007/978-3-031-36115-9_14
142
D. Tavrov et al.
Multidimensional performance measure approaches try to consider many aspects of organizational performance at once. For instance, a widely used performance prism [2] aims at measuring different facets, including defining stakeholder satisfaction, laying out strategies and processes needed to achieve it, defining the capabilities needed to achieve it, and involving contributions of stakeholders. Another popular approach called balanced scorecard [3] suggests a wide range of metrics to be tracked to assess organization’s success, including operational improvement metrics, metrics for innovation, employee capabilities, and others. Some authors argue [4] that performance management practices work better when combined with human resource management techniques, as maintaining organizational social climate is key to improving efforts of the employees. However, these and other approaches presented in the literature are not very formalized and often represent broad and generic recommendations. New challenges such as climate change, biodiversity deterioration, and sustainability issues require new approaches to measuring organizational performance. European Commission proposes a new concept of Industry 5.0 [5] in place of Industry 4.0, which was profit-driven and focused on enhanced efficiency by incorporating technological advancements, artificial intelligence, and digitalization into production processes [6]. The three pillars of Industry 5.0 are [5]: – human-centric approach, whereby technological advances are used to adapt production to workers’ needs, with an emphasis on training and well-being; – sustainability, which means that circular processes are needed to decouple resource and material use from negative impact on the environment; – resilience, which means that industrial production must become more robust against disruptions and crises. Thus, a new framework for evaluating organizational performance is needed. Industry 5.0 paradigm requires to re-imagine existing metrics and indicators, to allow [7] measurement of the above criteria and guide accelerated compliance to them. Many of new metrics and indicators will be qualitative. It is also recognized [7] that evaluation of compliance should be more “user-friendly.” In our opinion, a holistic approach, when objective parameters are supplemented with linguistic and qualitative data, is essential. The concept of Industry 5.0 is significantly understudied. Sustainability dimension of Industry 5.0 was analyzed in [8] using interpretive structural modeling. Authors of [9] propose a reference model for the human-centric dimension. However, no comprehensive approach exists for evaluating organizational performance in terms of Industry 5.0, taking into consideration its imprecise and ill-defined criteria. In our opinion, fuzzy methods should be applied, which are tailored to subjective and imprecise data. Using subjective measures for organizational performance is not new [10], and methods such as fuzzy TOPSIS [11], fuzzy AHP [12], and other fuzzy methods for multicriteria decision making [13, 14] were successfully used. In this work, we propose a novel framework for assessing organizational performance according to Industry 5.0 criteria. We formalize the whole process by offering an hierarchical structure of organizational performance, where each specific indicator describing the organization is evaluated using words, which are then processed using computing with words (CWW) methodology to arrive at the overall assessment. According to [15],
Perceptual Computing Based Framework for Assessing Organizational Performance
143
CWW is a way of “dealing with real-world problems [by means of] exploiting the tolerance for imprecision,” which is superior to applying ordinary fuzzy sets. Application of CWW can enable regulators to perform evaluation using words, treat subjective judgments in a mathematically coherent way, and reach easily interpreted decisions that can be used to formulate recommendations. The rest of the paper is organized as follows. Section 2 describes a proposed framework for assessing compliance to Industry 5.0 criteria. In Sect. 3, we discuss one specific area related to functional states of employees. We detail the inner workings of a CWW procedure and offer a model exercise that illustrates the proposed approach. We conclude in Sect. 4 and indicate directions for further research.
2 Framework for Assessing Organizational Performance 2.1 Outline of Assessment Process In the framework we propose, assessment of organizational performance for compliance with the criteria of Industry 5.0 is done through an hierarchical and distributed process [16] according to Fig. 1, which is based on [5, 17]. Evaluations at lower levels are aggregated and propagated forward, level by level, until a final decision is reached. Indicators numbered from 1 to N in Fig. 1, which are fed into the lowest-level nodes, can be evaluated as any of the following: – binary values corresponding to either absence or presence of a given characteristic (e.g., whether certain recycling practices are in place or not); – numbers, obtained by objective measurements (e.g., levels of carbon emissions). Every number needs to be converted to a [0; 1] scale, so that higher numbers contribute more to the overall performance level; – words, which correspond to subjective, ill-defined, or immeasurable indicators (e.g., employee satisfaction or engagement). Arcs connecting indicators to each other are weighted, to allow for possibility of some indicators having more influence on the overall decision. Weights are expressed as words, which are customary in practice of perceptual computing and reflects imprecise nature of interrelations between different indicators. Some weights can be set to 0, if lower-level indicators have no effect on a given middle-level indicator. Level of compliance of each middle-level indicator in Fig. 1 such as “Humantechnology collaboration” or “Supply chains” is obtained as a weighted average of all lower-level indicators it is connected to (not shown), weighted accordingly. Such indicators can be interesting in themselves, as corresponding performance levels can be used to identify weak links, or to make recommendations for improvement. The general architecture of the decision-making process shown in Fig. 1 captures all the main aspects of Industry 5.0 paradigm. Of course, for each specific industry and even individual enterprise, indicators can be added, removed, or modified. 2.2 Representation of Words as Type-2 Fuzzy Sets To process indicators expressed as words, we need to represent them in a computer. Often words are represented as fuzzy sets [18]. A fuzzy set A is characterized by a membership
144
D. Tavrov et al.
function µA (x): X → [0; 1]. Unlike an ordinary set, to which any element either belongs or does not, any element x can belong to a fuzzy set A with some membership degree µA (x).
Fig. 1. Framework for assessing organizational performance according to Industry 5.0 criteria
However, type-1 fuzzy sets are not sufficient to model words, and instead we use type-2 fuzzy sets [19], in particular the so-called interval type-2 fuzzy sets (IT2FS). Each IT2FS à is characterized by a lower membership function (LMF), µÃ (x), and an upper membership function (UMF), µÃ (x). Membership of each element x in an IT2FS is described by an interval [µÃ (x); µÃ (x)], which captures the uncertainty associated with a linguistic concept being modeled. When LMF and UMF are trapezoidal, i.e. when their functional form is ⎧ x−a ⎪ ,a≤x≤b ⎪ ⎪ b−a ⎨ 1, b≤x≤c µ(x; a, b, c, d ) = d −x (1) ⎪ ⎪ d −c , c ≤ x ≤ d ⎪ ⎩ 0 , otherwise we obtain what is called a trapezoidal IT2FS (Fig. 2). The shaded area is called the footprint of uncertainty (FOU). Thus, each trapezoidal IT2FS à used in this paper can be uniquely represented as a tuple (al , ar , ol , or , bl , br ). 2.3 Computing with Words in the Perceptual Computer As noted above, in this work, we propose to use one of the existing methodologies for CWW called perceptual computing, which is described in its entirety in [16]. A perceptual
Perceptual Computing Based Framework for Assessing Organizational Performance
145
computer (Per-C) is a specific architecture for CWW, in which words are used in decision making in three stages:
Fig. 2. Trapezoidal interval type-2 fuzzy set
– input words are mapped to their IT2FS models according to a codebook, which maps each word to an appropriate IT2FS; – at each node in Fig. 1, input IT2FSs are processed using linguistic weighted average (LWA): n
YLWA =
Xi Wi
i=1 n
(2) Wi
i=1
where X i are IT2FSs representing evaluations, and W i are IT2FSs representing weights on corresponding arcs. At each node we obtain output IT2FSs, which are then fed as inputs to upper-level nodes, and so on until the overall performance is determined. Discussion of LWA lies beyond the scope of our paper; – the IT2FS corresponding to “Overall performance” can be used to rank different companies, or it can be converted into an aggregate compliance level, expressed either as a word from the codebook or as a number from [0; 1]. To condense an IT2FS Ã to a number, average centroids can be applied: cl A˜ + cr A˜ c A˜ = 2
(3)
where cl (Ã) and cr (Ã) can be computed, e.g., using an algorithm described in [20].
3 Subdomain of Assessing Functional State of Employees 3.1 Assessing Functional State as Part of Human-Centric Approach In this section, we reuse and modify our previous approach in [21] and fully specify application of Per-C to a subdomain of assessing functional states (FuncS) of employees. According to Fig. 1, it is part of the human-centricity criterion, which is aligned with
146
D. Tavrov et al.
recommendations of [5], where it is stated that under Industry 5.0 guidelines, the wellbeing of employees should encompass not only physical health in the workplace, but also mental health, especially in the current era of digital technologies that can lead to frequent burnouts. By functional state we understand a complex of indicators shown in Fig. 3. Assessment of FuncS is carried out hierarchically, with each level corresponding to a certain group of indicators. These indicators are by no means exhaustive and/or obligatory, and can be modified depending on each specific application. Experts assess FuncS of employees either objectively (e.g., by conducting medical tests or examinations) or subjectively (e.g., to assess psychological state of an employee). Different indicators can be evaluated using either words (from a corresponding codebook) or numbers. Each indicator is evaluated independently. For the sake of concreteness, indicators in Fig. 3 can be evaluated as follows: – indicators A, B, E, F, L are evaluated using objective tests and protocols, and then mapped into [0; 1], where higher numbers stand for higher compliance; – indicators C, D, G, H, I, J, K are evaluated by experts using words, because they are highly subjective and do not lend themselves to objective testing; – indicators M–Q are computed using LWA of respective lower-level indicators, as shown by connecting lines in Fig. 3. We can use numbers instead of words because each number x can be considered an IT2FS with parameters al = ar = ol = or = bl = br = x. 3.2 Components of the Perceptual Computer In the Per-C discussed in this paper, words belong to two different categories: – words used by experts to set levels of compliance of individual indicators; – words assigned by (a different set of) experts to each arc in Fig. 3 as weights. All the words are represented as IT2FS with the same universe of discourse, [0; 1]. Adhering to recommendations from [21, 22], we chose the following words: – for assessment: good (G), normal (N), satisfactory (S), acceptable (A), unsatisfactory (U); – for weights: essential effect (Es), significant effect (Sg), average effect (Av), insignificant effect (In), little or no effect (Lt). The same words can have different IT2FS representation. Experts from different domains can have different levels of uncertainty associated with the same word. We propose to create two codebooks: one for evaluating indicators C–D and K from Fig. 3, and another one with the same words but used for evaluating indicators G–J. We will index these words by 1 or 2 to distinguish the codebooks they belong to. IT2FS models for each word should be created using, for instance, the Hao-Mendel approach [23] or similar. As we did not have access to many experts needed for this approach, we modeled words using a person FOU approach [24], which involves interviewing only one expert. The third author of this paper, based on her expertise, provided data to create trapezoidal IT2FS models (1) of all the words. Corresponding parameters
Perceptual Computing Based Framework for Assessing Organizational Performance
147
are given in Table 1 in reference to Fig. 2 (all the numbers in this paper are shown up to two significant digits). Weights corresponding to arcs from Fig. 3 were assigned as words from the codebook by the second author of this paper based on his expertise (Table 2).
Fig. 3. Graph model of the functional state of enterprise employees
Table 1. Parameters of IT2FSs used to represent words from the codebook Word
Parameters al
ar
ol
or
bl
br
G1
0.92
0.93
0.95
1.00
1.00
1.00
N1
0.79
0.81
0.87
0.93
0.95
0.97
S1
0.68
0.70
0.74
0.82
0.87
0.89
A1
0.60
0.60
0.64
0.70
0.74
0.76
U1
0.00
0.00
0.00
0.60
0.64
0.65
G2
0.83
0.86
0.92
1.00
1.00
1.00
N2
0.73
0.74
0.78
0.87
0.92
0.94
S2
0.62
0.63
0.63
0.74
0.78
0.79
A2
0.54
0.55
0.55
0.63
0.65
0.66
U2
0.00
0.00
0.00
0.54
0.57
0.59
Es
0.86
0.88
0.92
1.00
1.00
1.00
Sg
0.63
0.66
0.75
0.87
0.91
0.94
Av
0.37
0.40
0.50
0.67
0.76
0.78
In
0.19
0.25
0.35
0.43
0.52
0.55
Lt
0.00
0.00
0.00
0.25
0.37
0.39
148
D. Tavrov et al. Table 2. Linguistic weights of arcs in the Per-C
Arc
Weight
Arc
Weight
A–M
Sg
E–N
Sg
B–M
Es
F–N
Es
C–M
Av
G–N
Es
D–M
Lt
H–N
Av
Arc
Weight
Arc
Weight
I–O
Es
M–Q
Sg
J–O
Sg
N–Q
Es
K–P
In
O–Q
Sg
L–P
Av
P–Q
Av
3.3 Results for a Model Exercise To illustrate applicability of our approach to assessing subjective and imprecise criteria such as functional state, we created model data for five synthetic employees with different levels of compliance to specified indicators (Table 3). Evaluations for each indicator were assigned by the fourth author of this paper based on his experience. Resulting IT2FSs for each employee are given in Fig. 4. To complement visual representation, Table 3 shows resulting centroids (3), which can be used, for instance, to rank the employees, making this submodule useful in its own right. As part of a bigger framework in Fig. 1, LWA is applied to the IT2FS from Fig. 4 and corresponding IT2FS calculated at other nodes to get an aggregate level of “Safe and inclusive work environment,” and so on up the hierarchy until the overall performance is determined. Results for employees 2 and 5 are especially interesting. Their IT2FSs and centroids are very close, even though they differ in indicators B and F, which are “essential” for estimating the overall state. Thus, the Per-C takes into account the whole multitude of parameters and reaches conclusions that are robust to imprecise and subjective data.
Fig. 4. Interval type-2 fuzzy sets representing FuncS of five synthetic employees
Perceptual Computing Based Framework for Assessing Organizational Performance
149
Table 3. Evaluations of indicators from Fig. 3 for five synthetic employees and centroids of the corresponding output IT2FSs #
Evaluation
Centroid
A
B
C
D
E
F
G
H
I
J
K
L
cl
cr
1
0.95
0.97
N1
N1
0.93
0.95
G2
G2
G2
G2
N1
0.98
0.93
0.94
2
0.88
0.96
N1
S1
0.92
0.96
G2
N2
N2
N2
S1
0.75
0.86
0.87
3
0.70
0.73
A1
S1
0.82
0.85
S2
A2
A2
A2
S1
0.72
0.70
0.71
4
0.94
0.97
N1
N1
0.71
0.62
S2
A2
S2
S2
N1
0.98
0.78
0.79
5
0.89
0.87
N1
S1
0.99
0.87
G2
N2
G2
G2
S1
0.72
0.87
0.88
4 Conclusions and Further Research In this work, we introduced a new framework for assessing organizational performance in accordance with the concept of Industry 5.0. It captures three main pillars of Industry 5.0—human-centric view of production, sustainability, and resilience. The overall performance level is determined using perceptual computing that allows us to evaluate adherence to subjective and ill-defined criteria of Industry 5.0 using words. The proposed framework enables the regulators to compare different enterprise by ranking them according to the centroids of the overall assessment levels, as well as track the dynamics over some period of time, e.g. from year to year. We treated in more detail a subdomain of human-centricity related to assessment of functional states of employees and discussed how the obtained results fit in the general framework. To assess the FuncS, one needs to combine objective measurements and subjective evaluations of psychological and mental state of employees, which is why we decided to use perceptual computing. We described in detail a codebook for our perceptual computer and a method of converting the output IT2FS into a number. Model exercise related to assessing FuncS of synthetic employees showed that the perceptual computer is user-friendly, and the results produced by it are easily interpreted. Further research can be led in the direction of full specification of a working prototype of a perceptual computer that can be used to evaluate organizational performance using Industry 5.0 criteria. Work towards implementation of this approach in practice could help guide the process of defining appropriate metrics and measures to be used as inputs to such a system.
References 1. Langfield-Smith, K., Smith, D., Andon, P., et al.: Management Accounting: Information for Creating and Managing Value, 8th edn. McGraw-Hill, North Ryde (2017) 2. Neely, A., Adams, C., Crowe, P.: The performance prism in practice. Meas. Bus. Excell. 5(2), 6–12 (2001) 3. Kaplan, R.S.: Conceptual foundations of the balanced scorecard. In: Chapman, C.S., Hopwood, A.G., Shields, M.D. (eds.) Handbook of management accounting research, Vol. 3, pp. 1253–1269 (2009)
150
D. Tavrov et al.
4. Bourne, M., Pavlov, A., Franco-Santos, M., Lucianetti, L., Mura, M.: Generating organisational performance: the contributing effects of performance measurement and human resource management practices. Int. J. Oper. Prod. Manag. 33(11/12), 1599–1622 (2013) 5. Breque, M., De Nul, L., Petridis, A.: European Commission, Directorate-General for Research and Innovation. Industry 5.0: towards a sustainable, human-centric and resilient European industry, Publications Office (2021) 6. Aliyev, A.G., Shahverdiyeva, R.O.: Scientific and methodological bases of complex assessment of threats and damage to information systems of the digital economy. IJIEEB 14(2), 23–38 (2022) 7. Renda, A., Schwaag, S.S., Tataj, D., et al.: European Commission, Directorate-General for Research and Innovation. Industry 5.0, a transformative vision for Europe: governing systemic transformations towards a sustainable industry. Publications Office of the European Union (2022) 8. Ghobakhloo, M., Iranmanesh, M., Mubarak, M.F., et al.: Identifying industry 5.0 contributions to sustainable development: a strategy roadmap for delivering sustainability values. Sustain. Prod. Consumption 33, 716–737 (2022) 9. Lu, Y., Zheng, H., Chand, S., et al.: Outlook on human-centric manufacturing towards Industry 5.0. J. Manuf. Syst. 62, 612–627 (2022) 10. Singh, S., Darwish, T.K., Potocnik, K.: Measuring organizational performance: a case for subjective measures. Br. J. Manag. 27, 214–224 (2016) 11. Gupta, H.: Assessing organizations performance on the basis of GHRM practices using BWM and Fuzzy TOPSIS. J. Environ. Manage. 226, 201–216 (2018) 12. Modak, M., Pathak, K., Ghosh, K.K.: Performance evaluation of outsourcing decision using a BSC and Fuzzy AHP approach: a case of the Indian coal mining organization. Resour. Policy 52, 181–191 (2017) 13. Verma, R., Sharma, B.D.: A new inaccuracy measure for fuzzy sets and its application in multi-criteria decision making. IJISA 6(5), 62–69 (2014) 14. Dey, P.K., Ghosh, D.N., Mondal, A.C.: IPL team performance analysis: a multi-criteria group decision approach in fuzzy environment. IJITCS 7(8), 8–15 (2015) 15. Zadeh, L.A.: Fuzzy logic = computing with words. IEEE Trans. Fuzzy Syst. 4(2), 103–111 (1996) 16. Mendel, J.M., Wu, D.: Perceptual Computing. Aiding People in Making Subjective Judgments. John Wiley & Sons Inc, Hoboken, New Jersey (2010) 17. European Commission, Directorate-General for Communication, Towards a sustainable Europe by 2030: reflection paper, Publications Office (2019) 18. Zadeh, L.A.: Fuzzy sets. Inf. Control 8(3), 338–353 (1965) 19. Mendel, J.M.: Type-2 fuzzy sets and systems: an overview. IEEE Comput. Intell. Mag. 2(1), 20–29 (2007) 20. Wu, D., Nie, M.: Comparison and practical implementations of type-reduction algorithms for type-2 fuzzy sets and systems. In: FUZZ-IEEE 2011, pp. 2131–2138 (2011) 21. Tavrov, D., Temnikova, O., Temnikov, V.: Perceptual computing based method for assessing functional state of aviation enterprise employees. In: Chertov, O., Mylovanov, T., Kondratenko, Y., Kacprzyk, J., Kreinovich, V., Stefanuk, V. (eds.) ICDSIAI 2018. AISC, vol. 836, pp. 61–70. Springer, Cham (2019). https://doi.org/10.1007/978-3-319-97885-7_7 22. Pavlenko, P., Tavrov, D., Temnikov, V., Zavgorodniy, S., Temnikov, A.: The method of expert evaluation of airports aviation security using perceptual calculations. In: 2018 IEEE 9th International Conference on Dependable Systems, Services and Technologies (DESSERT), pp. 406–410 (2018)
Perceptual Computing Based Framework for Assessing Organizational Performance
151
23. Hao, M., Mendel, J.M.: Encoding words into normal interval type-2 fuzzy sets: HM approach. IEEE Trans. on Fuzzy Systems 24(4), 865–879 (2016) 24. Mendel, J.M., Wu, D.: Determining interval type-2 fuzzy set models for words using data collected from one subject: Person FOUs. In: 2014 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE), pp. 768–775 (2014)
Petroleum Drilling Monitoring and Optimization: Ranking the Rate of Penetration Using Machine Learning Algorithms Ijegwa David Acheme1 , Wilson Nwankwo3(B) , Akinola S. Olayinka2 , Ayodeji S. Makinde1 , and Chukwuemeka P. Nwankwo1 1 Department of Computer Science, Edo State University, Uzairue, Nigeria 2 Department of Physics, Edo State University, Uzairue, Nigeria 3 Department of Cyber Security, Delta State University of Science and Technology,
Ozoro, Nigeria [email protected]
Abstract. Oil prospection and exploration is a major economic venture requiring enormous expertise and technologies in developing countries. In this study, we set out to examine an important issue that affects the efficiency of oil drilling operations that is, the rate of penetration (ROP) during oil exploration. We attempt to apply different machine learning algorithms in order to discover the most efficient algorithm that could be used for the prediction of the ROP. We use the North Sea oil field data set while eighteen intelligent models were developed. On comparative analysis of the models, it is shown that the random forest algorithm exhibits the best efficiency in the prediction workflow having an RMSE value of 0 0010, and an R2 score of 0 891. Findings also showed that significant parameters of concern during drilling operations were the measured depth, bit rotation per minute, formation porosity, shale volume, water saturation, and log permeability respectively. We conclude that the result of this study provides a good template to guide the selection of machine learning algorithms to drive solutions for the optimization of oil drilling operations. Keywords: Petroleum drilling · Drilling operations · Rate of Penetration · Machine learning · Optimization
1 Introduction In different jurisdictions including developed countries oil and mineral drilling operations had been designated as very costly ventures in the oil and mineral production workflow [1–5]. Hence continuous research efforts, targeted towards the optimization of the drilling process so as to ensure a reduction in the total costs associated with the drilling process has been reported. However, despite the constant improvement in tools, products, and processes, machine learning methods have only very recently began to play © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 152–164, 2023. https://doi.org/10.1007/978-3-031-36115-9_15
Petroleum Drilling Monitoring and Optimization: Ranking the Rate of Penetration
153
significant roles in oil drill optimization, this has been mostly enhanced by the current availability of massive datasets [6]. Machine learning (ML) and Artificial Intelligence (AI) models have been recognized as pervasive tools that find interesting applications in the various sectors where intelligent computerization is required [7–19]. In the domain of oil exploration and drilling, ML and AI also hold promising expectations. This is because, drilling and oil exploration operations are data and knowledge-driven, hence, efficient processing of the massive amounts of data generated from several sensor networks at the oil rigs to aid cannot be overemphasized. Presently, there are implementations of Real-Time Operation Centers (RTOC) with major oil companies spending thousands of dollars, in the IT infrastructure to read drilling data from rigs in real-time. These readings enable experts to analyze real-time data in the centers, leading to faster decision making, reduction in stuck pipe incidents, less hole cleaning issues and fluid losses events, while also increasing the number of wells that can be monitored with the same number of personnel [20]. The availability of this data has also provided the necessary foundation for the implementation of artificial intelligence and machine learning techniques for the development of smart models for more accurate and robust real-time drilling performance monitoring and optimization. With current drilling rigs holding massive amounts of instrumentation for the collection of parameters from almost every equipment installed in the drilling rig, using sensors measuring their states, and allowing for remote and safe operations, there hasbeen an exponential growth of data generated at oil rigs which has laid the foundation for the development of machine learning models for predictive analytics and development of decision support systems. As researchers continue to explore these datasets generated at oil rigs, one challenge comes to bear, selecting the right machine learning algorithms and features for the effective prediction of ROP because the ROP is influenced by several factors which have complex relationships, the extent of the influence of these factors also vary as some are more relevant than others. A machine learning model built with many of the lowly influential factors or with a less efficient algorithm is not likely to give satisfactory results, hence this research work sets out to investigate and rank the factors that have been reported in literature for ROP prediction as well as implement and compare several ML algorithms. Identifying the most important factors will improve prediction performance, computation speed and training time while providing better explain ability for these models [21].
2 Related Literature The rate of penetration (ROP) can simply be defined as the speed at which a wellbore is being drilled. This can be manually calculated by measuring the depth per time intervals in units of feet or meters per hour. The high ROP values imply fast drilling which in turn emphasize an increased productivity of the drilling process. Because ROP is such a direct measurement of the overall time required to drill an oil well, reducing this time in order to achieve higher ROP is a key optimization strategy for oil companies. In this section, different strategies that have been employed in optimizing ROP are presented. These strategies have focused on modelling and predicting the ROP using selected drilling
154
I. D. Acheme et al.
parameters which can be controlled on the surface such as weight-on-bit (WOB), rotary speed (RPM) etc., and they can be broadly classified into traditional and data-driven models Traditional models are mathematical equations that have been formulated from experiments and field experience, while data- driven models refer to machine learning algorithms for the prediction of ROP. 2.1 Traditional ROP Models According to [22], one of the earliest mathematical models for ROP prediction was established by Maurer in 1962 who used a rock cratering technique to create a formula with the parameters as; bit diameter, rock strength, weight on bit (WOB), and rotations per minute (RPM). The Bingham model, reported in [23] is another early mathematical equation-based ROP prediction model, it uses comparable input parameters with an additional empirical constant, “k,” which represents a parameter that was dependent on formation. Eckel presented another early traditional model in [24] this model investigated the effects of mud on ROP. Among these early models, Bourgoyne and Young’s (BY) model [25] has received the most attention and been widely reported. It included additional geological and physical factors which included; the formation strength, normal compaction trend, under compaction, differential pressure, bit diameter and bit weight, rotational speed, tooth wear, and bit hydraulics. 2.2 Data Science Models The application of data science and machine learning techniques for the prediction and optimization of ROP seeks to build predictive models of ROP using data collected during drilling. Such models make use of parameters measured at the surface like weight on bit, rotations per minute, and flow rate as input variables in order to predict the rate of penetration. Generally, the process of oil drilling involves massive collection of data from both surface and sub-surface areas using IOT sensors. These sensors are able to collect large volumes of data about the state of the bit underground. The data collected are used to plot, analyze and control bit performance in this case the ROP. The availability of these datasets have laid the foundation for the building of ML models for the prediction of ROP because ML models the relationship between input variables in order to predict an output (target) variable. In these section we review different studies that have presented ML models for ROP prediction. Research works on the implementation of neural networks machine learning algorithm have been mostly reported. These neural network models have utilized different input parameters for ROP prediction [26]. [27] presented a hybrid neural network model that utilized Savitzky–Golay (SG) smoothing filter for removing noise from extracted data for the prediction of rate of penetration. [5] Presented a feedforward neural network model for the prediction of rate of penetration. The work utilized selected features such as; bit rotations per minute, mud flow, weight on bit, and differential pressures. These features comprised the input variables. Datasets utilized for the development of their model were obtained from laboratory simulations as well an oil field. The predicted ROP values are then used to identify bit malfunction or failure by comparison of the
Petroleum Drilling Monitoring and Optimization: Ranking the Rate of Penetration
155
predicted values with the actual measured value. Any observed deviation indicates an underperformance of the bit and this can serve as a warning flag. In the study by [28], an artificial neural network (ANN) model was used to predict hydraulics pump pressure and to provide early warnings, the model was implemented through the fitting tool of MATLAB, the sensitivity of the selected input parameters were analyzed by using a forward regression method. Data sets were obtained from selected well samples and were used for the validation of the model. The model predicted pump pressure against well depth in similar formations. While neural networks are powerful tools which have been very efficient in handling high dimensional modelling. The works in [29–35] argue that they generally underperform in comparison to simpler machine learning models such as random forest when applied to low dimensional problems, as such these simpler models have reported higher prediction accuracies. Traditionally, ROP is measured in real-time by instruments which carryout Measurement-While-Drilling (MWD). Higher ROP values means that faster distance is being achieved in drilling, and it is the desire of oil drilling companies to achieve deeper depths in shorter time intervals in order to save time and money, hence the need for the optimization of the rate of penetration. Several variables influence the rate ROP, some of them are WOB and RPM which can be controlled on the surface. The other variables (PHIF, VSH, SW, and KLOGH) are a function of the soil formation. Unfortunately, ROP does not always increase proportionally with increasing or decreasing values of these variables, as observation has shown that it increases initially until a point called the founder point or the sweet spot (optimum point) from where it begins to wane, The values of the outside variables must be increased from this point in order to maintain the optimum performance.
3 Proposed Methodology This work adopts the standard procedure for developing machine learning and data science models. The methodology as shown in Fig. 1 comprises of five (5) phases, these are: (1) Understanding the Business problem (2) Data cleaning and Preparation (3) Data modelling (4) Model Evaluation (5) Model Operationalization 3.1 Maintaining the Integrity of the Specifications The template is used to format your paper and style the text. All margins, column widths, line spaces, and text fonts are prescribed; please do not alter them. You may note peculiarities. For example, the head margin in this template measures proportionately more than is customary. This measurement and others are deliberate, using specifications that anticipate your paper as one part of the entire proceedings, and not as an independent document. Please do not revise any of the current designations.
156
I. D. Acheme et al.
Fig. 1. Data Science Methodology (Nwankwo, 2020)
3.2 Understanding the Problem and Data Collection The dataset utilized for this research is the open source volve data. This is a complete set of data from a North Sea oil field available for research, study and development purposes arranged from two original data; real-time drilling data and a Computed Petrophysical Output (CPO) log data from well number 15/9-F-15 in the Volve Oil Field in the North Sea [32] This dataset comprises of seven (7) input variables and one (1) target variable. These are. i. Depth (measured depth) ii. WOB (weight on bit) iii. SURF_RPM (rotation per minute at the surface) iv. PHIF (formation porosity) v. VSH (shale volume) vi. SW (water saturation) vii. KLOGH (log permeability) viii. ROP_AVG (rate of penetration average TARGET VARIABLE). The dataset contained a total of one hundred and fifty (150) records with eight (8) features. Table 1. Snapshot of the Dataset Depth WOB
SURF_RPM ROP_AVG PHIF
VSH
SW
KLOGH
0
3305
26217.864 1.314720
0.004088
0.086711 0.071719 1.000000 0.001000
1
3310
83492.293 1.328674
0.005159
0.095208 0.116548 1.000000 0.001000
(continued)
Petroleum Drilling Monitoring and Optimization: Ranking the Rate of Penetration
157
Table 1. (continued) Depth WOB
SURF_RPM ROP_AVG PHIF
VSH
SW
KLOGH
2
3315
97087.882 1.420116
0.005971
0.061636 0.104283 1.000000 0.001000
3
3320
54793.206 1.593931
0.005419
0.043498 0.110040 1.000000 0.001000
4
3325
50301.579 1.653262
0.005435
0.035252 0.120808 1.000000 0.001000
…
…
…
…
146 4065
71081.752 2.104258
…
…
0.008808
0.087738 0.291586 1.000000 0.162925
…
…
…
147 4070
72756.626 2.333038
0.008824
0.019424 0.503175 1.000000 −0.001124
148 4075
83526.789 2.333326
0.008799
0.054683 0.689640 1.000098 0.002261
149 4080
84496.549 2.334673
0.008375
0.022857 0.640100 1.000000 0.001000
150 4085
86658.559 2.331339
0.008454
0.022857 0.640100 1.000000 0.001000
3.3 Methodology In order to achieve the objectives of this research, the selected machine learning algorithms were built using the dataset described in Table 1. The steps of the entire process are shown in Fig. 2.
Fig.2. Proposed Architecture
158
I. D. Acheme et al.
3.4 Data Modelling and Evaluation Exploratory data analysis was then performed to obtain additional knowledge from the data and uncover hidden patterns. With the selected features, the data is then divided in the ratio of 70:30 for training and testing purposes. This is then passed on to machine learning algorithms. Table one shows the machine learning algorithms used and their performance comparison. In order to estimate outliers in the dataset, the cooks distance outlier detection was used, (Fig. 3). The Cook’s Distance is an estimate of a data point’s influence. It considers both the leverage and residual of each observation. Cook’s Distance is a calculation that summarizes how much a regression model changes when the ith observation is removed.
Fig. 3. Cook’s distance outlier detection
The models shown in Table 2. were implemented in order to compare the prediction accuracy of the target variable, ROP_AVG using the selected features. The metrics for comparison are the established metrics for evaluating regression models such as MAE, MSE, RMSE, R2 etc. From our results, the random forest regressor model outperformed all the others and it is the ranked number 1, while the passive aggressive regressor showed the least prediction accuracy and ranks number 18. Table 2. Selected Regression Models and their Performances Model
MAE MSE
RMSE R2
RMSLE MAPE TT (Sec)
rf
Random Forest Regressor
0.00 0.00 0.00 06 00 10
−0.089 0.00 1 10
0.1082 0.407
gbr
Gradient Boosting 0.00 0.00 0.00 Regressor 06 00 10
−0.107 0.00 6 10
0.1133 0.045 (continued)
Petroleum Drilling Monitoring and Optimization: Ranking the Rate of Penetration
159
Table 2. (continued) Model
MAE MSE
RMSE R2
RMSLE MAPE TT (Sec)
et
Extra Trees Regressor
0.00 0.00 0.00 06 00 09
−0.180 0.00 5 09
0.1064 0.363
huber
Huber Regressor
0.00 0.00 0.00 07 00 10
−0.185 0.00 6 10
0.1213 0.029
dt
Decision Tree Regressor
0.00 0.00 0.00 07 00 12
−0.361 0.00 7 12
0.1298 0.014
knn
K Neighbors Regressor
0.00 0.00 0.00 09 00 13
−0.536 0.00 4 12
0.1572 0.059
ridge
Ridge Regression
0.00 0.00 0.00 07 00 10
−0.551 0.00 6 10
0.1219 0.012
br
Bayesian Ridge
0.00 0.00 0.00 07 00 10
−0.553 0.00 0 10
0.1224 0.014
en
Elastic Net
0.00 0.00 0.00 09 00 12
−0.572 0.00 9 12
0.1531 0.013
lightg bm Light Gradient 0.00 0.00 0.00 Boosting Machine 07 00 11
−0.577 0.00 1 11
0.1281 0.046
lr
Linear Regression
0.00 0.00 0.00 07 00 10
−0.586 0.00 1 10
0.1227 0.304
lar
Least Angle Regression
0.00 0.00 0.00 07 00 10
−0.586 0.00 1 10
0.1227 0.013
lasso
Lasso Regression
0.00 0.00 0.00 09 00 12
−0.604 0.00 0 12
0.1532 0.014
llar
Lasso Least Angle 0.00 0.00 0.00 Regression 09 00 12
−0.692 0.00 9 12
0.1531 0.014
dum my
Dummy Regressor
0.00 0.00 0.00 09 00 12
−0.692 0.00 9 12
0.1531 0.013
omp
Orthogonal Matching Pursuit
0.00 0.00 0.00 07 00 11
−0.719 0.00 7 11
0.1262 0.012
ada
AdaBoost Regressor
0.00 0.00 0.00 07 00 11
−0.784 0.00 5 11
0.1267 0.067
par
Passive Aggressive Regressor
0.00 0.00 0.00 79 01 80
−138.4 0.00 456 80
1.0000 0.013
3.5 Evaluation of the Random Forest Regressor Having established the random forest (rf) algorithm as the most efficient among the chosen eighteen (18) machine learning algorithms tested with the dataset, further evaluation analysis of the algorithm was carried out, this includes the residual plot, error plot, learning and validation curves. Furthermore, feature importance ranking was also
160
I. D. Acheme et al.
carried out to rank the most important features for predicting ROP. These results are shown in Fig. 4, 5 and 6
Fig. 4. Residuals for the random forest algorithm
The residual plot was used to confirm the model’s fit with assumptions of constant variance, normality and independence of errors. The plot reveals the difference between observation and fitted values. The plot (Fig. 4) shows randomly scattered points which are maintain an approximately constant width about the line of identity, this is indicative of a good model as it is close to a null residual plot.
Fig. 5. Prediction error for the random forest algorithm
The model’s performance was further investigated with the learning and validation curves (Fig. 6). These plots show a model’s performance as the set of training data increases or over a defined period of time. They are useful for models built with datasets that are incremental. The training curves shows how well the model learns, while the validation curve shows how well the model with generalize with yet to be seen values. 3.6 Feature Importance and Ranking The decrease in node impurity weighted by the probability of reaching that node is used to calculate feature importance. The node probability can be calculated by the number of
Petroleum Drilling Monitoring and Optimization: Ranking the Rate of Penetration
161
(a)
(b) Fig. 6. Learning and validation curve for the random forest regressor
samples that reach the node, divided by the total number of samples. The more important features have higher values (see Fig. 7).
Fig. 7. Feature importance
162
I. D. Acheme et al.
KEY: Depth : Measured depth WOB: Weight on bit SURF_RPM: Rotation per minute at the surface PHIF: Formation porosity VSH: Shale volume SW: Water saturation KLOGH: Log permeability. As shown in Fig. 7, the features in increasing order of importance are; measured depth, rotation per minute on the surface, shale volume, weight on bit, formation porosity, water saturation and log permeability.
4 Conclusion The development and deployment of effective machine learning application for ROP prediction promises better results than other traditional techniques that have been reported. This is so because of more availability datasets generated at oil rigs, however, a clear challenge comes to bear, which is selecting the right machine learning algorithms and features. A model built with many of the lowly influential factors or with a less efficient algorithm is not likely to give satisfactory results. Hence in this research work, we investigated eighteen (18) machine learning algorithms by building these models from the Volve drilling dataset of the North Sea in order to comparatively study their performance and rank them. The result of this work provides a template for the choice of algorithms as well as selection of features for implementing ML solutions for oil drilling optimization. This can be helpful in hybridization and development of the real-time ROP prediction models. Future research will emphasize the development of innovative embedded systems that incorporate use specific optimized machine learning models for controlling drilling operations. These ML models will be designed to take into account various drilling conditions and factors such as geological formations, well depth, and drilling speed, among others, to improve drilling efficiency and reduce downtime. Additionally, the use of embedded systems in oil drilling operations will provide real-time monitoring and analysis of drilling parameters, enabling early detection of potential problems and minimizing the risk of equipment failure. Ultimately, this collaboration has the potential to revolutionize the oil drilling industry in Nigeria and significantly enhance its productivity and profitability.
References 1. Cao, J., Gao, J., Jiao, T., Xiang, R., Pang, Y., Li, T.: Feature investigation on the ROP machine learning model using realtime drilling data. J. Phys. Conf. Ser. 2024(1), (2021). IOP Publishing 2. Sircar, A., Yadav, K., Rayavarapu, K., Bist, N., Oza, H.: Application of machine learning and artificial intelligence in oil and gas industry. Petroleum Res. (2021)
Petroleum Drilling Monitoring and Optimization: Ranking the Rate of Penetration
163
3. Darwesh, A.K., Rasmussen, T.M., Al-Ansari, N.: Controllable drilling parameter optimization for roller cone and polycrystalline diamond bits. J. Petroleum Explor. Prod. Technol. 10(4), 1657–1674 (2019). https://doi.org/10.1007/s13202-019-00823-1 4. Ameloko, A.A., Uhegbu, G.C., Bolujo, E.: Evaluation of seismic and petrophysical parameters for hydrocarbon prospecting of G- field, Niger Delta, Nigeria. J. Petroleum Explor. Prod. Technol. 9(4), 2531–2542 (2019) 5. Takbiri-Borujeni, A., Fathi, E., Sun, T., Rahmani, R., Khazaeli, M.: Drilling performance monitoring and optimization: a data- driven approach. J. Petroleum Explor. Prod. Technol. 9(4), 2747–2756 (2019) 6. Braga, D.C.: Field Drilling Data Cleaning and Preparation for Data Analytics Applications. Louisiana State University and Agricultural & Mechanical College (2019) 7. Olayinka, A.S., Adetunji, C.O., Nwankwo, W., Olugbemi, O.T., Olayinka, T.C.: A study on the application of bayesian learning and decision trees IoT-enabled system in postharvest storage. In: Artificial Intelligence-based Internet of Things Systems, pp. 467–491. Springer, Cham (2022) 8. Nwankwo, W., Adetunji, C.O., Olayinka, A.S.: IoT-driven bayesian learning: a case study of reducing road accidents of commercial vehicles on highways. In: Pal, S., De, D., Buyya, R. (eds.) Artificial Intelligence-based Internet of Things Systems. Internet of Things. Springer, Cham (2022). https://doi.org/10.1007/978-3-030-87059-1_15 9. Adetunji, C.O., et al.: Machine learning and behaviour modification for COVID-19. Med. Biotechnol. Biopharmaceutics, Forensic Sci. Bioinform. 271–87 10. Adetunji, C.O., et al.: The role of an intelligent feedback control system in the standardization of bio-fermented food products. Fermentation and Algal Biotechnologies for the Food, Beverage and Other Bioproduct Industries 143–62 11. Nwankwo, W., Adetunji, C.O., Ukhurebor, K.E., Makinde, A.S.: Artificial intelligence-aided bioengineering of eco-friendly microbes for food production: policy and security issues in a developing society. In: Agricultural Biotechnology, pp. 301–313. CRC Press, Dec 21 2022 12. Nwankwo, W., Nwankwo, C.P., Wilfred, A.: Leveraging on artificial intelligence to accelerate sustainable bioeconomy. IUP J. Knowledge Manag. 20(2), 1 (2022) 13. Chinedu, P.U., Nwankwo, W., Masajuwa, F.U., Imoisi, S.: Cybercrime detection and prevention efforts in the last decade: an overview of the possibilities of machine learning models. Rev. Int. Geographical Educ. 11(7), 1 (2021) 14. Olayinka, T.C., Olayinka, A.S., Nwankwo, W.: Evolving feed-forward artificial neural networks using binary and denary dataset. SAU Sci. Tech J. 6(1), 96–108 (2021) 15. Nwankwo, W., et al.: The adoption of AI and IoT technologies: socio-psychological implications in the production environment. IUP J. Knowl. Manag. 19(1), 1 (2021) 16. Acheme, I.D., Makinde, A.S., Udinn, O., Nwankwo, W.: An intelligent agent-based stock market decision support system using fuzzy logic. IUP J. Inf. Technol. 16(4), 1 (2020) 17. Nwankwo, W., Umezuruike, C., Njoku, C.C.: Enhancing learning systems using interactive intelligent components. Int. J. 9(3) (2020) 18. Olayinka, A.S., Nwankwo, W., Olayinka, T.C.: Model based machine learning approach to predict thermoelectric figure of merit. Archive of Science and Technology. 1(1) (2020) 19. Nwankwo, W., Ukhurebor, K.E.: Web forum and social media: a model for automatic removal of fake media using multilayered neural networks. Int. J. Sci. Technol. Res. 9(1), 4371–4377 (2020) 20. Al-Khudiri, M.M., et al.: Application Suite for 24/7 Real Time Operation Centers. InSPE Saudi Arabia Section Annual Technical Symposium and Exhibition 2015 Apr 21. OnePetro 21. Acheme, I.D., Vincent, O.R., Olayiwola, O.M.: Data science models for short-term forecast of COVID-19 spread in Nigeria. In: Hassan, S.A., Mohamed, A.W., Alnowibet, K.A. (eds.) Decision Sciences for COVID-19. International Series in Operations Research & Management Science, vol. 320. Springer, Cham (2022). https://doi.org/10.1007/978-3-030-87019-5_20
164
I. D. Acheme et al.
22. Alsaihati, A., Elkatatny, S., Gamal, H.: Rate of penetration prediction while drilling vertical complex lithology using an ensemble learning model. J. Petrol. Sci. Eng. 1(208), 109335 (2022) 23. Hegde, C., Soares, C., Gray, K.: Rate of penetration (ROP) modeling using hybrid models: deterministic and machine learning. InUnconventional Resources Technology Conference, Houston, Texas, 23-25 July 2018 2018 Sep 28 (pp. 3220-3238). Society of Exploration Geophysicists, American Association of Petroleum Geologists, Society of Petroleum Engineers 24. Eckel, J.R.: Microbit studies of the effect of fluid properties and hydraulics on drilling rate. J. Petrol. Technol. 19(04), 541–546 (1967) 25. Bourgoyne, A.T., Young, F.S.: A multiple regression approach to optimal drilling and abnormal pressure detection. Soc. Petrol. Eng. J. 14(04), 371–384 (1974) 26. Jahanbakhshi, R., Keshavarzi, R., Jafarnezhad, A.: Real-time prediction of rate of penetration during drilling operation in oil and gas wells. In: 46th US Rock Mechanics/Geomechanics Symposium 2012 Jun 24. OnePetro 27. Ashrafi, S.B., Anemangely, M., Sabah, M., Ameri, M.J.: Application of hybrid artificial neural networks for predicting rate of penetration (ROP): a case study from Marun oil field. J. Petrol. Sci. Eng. 1(175), 604–623 (2019) 28. Wang, Y., Salehi, S.: Application of real-time field data to optimize drilling hydraulics using neural network approach. J. Energy Resour. Technol. 137(6), 1 (2015) 29. Hinton, G.E., Srivastava, N., Krizhevsky, A., Sutskever, I., Salakhutdinov, R.R.: Improving neural networks by preventing co- adaptation of feature detectors. arXiv preprint arXiv:1207. 0580. 3 Jul 2012 30. Hegde C, Wallace S, Gray K. Using trees, bagging, and random forests to predict rate of penetration during drilling. InSPE Middle East Intelligent Oil and Gas Conference and Exhibition 2015 Sep 15. OnePetro 31. Equinor. Volve data village dataset: released under a license based on CC BY 4.0 32. Makinde, A.S., Agbeyangi, A.O., Nwankwo, W.: Predicting mobile portability across telecommunication networks using the integrated-KLR. Int. J. Intell. Inf. Technol. (IJIIT) 17(3), 50–62 (2021). https://doi.org/10.4018/IJIIT.2021070104 33. Rajasekar, M., Geetha, A.: Comparison of machine learning algorithms in domain specific information extraction. Int. J. Math. Sci. Comput. (IJMSC) 9(1), 13–22 (2023). https://doi. org/10.5815/ijmsc.2023.01.02 34. Joseph, I., Imoize, A.L., Ojo, S., Risi, I.: Optimal call failure rates modelling with joint support vector machine and discrete wavelet transform. Int. J. Image Graph. Signal Process. (IJIGSP), 14(4), 46–57 (2022). https://doi.org/10.5815/ijigsp.2022.04.04 35. Abd El-Latif, E.I., Khalifa, N.E.: A model based on deep learning for COVID-19 X-rays classification. Int. J. Image graph. Signal Process. (IJIGSP), 15(1), 36–46 (2023). https://doi. org/10.5815/ijigsp.2023.01.04
Application of Support Vector Machine to Lassa Fever Diagnosis Wilson Nwankwo5(B) , Wilfred Adigwe2 , Chinecherem Umezuruike3 , Ijegwa D. Acheme1 , Chukwuemeka Pascal Nwankwo1 , Emmanuel Ojei4 , and Duke Oghorodi2 1 Department of Computer Science, Edo State University, Uzairue, Nigeria 2 Department of Computer Science, Delta State University of Science and Technology,
Ozoro, Nigeria 3 College of Computing and Communication Studies, Bowen University, Iwo, Nigeria 4 Department of Software Engineering, Delta State University of Science and Technology,
Ozoro, Nigeria 5 Department of Cyber Security, Delta State University of Science and Technology,
Ozoro, Nigeria [email protected]
Abstract. Lassa fever is a type of viral hemorrhagic fever with high fatality rate. Precise and prompt diagnosis and successful treatment of this disease is very important in the control and prevention of the spread as unrestricted spread can easily erupt into an epidemic. An insightful decision support system can help make quick and more exact findings on this disease. In this paper, an intelligent model is proposed for predicting Lassa fever is proposed. The Lassa fever dataset was extracted from the clinical records of patients from a known Specialist Teaching Hospital in Southern Nigeria. The relevant feature variables were identified and encoded in a manner appropriate for use in algorithmic analysis. The support vector classifier was used to develop a characterization model for the disease. Assessment and evaluation were based on exactness, accuracy, and affectability. Findings showed that the proposed system could improve and encourage quick decision-making as it utilizes real historical data to effectively characterize Lassa fever spread as well as promote the timely treatment and management of patients. Keywords: Diagnostics · Lassa fever · Machine learning · Support vector machine · Epidemic
1 Introduction The healthcare delivery sector has continued to suffer setbacks in providing quality healthcare services, especially in developing countries [1–7]. Quality connotes fitness for purpose hence, quality healthcare entailed correct and accurate diagnosis and the subsequent prompt treatment of diagnosed cases. Quality healthcare is largely a function of right and accurate clinical decision-making. Poor clinical decisions can lead to disaster and loss of lives in the healthcare space. Medical diagnosis is a subjective task; © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 165–177, 2023. https://doi.org/10.1007/978-3-031-36115-9_16
166
W. Nwankwo et al.
that is, it depends on the physician making the diagnosis [8]. Furthermore, the amount of data (signs, symptoms, laboratory tests, etc.) to be analyzed before arriving at the correct diagnosis may be enormous and, at times, unmanageable. Machine learning (ML) and deep learning have proved reliable and efficient in most socioeconomic spheres, especially in the provision of quick and high-precision prediction rules from historical data [9–19]. ML could help healthcare specialists make the diagnostic process more objective and more reliable [20]. Lassa fever (LF) is an acute and viral hemorrhagic disease with a cycle of between two and twenty-one days. The disease was first discovered in the North-East of Nigeria in 1969 and the virus (an RNA Virus) was named the lassa virus after a town where it was discovered in Nigeria. The LF virus has been isolated in other West African countries such as Sierra Leone, Liberia, Mali, Cote d’Ivoire, and Guinea. Similarly, the virus has also been isolated in some countries in eastern Africa. As a disease with high mortality rates, effective and efficient approaches need be explored in controlling the spread of the disease [21]. Usually, human infections is through direct contact with items such as food, water, etc. contaminated with the urine or feces of the multimammate rat, a rodent from Mastomys family [22, 23]. LF may be transmitted from human to human which may often result to epidemics with case fatality rates as high as 60%. LF may be difficult to diagnose in its early stages in hospitals [21]. The common symptoms include pharyngitis, cough, and gastrointestinal symptoms [22]. It has been noted that the early symptoms of the disease are not specific but include fever, sore throat, malaise, myalgia, cough, nausea, headache, chest pain, abdominal pain, diarrhea, and vomiting. Late signs of infection or disease are facial oedema, bleeding, effusions and convulsion, coma, pleural, and pericardial. With these known symptoms, it is possible to develop and deploy a system that is capable of predicting the spread of the disease so that Government agencies, medical experts, and individuals can take precautionary measures in the fight of the disease. A successful prediction would help healthcare practitioners and other stakeholders to manage emergencies such as outbreaks of the disease. Current researches focus on intelligent insights drawn through spatiotemporal infectious diseases outbreak modeling and prediction [24, 25]. This study aims to develop an intelligent diagnostic model for Lassa fever disease. The specific objectives are to: (1) design a machine learning model that can serve as a decision support system for health experts in the diagnosis of Lassa fever; (2) implement the design in (i) for the diagnosis of Lassa fever using the python programming language.
2 Related Works Owing to the fatality of the Lassa fever disease, research is ongoing in the area of developing sophisticated tools and technologies such as machine learning applications and mathematical models for managing Lassa fever outbreaks. However, due to the limited localization of the disease, not much studies have been documented especially in the domain of machine learning applications and algorithmic modeling. Some of the related works in the domain of disease management using intelligent and smart approaches are briefly discussed herein. Madueme and Chirove [22] developed a mathematical model for the study of LF epidemiology. Their model comprises a system of nonlinear ordinary
Application of Support Vector Machine to Lassa Fever Diagnosis
167
differential equations (ODE) which were aimed at determining the effect of the various LF transmission pathways as well as the progression of the infection in humans and rodents. Their model analysis and numerical simulations showed that the rate of LF infection increases if there are different transmission routes. Bakere et al. [26] developed a mathematical model which they used to analyse the transmission of Lassa fever. They employed a periodically forced seasonal nonautonomous system of a nonlinear ordinary differential equation to capture the dynamics of Lassa fever transmission as well as the seasonal variation in the birth of Mastomys rodents which are the main vectors of the Lassa fever virus. The researchers showed that their model could offer epidemiological insights into the control of Lassa fever spread. Ossai et al. [27]. Examined the preventive measures against Lassa fever among heads of family units in Abakaliki city, Southeast Nigeria using an expressive cross-sectional study. A four-phase examining configuration was utilized to choose 420 respondents from Abakaliki city. They concluded that though most of their respondents exhibit a good understanding of preventive measures against Lassa fever, there is a need to involve more health workers in community awareness programmes to educate the population on appropriate measures. Alabdulkarim et al. [28] proposed a privacy-policy single decision tree model for clinical decision support which could help healthcare officers to protect patients’ data from privacy breaches. The system is secure and the patients’ data are protected through encryption using homomorphic encryption code in such a way that only authorized organizations utilizing the Internet of Things could access the patients’ information. Oonsivilai et al. [29] applied ML algorithms to guide the empirical antibiotic prescription process in children. They deduced that the intelligent model could provide exceptionally instructive predictions on anti-microbial susceptibilities using patients’ data thereby aiding in controlling experimental anti-toxin therapy. Osaseri and Osaseri [30]. Proposed an Adaptive Neuro-Fuzzy Inference System (ANFIS) for use in Lassa Fever prediction among patients. They adopted four input parameters: White Blood Count, Temperature on Admission, Abdominal Pain, and Proteinuria respectively. They appropriated their LF dataset into a 61: 59 ratio for training and testing of their model respectively. Roosan et al. [31] examined the multifaceted nature of clinical thinking in infections utilizing subjective topical examination, that is a meeting of coauthors to autonomously note pertinent ideas joined with multifaceted nature, psychological objectives, versatile systems, and sense-production. Shoaib et al. [32] discussed a predictive computational network that is aimed at predicting the transmission of Lassa fever in Nigeria. They designed a mathematical model comprising an artificial neural networks-based hybrid algorithm of Genetics Optimization and Sequential Quadratic Programming to predict the dynamics of Lassa fever. The model emphasized the transmission patterns of Lassa fever.
168
W. Nwankwo et al.
3 Methodology The Lassa fever clinical dataset was obtained from the epidemiological unit of a specialist hospital designated by the Nigerian Government for the management of and research on Lassa fever infections. The dataset contains a thousand seven hundred and twenty-eight records. Following data cleaning, the dataset was split into 80: 20 ratios for training, testing, and evaluation purposes. Figure 1 shows the methodology adopted for this study.
Fig. 1. Methodology for developing the prediction model
3.1 Data Collection The dataset collected was unstructured with some columns not required for training and testing; such columns were the patients’ name ID and home address. As part of that data preparation process, these columns were removed from the dataset. Table 1 presents the useful features selected for the machine learning model. From the raw dataset, essential features relevant to the prediction/diagnosis of Lassa fever disease were identified (Table 1).
Application of Support Vector Machine to Lassa Fever Diagnosis
169
Table 1. Description of selected variables S/N
Condition
1
Lassa Fever (Target Variable: Yes or No)
2
Date/Time
3
Fever (Body Temperature)
4
Body Pain
5
Headache
6
Vomiting
7
Jaundice
8
Bleeding
9
Age
10
Gender
11
Underlying health condition (Diabetes/Hepatitis/High blood pressure)
12
Body Mass Index
13
Alcohol or cigarette intake
3.2 Data Preprocessing During this stage, the raw data was transformed into a machine-usable format since the raw data contained incomplete inconsistent values and outliers. The dataset was also labeled and encoded appropriately. Hence the resulting dataset was more reliable and fit for knowledge discovery. Furthermore, in a randomized manner, the entire dataset was split into a train and test set. 3.3 Input Variable Transformation and Target Variable The symptoms manifested by patients are the input variables. These are transformed into two classes (yes or no) and three classes depending on the nature of the manifestation, while the target/predicted variable is a diagnosis of suspected Lassa fever which is a binary classification). Table 2 presents the transformed dataset for the input variables. The data transformation and encoding, as shown in Table 2, are essential for the dataset used for training machine learning models. This is because machine learning algorithms require that input and output variables are numeric values. As a result, categorical data such as is in this study must be encoded to numbers before we can use it to train and test a model.
170
W. Nwankwo et al. Table 2. Input Variables Transformation
S/N
Variable definition
Domain 1
Domain 2
Domain 3
1
Lassa Fever (Target Variable: Yes or No)
Yes = 1
No = 0
NA
2
Date/Time
NA
NA
NA
3
Fever (Body Temperature)
High = 1
Medium = 0.5
Low = 0
4
Body Pain
Yes = 1
No = 0
NA
5
Headache
High = 1
Moderate = 0.5
Low = 0
6
Vomiting
Yes = 1
No = 0
NA
7
Jaundice
Yes = 1
No = 0
NA
8
Haemorrhage
Yes = 1
No = 0
NA
9
Age
NA
NA
NA
10
Gender
NA
NA
NA
11
Underlying Health condition (Diabetes/Hepatitis/High blood pressure)
Yes = 1
No = 0
NA
12
Body Mass Index
NA
NA
NA
13
Alcohol or cigarette intake
Yes = 1
No = 0
NA
3.4 Use Case Diagram of the System The use case diagram is used for representing a user’s interaction with the system, in this case, medical personnel since this system is built to aid in the diagnosis of Lassa fever disease. The use case diagram helps in understanding and identifying the operation and users’ role in the entire system as well as showing the relationship between the user and designated functions. Figure 2 is the use case diagram of the proposed method.
Fig. 2. Use case Diagram
Application of Support Vector Machine to Lassa Fever Diagnosis
171
3.5 The Support Vector Machine Algorithm The Support Vector Machine (SVM) belongs to the family of supervised machine learning algorithms used for regression and classification problems [33–35]. The SVM algorithm plots each data item as a point in n-dimensional space with the value of each feature represented by the value of a coordinate and n refers to the number of features in the dataset. Thus, classification is performed by finding the hyper-plane that most distinguishes the two classes. The study utilized the Lassa fever patients’ dataset in the form: {(xk , yk )}nk=1
(1)
where x is the features, and y is a 1x n vector; n is the training dataset. Given that training variable x and output variable y are defined by Eq. (2) and (3). {x1 , x2 , x3 . . . } ∈ RN
(2)
{y, y2 , y3 . . . } ∈ R
(3)
Equation (1) represents the training variable while Eq. (2) is the output variable. The SVM algorithm is the function y(x) that minimizes the error for all the learning features xi , hence the objective is to: minimize :
1 ¨ l ||w||2 + c (αl + α) l=1 2
(4)
Subject to: w.xi + b-y ≤ e + αi , yi -w.xi -b < e + αi . Where w is an n-dimensional vector representing the weight i = 1, …l; c > 0 determines the trade-off between the differences in the decision function and α, α≥1. 3.6 Proposed Evaluation The results of the SVM model are evaluated by the prediction accuracy, represented by a confusion matrix. Other evaluation metrics used include AUC-ROC, F1 Score, precision, and recall though these are more suitable for binary classification problems, unlike a multiclass problem discussed in the present study. 3.7 Exploratory Analysis The dataset comprises over one thousand seven hundred records of patients admitted for treatment of Lassa fever. Figure 3–4 show the summary and description of the dataset in relation to the member variables: ‘Lasssa_Fever_Case’, and ‘Not_Lasssa_Fever_Case’ respectively.
172
W. Nwankwo et al.
Fig. 3. Description of dataset
Fig. 4. Pie Chart showing the percentage distribution of Lassa case
3.8 Evaluation Metrics To establish the efficiency of this model, well-known and generally acceptable evaluation metrics are used. First, the prediction accuracy and confusion matrix are considered, as shown in Fig. 5. The “Area Under the Curve (AUC)” of the “Receiver Operating Characteristics(ROC)” i.e. AUC-ROC curve that evaluates the number of true positives against false positives is then plotted. This model is also evaluated with the precision and recall plots. Figure 6 shows the AUC-ROC and Precision/Recall plots. Confusion Matrix It is a productivity metric for the problem of classification of machine learning where two or more classes can be tested. It is a table comprising four different expected and actual value combinations. Figure 5 and Table 3 show the confusion matrix of the support vector machine of the test data.
Application of Support Vector Machine to Lassa Fever Diagnosis
173
Fig. 5. Prediction Accuracy and Confusion Matrix
Table 3. Confusion Matrix computation N = 519
Predicted: 0
Predicted: 1
Actual: 0
TN = 294
FP = 69
363
Actual: 1
FN = 72
TP = 84
98
366
153
The rates computed from the confusion matrix for the classifier model are presented in Table 4. The plots of AUC-ROC, precision, and recall are shown in Fig. 6. Table 4. Rates Computed from the confusion matrix Rates
Values
Accuracy
0.73
Misclassification Rate
0.27
True Positive Rate
0.86
False Positive Rate
0.19
True Negative Rate
0.81
Precision
0.55
Prevalence
0.19
Recall
0.80
174
W. Nwankwo et al.
Fig. 6. AUC-ROC; Precision and Recall
One of the major advantages of SVM is relatively high accuracy and speed (efficient use of processing resources) compared to many fundamental machine learning algorithms. This constitutes the bedrock for the choice of the SVM. However, we are aware that different SVM kernels may perform differently on same dataset. In the above computation, the linear kernel was used. From Table 4, the accuracy of predictions on cases of Lassa fever is placed at 73% however other metrics such as the true positive and true negative predictions are placed at 86% and 81% respectively. The model in its present state predicts 80% of the true Lassa fever cases (recall = 0.80). The recall is considered very important to us considering the fact that we are more interested in identifying and managing a Lassa fever case. This shows that the model exhibits the potential for improved accuracy and precision which might be attained on carrying out model tuning by way of kernel modification, regularization, modifying the gamma value(increasing it). It is also instructive to note that though model tuning is ideal, it does not necessarily guarantee much significant performance improvements considering the small dataset. Furthermore, in Fig. 6, we got 0.80 as the AUC, and we considered this a pretty good score in the sense that the model could differentiate those patients with Lassa fever and those patients without Lassa fever 80% of the time. This percentage/score may be improved by applying different hyperparameter values.
Application of Support Vector Machine to Lassa Fever Diagnosis
175
4 Conclusion Clinical decision support and intelligent systems have a history of enhancing the efficiency of valuable socio-economic workflows. The application of AI-based systems in medical diagnosis and disease management cannot be overemphasized. In this paper, we set out to examine the applicability of support vector machine algorithms in the detection, diagnosis, and profiling of Lassa fever in patients using historical clinical data from a specialist hospital. The developed prediction model shows that valuable insights could be drawn from historical data and used to drive new diagnoses. The proposed system improves and encourages decision-making in this regard and would be a valuable tool in the hands of medics who are saddled with the task of providing care for the helpless during Lassa fever outbreaks. The findings in this study notwithstanding, we recognize the limitation in the sense that the dataset available was not detailed enough and also fall short of satisfying the adequacy metric in terms of the volume of data available. However, we believe that this study would constitute a good cornerstone for conducting future research in the construction of hybrid models that could provide better results for Lassa disease diagnosis and management.
References 1. Victor-Ikoh, M.I., Moko, A., Nwankwo, W.: Towards the Implementation of a Versatile Mobile Health Solutions for the Management of Immunization Against Infectious Diseases in Nigeria. In: Salvendy, G., Wei, J. (eds.) Design, Operation and Evaluation of Mobile Communications. HCII 2022. Lecture Notes in Computer Science. Springer, Cham (2022) 2. Umezuruike, C., Nwankwo, W., Tibenderana, P., John, P.A., Muhirwa, R.: Corona Virus Disease (COVID 19): analysis and design of an alert and real-time tracking system. Int. J. Emerg. Trends Eng. Res. 8(5), 1743–1748 (2020) 3. Umezuruike, C., Nwankwo, W., Okolie, S.O., Adebayo, A.O., Jonah, J.V., Ngugi, H.: Health informatics system for screening arboviral infections in adults. Int. J. Inf. Technol. Comput. Sci. (IJITCS), 11(3), 10–22 (2019) 4. Nwankwo, W., Umezuruike, C.: An object-based analysis of an informatics model for Zika virus detection in adults. Comput. Biol. Bioinform. 6(1), 1–20 (2018) 5. Nwankwo, W.: Harnessing e-healthcare technologies for equitable healthcare delivery in Nigeria: the way forward. Int. J. Sci. Res. 6(3), 1875–1880 (2017) 6. Umezurike, C., Nwankwo, W., Kareyo, M.: Implementation challenges of health management information systems in Uganda: a review. J. Multidisciplinary Eng. Sci. Technol. 4(7), 7726– 7731 (2017) 7. Umezurike, C., Nwankwo, W., Okolie, S.O., Adebayo, A.: Developing an informatics model for effective healthcare in military health facilities in Nigeria. World J. Eng. Res. Technol. 3(4), 69–99 (2017) 8. Das, R., Turkoglu, I., Sengur, A.: Effective diagnosis of heart disease through neural networks ensembles. Expert Syst. Appl. 36(4), 7675–7680 (2009) 9. Nwankwo, W., Chinedu, U.P., Aliu, D., et al.: Integrated FinTech solutions in learning environments in the post-COVID-19 era. IUP J. Knowl. Manag. 20(3), 1–22 (2022) 10. Nwankwo, W., Nwankwo, C.P., Wilfred, A.: Leveraging on artificial intelligence to accelerate sustainable bioeconomy. IUP J. Knowl. Manag. 20(2), 38–59 (2022)
176
W. Nwankwo et al.
11. Acheme, D.I., Makinde, A.S., Osemengbe, U., Nwankwo, W.: An intelligent agent-based stock market decision support system using fuzzy logic. IUP J. Inf. Technol. 16(4), 1–20 (2020) 12. Nwankwo, W., Adetunji, C.O., Olayinka, A.S.: IoT-driven bayesian learning: a case study of reducing road accidents of commercial vehicles on highways. In: Pal, S., De, D., Buyya, R. (eds.) Artificial Intelligence-based Internet of Things Systems. Internet of Things. Cham: Springer, pp. 391–418 (2022) 13. Chinedu, P.U., Nwankwo, W., Masajuwa, F.U., Imoisi, S.: Cybercrime detection and prevention efforts in the last decade: an overview of the possibilities of machine learning models. Rev. Int. Geographical Educ. (RIGEO) 11(7), 956–974 (2021) 14. Nwankwo, W., Adetunji, C.O., Olayinka, A.S., et al.: The Adoption of AI and IoT technologies: socio-psychological implications in the production environment. IUP J. Knowl. Manag. 19(1), 50–75 (2021) 15. Olayinka, A.S., Adetunji, C.O., Nwankwo, W., et al.: A study on the application of bayesian learning and decision trees IoT-enabled system in postharvest storage. In: Pal, S., De, D., Buyya, R. (eds.) Artificial Intelligence-based Internet of Things Systems. Internet of Things. Cham: Springer, 467–491 (2022) 16. Osikemekha, A.A., Adetunji, C.O., Olaniyan, T.O., Hefft, D.I., Nwankwo, W., Olayinka, A.S.: IoT-based monitoring system for freshwater fish farming: analysis and design. In: Abraham, A., Dash, S., Rodrigues, J.J.P.C., Acharya, B., Pani, S.K. (eds.) Intelligent Data-Centric Systems: AI, Edge and IoT-based Smart Agriculture, pp. 505–515. Academic Press, Amsterdam (2022) 17. Adetunji, C.O., Osikemekha, A.A., Olaniyan, T.O.: Toward the design of an intelligent system for enhancing salt water shrimp production using fuzzy logic. In: Abraham, A., Dash, S., Rodrigues, J.J.P.C., Acharya, B., Pani, S.K. (eds.) Intelligent Data-Centric Systems: AI, Edge and IoT-based Smart Agriculture, pp. 533–541. Academic Press, Amsterdam (2022) 18. Nwankwo, W., Ukhurebor, K.E.: Big data analytics: a single window IoT-enabled climate variability system for all-year-round vegetable cultivation. IOP Conference Series: Earth and Environmental Science, 655, 012030 (2021) 19. Nwankwo, W., Ukhurebor, K.E., Ukaoha, K.C.: Knowledge discovery and analytics in process re-engineering: a study of port clearance processes. In: International Conference in Mathematics, Computer Engineering and Computer Science (ICMCECS). Lagos: IEEE Explore, pp. 1–7 20. Adetunji, C.O., Nwankwo, W., Olayinka, A.S., Olaniyan, O.T., et al.: Machine learning and behaviour modification for COVID-19. In: Inuwa, H.M., Ezeonu, I.M., Adetunji, C.O., Ekundayo, E.O., Gidado, A., Ibrahim, A.B., Ubi, B.E. (eds.) Medical Biotechnology, Biopharmaceutics, Forensic Science and Bioinformatics. Florida: CRC Press, pp. 271–87 (2020) 21. Asogun, D.A., Günther, S., Akpede, G.O., Ihekweazu, C., Zumla, A.: Lassa fever: epidemiology, clinical features, diagnosis, management and prevention. Infect. Dis. Clin. North Am. 33(4), 933–951 (2019) 22. Madueme, P.U., Chirove, F.: Understanding the transmission pathways of Lassa fever: a mathematical modeling approach. Infectious Dis. Model. 8(1), 27–57 (2023) 23. Musa, S.S., Zhao, S., Abdullahi, Z.U.: COVID-19 and Lassa fever in Nigeria: a deadly alliance? Int. J. of Infectious Diseases 117, 45–47 (2022) 24. Li, Y.: Genetic basis underlying Lassa fever endemics in the Mano River region, West Africa. Virology 579, 128–136 (2023) 25. Okokhere, P., Colubri, A., Azubike, C., et al.: Clinical and laboratory predictors of Lassa fever outcome in a dedicated treatment facility in Nigeria: a retrospective, observational cohort study. Lancet. Infect. Dis 18(6), 684–695 (2018)
Application of Support Vector Machine to Lassa Fever Diagnosis
177
26. Bakare, E.A., Are, E.B., Abolarin, O.E., et al.: Mathematical modelling and analysis of transmission dynamics of Lassa Fever. J. Appl. Math. 2020, 1–18 (2020) 27. Ossai, E.N., Onwe, O.E., Okeagu, N.P., et al.: Knowledge and preventive practices against Lassa fever among heads of households in Abakaliki metropolis, Southeast Nigeria: a crosssectional study. Proc. Singapore Healthc. 29(2), 73–80 (2020) 28. Alabdulkarim, A., Al-Rodhaan, M., Ma, T., Tian, Y.: PPSDT: a novel privacy-preserving single decision tree algorithm for clinical decision-support systems using IoT devices. Sensors 19(1), 142 (2019) 29. Oonsivilai, M., et al.: Using machine learning to guide targeted and locally-tailored empiric antibiotic prescribing in a children’s hospital in Cambodia. Wellcome Open Res. 3(131), 1–18 (2018) 30. Osaseri, R.O., Osaseri, E.I.: Soft computing approach for diagnosis of Lassa fever. Int. J. Comput. Inf. Eng. 3(11) (2016) 31. Islam, R., Weir, C.R., Jones, M., Del Fiol, G., Samore, M.H.: Understanding complex clinical reasoning in infectious diseases for improving clinical decision support design. BMC Med. Inform. Decis Making 15(101), 2–12 (2015) 32. Shoaib, M., Tabassum, R., Raja, M.A.Z., Nisar, K.S., Alqahtani, M.S., Abbas, M.: A design of predictive computational network for transmission model of Lassa fever in Nigeria. Results Phys. 39 (2022) 33. Rajasekar, M., Geetha, A.: Comparison of machine learning algorithms in domain specific information extraction. Int. J. Math. Sci. Comput. (IJMSC) 9(1), 13–22 (2023) 34. Joseph, I., Imoize, A.L., Ojo, S., Risi, I.: Optimal call failure rates modelling with joint support vector machine and discrete wavelet transform. Int. J. Image Graph. Signal Process. (IJIGSP), 14(4), 46–57 (2022) 35. Abd El-Latif, E.I., Khalifa, N.E.: A model based on deep learning for COVID-19 x-rays classification. Int. J. Image Graph. Signal Process. (IJIGSP), 15(1), 36–46 (2023)
A Novel Approach to Bat Protection IoT-Based Ultrasound System of Smart Farming Md. Hafizur Rahman1 , S. M. Noman2 , Imrus Salehin3(B) and Tajim Md. Niamat Ullah Akhund4
,
1 Department of Digital Anti-Aging Healthcare, Inje University, Gimhae, Republic of Korea 2 Faculty of Computer Science and Engineering, Frankfurt University of Applied Sciences,
Frankfurt, Germany 3 Department of Computer Engineering, Dongseo University, Busan, Republic of Korea
[email protected] 4 Department of Computer Science and Engineering, Daffodil International University, Dhaka,
Bangladesh
Abstract. In past decades, scientific research has explored various bat detection systems methods, a common trend to identify heterogeneous species of bats. In this article, we represented an ultrasonic bat detection and protection method by interfacing an IoT automated device for the first time in agriculture development to introduce steady security in a preventive way. The automated bat banisher system is an effectual model for farmers to protect their land at a very affordable cost from nuisance bats. Moreover, the highest-quality fruits are the ones that bats choose to eat when they attack. This prevents the farmer from getting the best possible yield, Farmers have tried everything from handcrafted sound systems and medication to nets and traps to get rid of the bats, but they have yet to be effective. We created an Internet of Things-based automated bat banisher to combat bats invading farms. In this research, our High-Frequency Ultrasound (IoT automated) System has the potential to save farmers money, time, and effort. Farmers may increase agricultural output to provide the highest suitable fruit supply. Therefore, an ultrasonic IoT-based bat Banisher device is the most straightforward and functional means of dealing with nuisance bats. Keywords: Ultrasonic Waves · High-Frequency Ultrasounds · IoT · Smart Farming · Agriculture Development · Bat Banisher
1 Introduction The Bat discharges ultrasonic waves with high frequencies. Its calls are pitched at 20– 200 Kilohertz (kHz), usually absurdly sharp for people to hear [1]. Their sounds are reflected in the earth, hitting different articles, and returning to the Bat as echoes. The resounding sign empowers the Bat to lay out a psychological guide to its condition. Sound is prompted through a medium as the particles are vibrated, making weight waves with the locale of weight and rarefaction [2]. The waves have trademark highlights of © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 178–186, 2023. https://doi.org/10.1007/978-3-031-36115-9_17
A Novel Approach to Bat Protection IoT-Based Ultrasound System of Smart Farming
179
wavelength, rehash, and ample. Wavelength (λ) separates two zones of maximal weight (or rarefaction). The immensity of wavelength is that the attack of the ultrasound wave is as for wavelength, and picture targets are close to 1–2 wavelengths [3]. Rehash (f) is the number of wavelengths that appreciate relief. It is reliably evaluated as cycles (or wavelengths); the unit is hertz (Hz). It is a particular portion of the critical stone utilized in the ultrasound transducer. It may be moved by the administrator inside the set reasons for constraint – the higher the rehash, the better the targets, yet the lower the entrance. The level of weight change is given by the ample. It is passed on in decibels on a logarithmic scale [21]. Farmers can employ ultrasonic bat banisher devices to regulate bat communities because they emit sound waves that interfere with echolocation, making the region inhospitable to bats and creating a friendly environment for the farming system. The ultrasonic sound waves produced by bats are protected barriers that do not harm people [22]. Farmers have suffered from their traditional method of protecting their crops for a few decades, but they may significantly improve by adopting a bat banisher device. The bat colony will look for a new home in unremarkable structures or areas. This paper presents the technique and structures for ousting bats from agriculture. In one putting, the procedure that may accomplish on a system fabricated perceiving from an ultrasound structure for Bat removing to guarantee agriculture field. Much more research has to be done on the pervasive use of ultrasonic sound as a monitoring technique. Many bats, cetaceans, certain rodents, insectivores, birds, and insects are among the numerous animals that create ultrasound [4]. O’Donnell et al. described their papers BATBOX III bat detector worked in two different types of bats, short-tailed and long-tailed bats, with detection frequency levels 27 kHz and 40 kHz [5]. In contrast, our bat banisher devices are more effective in the 20 kHz to 60 kHz range in recognizing ultrasound frequencies that accelerate from various bat categories. This article suggests an IoT-based methodology to guard against possible harm to agriculture from wild animals and weather. Using IoT (Internet of Things) agricultural applications, the economy could improve crop quality, save expenses, and raise operational efficiency [6]. Various methods are used in bat detection systems of ultrasound conversion techniques, such as heterodyning, frequency division, and time expansion [19]. Through ultrasonic recording technology, analysis software, and conservation, the UK’s iBat program has devised a system for monitoring ultrasound biodiversity in three bat species. According to their research, agricultural land contains 42.4% of Pipistrellus species and 21.7% of Myotis species [20]. This paper addresses a prototype ultrsound IoT device system for banishing bats and a procedure for growing the agricultural rate that leads to our country’s economic growth. • To reduce time, cost, and human resources • To increase agricultural production • To mitigate many types of virus attacks, including Nipah virus (NIV) from bats.
2 Related Work Bats pose a significant risk, harm to fruit growth and infecting viruses. The technology is advertised as simple, affordable, and autonomous. A watertight container houses a Batbox III bat detector, a tiny voice-activated tape recorder, a talking clock, and an
180
Md. H. Rahman et al.
optional long-life battery are a bat detecting methods [5]. Oisin Mac Aodha et al. identified echolocating bats’ ultrasonic, full-spectrum sounds during the search phase and created an open-source pipeline based on convolutional neural networks [7]. Ultrasonic signal transmission via a wireless connection in the commercial radio frequency range is the subject of this paper’s early investigation (UHF). Herman et al. used in the building of massive wireless sensor networks for the study of bat behaviour [8]. Tomas et al. described in their paper superheterodyne QMC Mini Bat Detector inaccuracies action due to an adequate calibration system. The echolocation call features of bats cannot be used to identify them as a result [9]. Using bat detectors to record bat behavior, such as flight patterns and hunting strategy, the researchers provided examples of their research [17]. According to the study by Griffin et al., heterodyne bat detectors have had trouble detecting some species of bats since the 19th century because of their poor performance at 12 kHz, even though bat noises have a frequency range of up to 120 kHz. They implemented a bat detector at frequencies between 35 kHz and 45 kHz audible [18]. S. Noman et al. designed a prototype IoT device to detect insects from farmer’s crop fields. In their study, they simulated agriculture automation systems to dispose of conventional farming through the blessing of modern technology [16]. Besides, bats employ an echo they make to forage for food at night. The primary diet of bats consists of insects and a variety of fruits, including mango, litchi, guava, plum, banana, white Jamun, Ziziphus mauritiana, tamarind, olive, and betel nuts. IoT-based bat detection and protection device is a unique method to redress conventional man-made systems to enlarge fruit production in farming areas. According to S. Sofana Reka et al., who utilized an IoT-Based Smart Greenhouse Farming System, food production may be boosted by using even dry and infertile terrain, improving output. Higher quality yields are produced by eliminating wastage and keeping greater control over crops’ operations and environment [11]. Sensor technology and wireless IOT network integration have been researched and examined in the context of the current state of smart agricultural farming systems using the internet and cellular communications. The Remote Monitoring System (RMS) is suggested [12]. I. Salehin et al. proposed an intelligent device, interfacing between Arduino UNO and ultrasonic sensors to get an overload water alert in the drainage system. Their motive was to reduce the time and cost of innovating IoT to enable a model for a smart city [10]. Ryu et al. demonstrate an IoT-based linked farm that seeks to provide consumers with smart agricultural solutions. The benefits of linked farms over earlier smart farms are described with service scenarios and a complete design and execution for connected farms [13]. The proposed Farm as a Service (FaaS) integrated system offers high-level application services by managing connected equipment, data, and models and running and monitoring farms. This system analyses environmental and growth data as well as registers, connects and controls Internet of Things (IoT) devices [14]. An in-depth analysis of many techniques, including drip irrigation, greenhouse cultivation, IoT-based monitoring systems, wireless networks, smart agriculture, and precision farming, is addressed in this research [15].
A Novel Approach to Bat Protection IoT-Based Ultrasound System of Smart Farming
181
3 Methodology of Circuit Implementation We used an Arduino UNO R3, a speaker 8, a Voice Recorder Module ISD1820, connection cables, an IR sensor, a motion sensor PIR, red and green LED lights, a breadboard, and a 5-V Lithium-ion battery, power source to build the system. The thickness and compressibility of the medium affect the propagation velocity (v) of the sound waves. The link between these variables is transmitted via the Wave Eq. 1. The bat behavior expressed in terms of frequency of "f" is caused by wavelength λ (because λf = v, v is the velocity of sound waves in the air) in the Bat algorithm. The magnitude of the wavelength denoted the shape of the aliment that bats are hunting to their focusing target [21]. v = λf
(1)
3.1 Model Selection Diagram Figure 1 illustrates the activity completed from beginning to end. At that time, we need first to tap the switch start. The PIR sensor is quickly approached. Whether the IR sensor receives signals from PIR sensor to differentiate an object, the recording of the article’s voice transfers to the ultrasound sensor, the ultrasound sensor expands the recorded sound by voice recorder module between 20 kHz up to 60 kHz, and the ultrasound sensor moves to play, after which the recording stops, and the process ends. An algorithm for global streamlining is called the Bat calculation. It was propelled by the echolocation of microbats, whose heartbeats fluctuated and produced noise and a lot of noise. [22].
Fig. 1. Flow chart diagram of system activities.
The activation process shows in Fig. 2, after activated the device first emits an IR signal and, if an ultrasonic echo is detected, PIR sensor sends signal in Arduino device and receives an IR signal from the opposite side. If echoes are found, it will work red light in system and record then play back to transfer at the ending of the framer’s field.
182
Md. H. Rahman et al.
Start process
PIR send
IR Receive (if detected)
If Eco (Record)
Play
Go to 2
End
Fig. 2. Block diagram of system activities.
3.2 System Circuit Implementation We integrated the ultrasonic sound speaker terminals connected to the SP + and SP-pins of the ISD1820 Voice Recorder Module. The module’s VCC and GND are linked to its +5V and GND. Arduino’s Computerized IO Pins 2 and 3 are connected to the REC and PLAYE pins. Here, two advanced infrared sensors are used, PIR and IR sensors. Its driven output is connected to Arduino Sticks 0 and 4. In this experiment, we used a 5-V Lithium-ion battery connected to the battery connection port. We create the connection as the circuit schematic specifies, then turn on the circuit. The PIR Sensor output is LOW, and the Arduino is idle when there is no object in front of it. The IR Sensor’s output rapidly rises when there is an object in front of it, and Arduino then begins recording a message by setting the REC Pin HIGH for around 5 s. The LED connected to Pin 13 is now lit RED to show that the module is currently encoding a message. Following the strategy, the message is replayed by raising the play stick for around five seconds. The LED connected to Pin 12 is now lit GREEN after exterminating the detected object. Figure 3 shows the circuit diagram of bat detection and protection system.
Fig. 3. Circuit diagram of bat detection and protection system.
3.3 Architecture Model Implementation A particular form of a sound wave is sent into the air as a bat moves. The sound is reflected off buildings, trees, mountains, and other large obstructions and returns to the
A Novel Approach to Bat Protection IoT-Based Ultrasound System of Smart Farming
183
bat’s ears. The bat’s brain knows the distance between the obstacle and the reflected sound. When they join our agricultural fields, the bat will make high-frequency sounds ranging from 40 to 200 kHz [2]. The Voice Recorder Module in our system will capture the high-frequency sound made by bats. The IR Sensor and Motion Sensor detect the sound when bats come within range of the device up to 110 m. The Voice Recorder Module starts recording the sound once the sensor sends a signal to the Red LED. The Red LED will then flash in a millisecond, and the speaker will begin producing the same sound frequency. Bats will assume this passage is dangerous if they hear the same echo frequency. So, they’ll stay away from this route at this time, and LED Green will blink when an object is finished off. Our agricultural land will be secure at the same time. The following Fig. 4 shows architecture model of Bat detection and protection system in real image.
Fig. 4. Architecture model of Bat detection and protection system in real image.
4 Experimental Result and Discussion In our experiment, the PIR sensor’s maximum range outside is from 10 to 150 m. The PIR sensor will signal the voice recorder module when the bats come into the range. The red light will be ON, meaning the sound is recorded with the help of an IR sensor. Then within a few seconds green lights is ON when the tweeter speaker (high-frequency speaker) emits the sound into the environment and blows away the object. After bat banishes, our agricultural fruit will be safe, and we will get the proper output from future utilization. Olive, litchi, Ziziphus Mauritiana, and plum trees are used in our investigation of farming land. After conducting fieldwork, we discovered that out of 68 bat detection alerts, 54 bats flew alerts using red and green lights (on/off), whereas device accuracy was an average of 79.41%. The sensing number of bats depended on various time durations in different locations in tree fields. As we see, Litchi tree has the highest number of detection alerts producing 84% accuracy. In the morning, to verify the field status, we haven’t seen damaged fruits in the ground as a before, comparatively past time.
184
Md. H. Rahman et al.
We set up the device in a 100-m farmers field for the experiment and observed the bat’s activity by seeing the green and red lights on and off. Table 1 shows the experimental results of four variant trees on farming land. Table 1. Experimental results of IoT Bat Banisher Tree Name
Time (hour)
RED Light on/off (time)
GREEN Light on/off (time)
IoT Device Accuracy (%)
Olive
9pm–10pm
12
9
75
Litchi
10pm–11pm
25
21
84
Ziziphus Mauritiana
11pm–12am
13
10
76.92
Plum
12am–1am
18
14
77.77
Total
4h
68
54
79.41
We had a pleasant, successful outcome after deploying our system in a field of agricultural trees. For better accuracy, more than one device above 50 m of land space is suggested depending on the farmer’s land space. Due to the different genetics of bat species and producing high frequency, some bat was unaffected by the bat banisher. A man can hear the range of frequencies between 20 Hz and 20 kHz [22]. As a result, people won’t be harmed by this system of ultrasonic sound. Additionally, this technology does not affect the biosphere, as we saw when we employed it in a real-world setting.
5 Conclusion Innovative tools are essential to the success of today’s agricultural methods. As a result, our Bat Banisher may be a useful tool for farmers. From this experiment, we might infer that protecting agricultural land is the farmer’s responsibility and that doing so is essential to successfully implementing the novel approach. Bat detection and protection via the Internet of Things is a versatile and valuable addition. The bats are driven away by the ultrasonic barrier created by the continuous transmission of ultrasonic sound waves from the ultrasonic bat banisher. The Internet of Things Bat Banisher is the most recent method developed to remove bats from agricultural locations. Adopting this strategy will boost the agricultural sector’s gross domestic product (GDP), allowing us to raise the country’s total GDP each year. This proposed device would also be beneficial in eliminating bat-borne viruses like Covid-19 and Nipah. Acknowledgments. The Daffodil International University Innovation Lab and Dongseo University Machine Learning /Deep Learning lab assisted in the development of the suggested framework.
A Novel Approach to Bat Protection IoT-Based Ultrasound System of Smart Farming
185
References 1. Brigham, R., Elizabeth, K., Gareth, J., Stuart, P., Herman, L.: Bat Echolocation Research. Tools, Techniques, and Analysis (2004) 2. Dekker, J., Steen, W., Bouman, H.B., van der Vliet, R.E.: Differences in acoustic detectibility of bat species hamper environmental impact assessment studies. Eur. J. Wildl. Res. 68(2), 1–8 (2022). https://doi.org/10.1007/s10344-022-01562-1 3. Fenton, M.B., Bouchard, S., Vonhof, M.J., Zigouris, J.: Time-expansion and zero-crossing period meter systems present significantly different views of echolocation calls of bats. J. Mammal. 82(3), 721–727 (2001) 4. Pettorelli, N., Baillie, J.E., Durant, S.M.: Indicator bats program: a system for the global acoustic monitoring of bats. Biodiversity monitoring and conservation: bridging the gap between global commitment and local action, 211–247 (2013) 5. O’Donnell, C.F.J., Sedgeley, J.: An automatic monitoring system for recording bat activity, 1–16 (1994) 6. Haque, M.A., Haque, S., Sonal, D., Kumar, K., Shakeb, E.:. Security enhancement for IoT enabled agriculture. Materials Today: Proceedings (2021) 7. Mac Aodha, O., et al.: Bat detective—Deep learning tools for bat acoustic signal detection. PLoS Comput. Biol. 14(3), e1005995 (2018) 8. Herman, K., Debski, M., Furmankiewicz, J.: A preliminary analysis of wireless link for bats signals acquisition system. Acta Phys. Pol. A 120(4), 635–638 (2011) 9. Thomas, D.W., West, S.D.: On the use of ultrasonic detectors for bat species identification and the calibration of QMC Mini Bat Detectors. Can. J. Zool. 62(12), 2677–2679 (1984) 10. Salehin, I., Baki-Ul-Islam, S.M., Noman, M.M., Hasan, S.T., Hasan, M.: A smart polluted water overload drainage detection and alert system: based on IoT. In: 2021 International Mobile, Intelligent, and Ubiquitous Computing Conference (MIUCC), pp. 37–41 (2021). https://doi.org/10.1109/MIUCC52538.2021.9447664 11. Reka, S.S., Chezian, B.K., Chandra, S.S.: A novel approach of IoT-based smart greenhouse farming system. In: Green Buildings and Sustainable Engineering, pp. 227–235. Springer, Singapore (2019) 12. Patil, K.A., Kale, N.R.: A model for smart agriculture using IoT. In: 2016 International Conference on Global Trends in Signal Processing, Information Computing and Communication (ICGTSPICC). IEEE (2016) 13. Ryu, M., Yun, J., Miao, T., Il-Yeup, A., Choi, S.H., Kim, J.: Design and implementation of a connected farm for smart farming system. In: 2015 IEEE SENSORS, pp. 1–4. IEEE (2015) 14. Kim, S., Lee, M., Shin, C.: IoT-based strawberry disease prediction system for smart farming. Sensors 18(11), 4051 (2018) 15. Prasad, G., Vanathi, A., Devi, B.S.: A Review on IoT Applications in Smart Agriculture (2023). https://doi.org/10.3233/ATDE221332 16. Salehin, I., et al.: IFSG: intelligence agriculture crop-pest detection system using IoT automation system. Indonesian J. Electr. Eng. Comput. Sci. 24(2), 1091 (2021). https://doi.org/10. 11591/ijeecs.v24.i2.pp1091-1099 17. Limpens, H.J.G.A.: Field identification: using bat detectors to identify species. Bat Echolocation Res. Tools, Tech. Anal. 46 (2004) 18. Griffin, D.R.: The past and future history of bat detectors. Bat echolocation research: tools, techniques, and analysis. Bat Conservation International, Austin, Texas, USA, 6–9 (2004) 19. Pettersson, L.: The properties of sound and bat detectors. Bat Echolocation Research: tools, techniques and analysis, 9 (2004) 20. Linton, D.M.: Bat ecology & conservation in lowland farmland. Diss. University of Oxford (2009)
186
Md. H. Rahman et al.
21. Yong, J., He, F., Li, H., Zhou, W.: A novel bat algorithm based on collaborative and dynamic learning of opposite population. In: 2018 IEEE 22nd International Conference on Computer Supported Cooperative Work in Design ((CSCWD)), pp. 541–546. IEEE (2018) 22. Britzke, E.R., Gillam, E.H., Murray, K.L.: Current state of understanding of ultrasonic detectors for the study of bat ecology. Acta Theriologica 58, 109–117 (2013)
Synthesis and Modeling of Systems with Combined Fuzzy P + I - Regulators Bohdan Durnyak1 , Mikola Lutskiv1 , Petro Shepita1 , Vasyl Sheketa2 , Nadiia Pasieka2(B) , and Mykola Pasieka2 1 Ukrainian Academy of Printing, Pid Goloskom Str., 19, Lviv 79020, Ukraine 2 National Technical University of Oil and Gas, 76068 Ivano-Frankivsk, Ukraine
[email protected]
Abstract. A mathematical model of the automatic control system with a combined fuzzy P + I - controller with an unclouded I - component, which allows eliminating the static error of regulation of static objects, has been developed. The method of determining the parameters of the controller adjustment based on the identification of the object by the transient characteristic, under the condition when the transfer coefficient of the object is equal to one by conducting an extended tangent through the point of intersection determined the parameters of adjustment of the controller. The blurring of the P - component of control has been carried out based on three linguistic variables of the normalized error signal. The fuzzy control described by the fuzzy base of rules has been given, three triangular membership functions have been chosen, and their parameters have been adjusted. MatLab Simulink package on the basis of which the structural scheme of the system was developed was used for modelling and research of fuzzy automatic control systems. The combined fuzzy P + I - regulator consists of two main blocks: the algorithm of normalized control, fusion units, logic output and denormalization, on the basis of which the regulated action on the object has been formed. For convenience of adjustment and research of separate blocks and system visualization blocks have been provided. An example of synthesis and modelling of a combined fuzzy system for inertial objects of the third and fourth order has been considered. The settings for adjusting the controller have been defined and adjusted. The results of simulation modelling in the form of transient characteristics for objects of different dimensions that satisfy the specified parameters of control quality and provide better quality indicators than with a traditional controller have been given. The maximum value of the regulatory action is small, so it is not physically limited, which is an advantage. Keywords: Synthesis · Modelling · Fuzzification · Fuzzy controller · Parameters · Inference · Accuracy
1 Problem Setting Growing demands on the quality of manufactured products place new demands on systems of automatic control of technological processes and objects in the presence of various influences. The main disadvantage of traditional standard industrial regulators is © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 187–196, 2023. https://doi.org/10.1007/978-3-031-36115-9_18
188
B. Durnyak et al.
that they do not provide quality control when changing the parameters of the object and limiting the regulatory effect on the object. Instead, fuzzy regulators control processes and objects much better with incomplete information about the object, changes in its parameters and the action of various influences. Today there is a development of various algorithms and fuzzy automatic control systems in various industries [1–5, 7, 8, 10]. In the available sources, little attention is paid to the synthesis of systems with fuzzy controllers and the formation of fuzzy control on an object that is fundamentally different from the classical theory of automatic control. Traditional control algorithms are described by an analytical expression and have two or three debug parameters. Instead, fuzzy regulators are described by a base of fuzzy rules, linguistic variables, a series of fuzzy transformations (error normalization, fuzzification, defuzzification, fuzzy inference, which are different and have different implementation schemes), and so on. Therefore, the solution of the synthesis problem and the definition of the parameters of the fuzzy controller is ambiguous and complex. To date, there are no generally accepted sound methods for the synthesis and determination of the parameters of fuzzy regulators, which makes it impossible to optimize them and complicates their development and implementation. Thus, the task of synthesis and modelling of automatic control systems of the combined fuzzy controller is relevant.
2 Analysis of Recent Research and Publications The demand for fuzzy systems is because they are developed faster, simpler, and cheaper than traditional controllers [6, 9, 11–14]. Works [15–17] are devoted to the synthesis and calculation of fuzzy digital controllers in automatic control systems. The methods of designing fuzzy controllers are based on the analytical expression for the regulatory action at the output of the fuzzy controller for different membership functions and the general functional and structural schemes, based on which it is possible to implement fuzzy program controllers by software and hardware. Schematic diagrams of fuzzy controllers with different membership functions are presented. The study of automatic control systems was conducted by modeling in the MatLab: Simulink package. For the implementation of the given task, numerous complex schemes were used, which additionally contain a scheme for estimating the maximum value of error and its scaling, and the I-component that introduces inertia into the system is not used, but the first and second derivatives of the error signal complicating the system are considered. The monographs [17, 19–24] present the tasks of designing fuzzy systems in the package MatLab: Simulink using Fuzzy Logic Toolbox. Information on fuzzy sets, fuzzy logic and the application of fuzzy rules and transformations in different systems has been given. Books [18, 25, 26] present different versions of fuzzy controllers, rule bases, block diagrams of different regulators, their analysis, and debugging choices. In the works of the authors [27, 28, 31] the analysis of automatic control systems with simplified versions of fuzzy controllers has been carried out, and the simulation results have been presented. It has been established that the quality of the transient process changes a little when the object transfer coefficient is doubled. Significant fluctuations occur in a system with a traditional regulator, instead [29, 30]. Purpose of the article: To synthesize an automatic control system with a simple version of the combined P + I -
Synthesis and Modeling of Systems with Combined Fuzzy P + I - Regulators
189
controller for static objects, to determine the debugging parameters, and to investigate its properties.
3 Presentation of the Main Research Material The problem of synthesis of automatic control systems with the fuzzy controller is complex and multifaceted, and contains a significant number of partial problems and possible ways to solve them, so we have chosen a simple algorithm for fuzzy combined P-controller with I-component that does not blur and eliminates static control error in static control objects. We set the basic requirements for the quality of regulation: to ensure sufficient system speed and 20% overshoot when changing the transfer rate of the object. It is necessary to determine the debugging parameters of the controller, which provide quality indicator tasks for a given object of regulation. Debugging parameters will be determined based on the identification of the object by the transient characteristic, provided that the transfer factor of the object is equal to one and its parameters will be determined on the basis of tangent to the transition characteristic through the point of intersection, which is shown in Fig. 1. The simplest substitute model of the object of regulation is the inertial link with delay, which is widely used in determining the parameters of traditional regulators [6, 7]. W0 (S) =
k0 e−τ s , To S + 1
(1)
where T0- is the substitution constant of the object time, τ - is the delay time, k0 - is the transmission coefficient of the object.
Fig. 1. Transient characteristics of the object
To determine the debugging parameters of the P + I – fuzzy controller, we additionally perform a tangent to the intersection with the axis and determine the parameters a, and τ as shown in Fig. 1. Based on the identification of the object by the transient characteristic in Fig. 1 for the fuzzy P + I - regulator, the following values of the debugging parameters were obtained: the scale P of the component Mp = 0,17/a Mr = and I component Ki = 0.125/τ . The presented methods of determining the parameters of adjustment of the controller are approximate and therefore require additional experimental tuning or simulation.
190
B. Durnyak et al.
For the convenience of presentation, we give the algorithm of normalized control of P + I – fuzzy controller in operator form. U = F(e) +
E ki e, if e = , S y0
(2)
where E = y0 -y is the control error, k i is the transmission factor I of the algorithm component, y0 , y is the output and the set point of the controlled value, F(e) is the fuzzy transformations of the normalized e-error signal. To synthesize the fuzzy P-component of regulation, we accept three linguistic variables of the normalized error signal e: negative B, zero H, positive D, which correspond to three functions of belonging to fuzzy sets of left type - right and triangular symmetric, then control is described by fuzzy rule base. R1 : if (e = B)then(U = B1) R2 : if (e = H )then(U = B2), R3 : if (e = D)then(U = B3)
(3)
where Bi are blurred controls. Each rule corresponds to the initial sets Bi, i = 1, 2, 3. To model and study the fuzzy automatic control system, use the package MatLab: Simulink, which developed a scheme of combined P + I-fuzzy controller, given in Fig. 2.
Fig. 2. Block diagram of the model of automatic control system with fuzzy controller
Fuzzy P + I– regulator consists of two main units: the normalized control algorithm developed by expression (2) for competence is located in the Subsystem unit and the fuzzification, logical output and denormalization units located in the Enabled Subsystem unit. The inertial static object of the third dimension is located on the right. At the bottom there is a copy of the fuzzy control system of the fourth-dimension object, necessary for comparing the results of modeling and analysis.
Synthesis and Modeling of Systems with Combined Fuzzy P + I - Regulators
191
The scheme of the algorithm of normalized control has been developed on the basis of the expression (2) shown in Fig. 3. To determine the normalized signal of e – error, its scaling is carried out by dividing the adjustment error E by a given value of the input task y0, which ensures its rationing regardless of the value of the task, which may be different, which simplifies rationing. The normalized signal is sent to the input of the next unit for fuzzification and fuzzy transformations. The general scheme of the unit of fuzzification of logical output and denormalization is shown in Fig. 4. The scheme consists of three triangular functions of accessories such as Triangular MF, the following parameters B [-2000; -1; 1], H [-1; 0; 1], D [0; 1; 2000] have been adjusted. Conclusion is carried out by optimizing the function of belonging to the Up level, which is set by a normalized e-error signal, and is limited to the Saturation Dynamic unit.
Fig. 3. Scheme of a normalized control algorithm
Fig. 4. Scheme of a logical output fuzzification unit
The functions of belonging to B*, H*, D* fuzzy sets are modified in this way at the input of the MAX operator at the input of which the normalized fuzzy control Un is obtained, which is fed to the input of the Gain unit, in which the Mp debugging parameter
192
B. Durnyak et al.
of the P-component of the regulator is set, the output of which is fed to the first input of the Produkt unit, and a displacement of U0 = 1/k0 y0 is submitted to its second input as a result of this, denormalization (scaling) of the U control is carried out. To create an adjusting action V on the object on the ground of the addition unit, three components are supplied: denormal U control, U0 displacement and I-component of Ui control. V = U + U0 + Ui , if U0 =
1 y0 then U = Mp Un . k0
For the convenience of debugging individual units of the regulator and the automatic control system, scope and display visualization units are provided. An example of synthesis and modeling of a combined fuzzy system with a P + I-regulator for inertial static objects of the third and fourth order with a transmission coefficient k 0 = 10 and time levels T 1 = 5c, T 2 = 3c, T 3 = 2c, T 4 = 1c. is considered. The 20% walking, and the y0 = 100 input task has been set. Based on the identification of the object by transitional characteristic (Fig. 1), the regulator debugging parameter x + rs for objects of the third M p = 1,9, k i = 0,60 and the fourth order M p = 1,0, k i = 0,25 have been determined and set. More complete parameters of the fuzzy automatic control system are presented directly on the corresponding diagrams. The results of simulation modeling of transition processes in a combined fuzzy control system for objects of different orders are shown in Fig. 5.
Fig. 5. Transition processes in systems
Transition processes have a specified 20%. Walking the growth time in the system with the object of the third order t n = 6c, the adjustment time is t p = 14c, and t n = 10c, t p = 26c with the objects of the fourth order. Transition processes quickly disappear, and there is no static error. At the moment of time, t = 30c, a degree perturbation z0 = 20 is presented on an object that, with a slight reregulation, goes away. Consequently, the combined fuzzy automatic control system with the P + I regulator provides the specified indicators of the quality of regulation. The results of the study of the regulatory action quality on the object at the step task y0 = 100 are presented in Fig. 6.
Synthesis and Modeling of Systems with Combined Fuzzy P + I - Regulators
193
Fig. 6. Regulatory action on the object
When the jumping task y0 = 100, the combined fuzzy regulator forms a regulatory effect that differs significantly from the traditional one. At the initial moment of time, a short-term amplitude pulse of which is 30 is created and gradually approaches the established value. Consequently, the maximum value of the adjustable action is small, so it is not physically limited, which is an advantage. Studies confirm that automatic control systems with a combined P + I fuzzy regulator with inertial static objects provide better quality indicators of regulation than with traditional PI regulator. In particular, with small transfer coefficients of the object (k 0 ≈ 1), the maximum value of the adjustable action increases significantly (V H ≥ 250), which can be physically limited, which impairs achievable quality indicators. It is established that systems with a traditional PI regulator are quite sensitive to changes in the transfer coefficient of the object. With a twofold increase in the coefficient in the system, there is a significant re-regulation and oscillation. Instead, with a twofold decrease in the transmission coefficient, the transition process becomes periodic and significantly decreases the performance of the system. To eliminate the impact of changes in the transfer coefficient of the object, it is necessary to re-determine and set the parameters for debugging the PI regulator, which is a disadvantage. Instead, systems with a combined fuzzy regulator are little sensitive to changes in the parameters of the object, which is their advantage.
4 Conclusions Nowadays, there are no generally accepted reasonable methods of synthesis and determination of parameters for setting fuzzy regulators, which makes it impossible to optimize them, complicates their development and implementation. The problem of synthesis of the automatic control system with fuzzy regulators is complex and multifaceted, it contains a significant number of partial problems and possible ways to solve them. A fuzzy combined P-regulator with the I-component is proposed, which does not heat up and eliminates the static error of regulation of static regulatory objects, which simplifies the regulator’s system. It is proposed to determine the parameters of setting the regulator on the basis of identification of objects by transitional characteristics, provided that the transfer coefficient of the object is equal to one that ensures the choice of parameters of the regulator, regardless of the transfer factor of the object. A structural diagram of the
194
B. Durnyak et al.
fuzzy system model in the MatLab: Simulink package has been developed, which makes it possible to calculate and build transitional characteristics of the system to analyze its properties and interactively test the optimal adjustment parameters of the regulator. The results of simulation modeling in the form of transitional characteristics of the system for objects of different order are presented and the quality indicators of regulation are determined. It is established that automatic control systems with combined fuzzy P + I -regulator with inertial objects provide better indicators of quality of regulation than with traditional regulators. With small transfer coefficients of the object (k 0 ≈ 1), the maximum value of the adjusting action increases significantly (V H ≥ 250), which can be physically limited, which impairs achievable quality indicators. It has been established that systems with a traditional PI regulator are quite sensitive to changes in the transfer coefficient of the object. With a twofold increase in the coefficient of the system, significant reregulation and oscillation occurs, and with a decrease in the coefficient, the performance decreases significantly. To eliminate the effect of changing the coefficient, it is necessary to re-determine and set the adjustment parameters of the regulator, which is a disadvantage. Instead, systems with a combined fuzzy regulator are little sensitive to changes in object parameters, which is their advantage.
References 1. Durnyak, B., Lutskiv, M., Shepita, P.: Fuzzy model of raster transformation of square elements. Paper presented at the CEUR Workshop Proceedings, 3156, pp. 140–149 (2022) 2. Durnyak, B., Lutskiv, M., Shepita, P., Nechepurenko, V.: Simulation of a Combined Robust System with a P-Fuzzy Controller. Intellectual Systems of Decision Making and Problems of Computational Intelligence: Proceedings of the XV International Scientific Conference. 1020, 570–580 (2019) 3. Imamovi´c, B., Halilˇcevi´c, S.S., Georgilakis, P.S.: Comprehensive fuzzy logic coefficient of performance of absorption cooling system. Expert Syst. Appl. 190, art. no. 116185 (2022). https://doi.org/10.1016/j.eswa.2021.116185 4. Pasieka, N., Sheketa, V., Romanyshyn, Y., Pasieka, M., Domska, U., Struk, A.: Models, methods and algorithms of web system architecture optimization. In: IEEE International Scientific-Practical Conference: Problems of Infocommunications Science and Technology, PIC S&T 2019 - Proceedings, pp. 147–152. https://doi.org/10.1109/PICST47496.2019.906 1539 5. Salem, A., Desouky, A., Alaboudy, A.: New analytical assessment for fast and complete pre-fault restoration of grid-connected FSWTs with fuzzy-logic pitch-angle controller. Int. J. Electr. Power Energy Syst. 136, art. no. 107745 (2022). https://doi.org/10.1016/j.ijepes.2021. 107745 6. Yoo, J., et al.: RaScaNet: learning tiny models by raster-scanning images. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 13673–13682 (2021) 7. Durnyak, B., Lutskiv, M., Shepita, P., Karpyn, R., Savina, N.: Determination of the optical density of two-parameter tone transfer for a short printing system of the sixth dimension. In: IntelITSIS, pp. 134–140 (2021) 8. Xia, J., Zhang, J., Feng, J., Wang, Z., Zhuang, G.: Command filter-based adaptive fuzzy control for nonlinear systems with unknown control directions. IEEE Trans. Syst. Man Cybern. Syst. 51(3), 1945–1953 (2019)
Synthesis and Modeling of Systems with Combined Fuzzy P + I - Regulators
195
9. Kurniawan, E., Widiyatmoko, B., Bayuwati, D., Afandi, M., Suryadi, Rofianingrum, M.: Discrete-time design of model reference learning control system. In: 24th International Conference on Methods and Models in Automation and Robotics (MMAR), pp. 337–341 (2019). https://doi.org/10.1109/MMAR.2019.8864713 10. Brik, A., Labrak, L., O’Connor, I., Saias, D.: Fast hierarchical system synthesis based on predictive models. In: 18th IEEE International New Circuits and Systems Conference (NEWCAS), pp. 70–73 (2020). https://doi.org/10.1109/NEWCAS49341.2020.9159765 11. Nejati, A., Zhong, B., Caccamo, M., Zamani, M.: Controller synthesis for unknown polynomial-type systems: a data-driven approach. In: The 2nd International Workshop on Computation-Aware Algorithmic Design for Cyber-Physical Systems (CAADCPS), pp. 11–12 (2022). https://doi.org/10.1109/CAADCPS56132.2022.00007 12. Allaboyena, G., Yilmaz, M.: Robust power system stabilizer modeling and controller synthesis framework. In: IEEE International Conference on Electro Information Technology (EIT), pp. 207–212 (2019). https://doi.org/10.1109/EIT.2019.8833656 13. Pasyeka, M., Sheketa, V., Pasieka, N., Chupakhina, S., Dronyuk, I.: System analysis of caching requests on network computing nodes. In: The 3rd International Conference on Advanced Information and Communications Technologies, AICT2019 - Proceedings, pp. 216–222 (2019). https://doi.org/10.1109/AIACT.2019.8847909 14. Valenzuela, P., Ebadat, A., Everitt, N., Parisio, A.: Closed-loop identification for model predictive control of hvac systems: from input design to controller synthesis. IEEE Trans. Control Syst. Technol. 28(5), 1681–1695 (2020) 15. Danilushkin, A.. Closed-loop optimal system :synthesis in the mode of “special” control of a continuous induction heater. In: 2019 XXI International Conference Complex Systems: Control and Modeling Problems (CSCMP), pp. 214–217 (2019). Doi: https://doi.org/10.1109/ CSCMP45713.2019.8976505 16. Pasyeka, M., Sviridova, T., Kozak, I.: Mathematical model of adaptive knowledge testing. In: 5th International Conference on Perspective Technologies and Methods in MEMS Design, MEMSTECH 2009, pp. 96–97 (2009) 17. Oleg, B., Marina, T.: Regulator synthesis for the proportional electromagnet control system. In: 2022 International Conference on Industrial Engineering, Applications and Manufacturing (ICIEAM), pp. 610–614 (2022). https://doi.org/10.1109/ICIEAM54945.2022.9787185 18. Loubach, D.: A high-level synthesis approach applicable to autonomous embedded systems. In: 2022 IEEE XXIX International Conference on Electronics, Electrical Engineering and Computing (INTERCON), pp. 1–4 (2022). https://doi.org/10.1109/INTERCON55795.2022. 9870103 19. Romanyshyn, Y., Sheketa, V., Poteriailo, L., Pikh, V., Pasieka, N., Kalambet, Y.: Socialcommunication web technologies in the higher education as means of knowledge transfer. In: IEEE 14th International Scientific and Technical Conference on Computer Sciences and Information Technologies (CSIT). – 2019(3), pp. 35–39 (2019) 20. Shebanin, V., Atamanyuk, I., Kondratenko, Y.: Synthesis of control systems on the basis of a canonical vector decomposition of random sequences. In: 2020 IEEE Problems of Automated Electrodrive. Theory and Practice (PAEP), pp. 1–5 (2020). doi: https://doi.org/10.1109/PAE P49887.2020.9240808 21. Liu, K., Teel, A., Sun, X., Wang, X.: Model-based dynamic event-triggered control for systems with uncertainty: a hybrid system approach. IEEE Trans. Autom. Control 66(1), 444–451 (2021). https://doi.org/10.1109/TAC.2020.2979788 22. Vodyaho, A., Zhukova, N., Abbas, S., Kulikov, I., Annam, F.: On one approach to the dynamic digital twins models synthesis. In: 2022 XXV International Conference on Soft Computing and Measurements (SCM), pp. 126–128 (2022). doi: https://doi.org/10.1109/SCM55405. 2022.9794895
196
B. Durnyak et al.
23. Botan, C., Ostafi, F.: Discrete time model-following problem for linear systems with variable disturbances. In: The 24th International Conference on System Theory, Control and Computing (ICSTCC), 2020, pp. 54–59 (2020). https://doi.org/10.1109/ICSTCC50638.2020.925 9778 24. Khalifa, T., El-Nagar, A., El-Brawany, M., El-Araby, E., El-Bardini, M.: A novel hammerstein model for nonlinear networked systems based on an interval type-2 fuzzy takagi–sugeno–kang system. IEEE Trans. Fuzzy Syst. 29(2), 275–285 (2021). https://doi.org/10.1109/TFUZZ. 2020.3007460 25. Rashkevich, Y., Peleshko, D., Pasyeka, M., Stetsyuk, A.: Design of web-oriented distributed learning systems. Upravlyayushchie Sistemy i Mashiny, (3–4), 72–80 26. Guo, Y., Hou, Z., Liu, S., Jin, S.: Data-driven model-free adaptive predictive control for a class of MIMO nonlinear discrete-time systems with stability analysis. IEEE Access 7, 102852–102866 (2019) 27. Zeng, Y., Lam, H., Wu, L.: Hankel-norm-based model reduction for stochastic discrete-time nonlinear systems in interval type-2 T-S fuzzy framework. IEEE Trans. Cybern. 51(10), 4934–4943 (2021). https://doi.org/10.1109/TCYB.2019.2950565 28. Sheketa V., Poteriailo L., Romanyshyn Y., Pikh V., Pasyeka M., Chesanovskyy M.: Case-based notations for technological problems solving in the knowledge-based environment.” Paper presented at the International Scientific and Technical Conference on Computer Sciences and Information Technologies, 1, pp. 10–14 (2019). doi:https://doi.org/10.1109/STC-CSIT.2019. 8929784 29. Hussain, N., Ali, S., Ridao, P., Cieslak, P., Al-Saggaf, U.: Implementation of nonlinear adaptive U-model control synthesis using a robot operating system for an unmanned underwater vehicle. IEEE Access 8, 205685–205695 (2020). Doi:https://doi.org/10.1109/ACCESS.2020. 3037122 30. Pasieka, M., Sheketa, V., Pasieka, N., Chupakhina, S., Dronyuk, I.: System analysis of caching requests on network computing nodes. In: 3rd International Conference on Advanced Information and Communications Technologies, AICT 2019, pp. 216–222 (2019) 31. Mushar, K., Hote, Y., Pillai, G.: New mixed model order reduction approach for linear system. In: 2022 IEEE International Conference on Signal Processing, Informatics, Communication and Energy Systems (SPICES), pp. 343–348 (2022). https://doi.org/10.1109/SPICES52834. 2022.9774083
Protection of a Printing Company with Elements of Artificial Intelligence and IIoT from Cyber Threats Bohdan Durnyak1 , Tetyana Neroda1 , Petro Shepita1 , Lyubov Tupychak1 , Nadiia Pasieka2(B) , and Yulia Romanyshyn2 1 Ukrainian Academy of Printing, Pid Goloskom str., 19, 79020 Lviv, Ukraine 2 National Tech, University of Oil & Gas, Ivano-Frankivsk 76068, Ukraine
[email protected]
Abstract. The article considers the actual challenges of cyber security that arise at the current stage of information technology development, with the introduction of artificial intelligence and machine learning. On the basis of the received information, it is proposed to carry out the design of the management system of the printing enterprise with the imitation of an expert. When implementing this function, it was decided to use an element of artificial intelligence, namely neural networks. When designing the proposed analytical apparatus, two alternative options for ANN training were developed for comparison. As a result of the simulation of disturbance perception, the ANN trained using TrustScore showed its effectiveness and superiority. Keywords: Printing company · Internet of things · Artificial Neural network · Cyber attack · Knowledge base
1 Introduction The introduction of existing control and data collection systems into the technological process of small printing enterprises, along with the significant cost of highly specialized, mostly built-in equipment of a limited range, provision of equipment for the placement of the control room and server room, and the introduction of separate positions, requires high qualification of operators and free service, which significantly increases the cost of manufactured products. And services provided. The problems of the application of information technologies in printing are focused on the study of materials and equipment, the analysis of individual effects on the quality of printing, methods of rasterization and their influence on the transfer of image color, the development of means of automation and informatization of printing production. In the conditions of innovative reforming of enterprises, there is a growing shortage in the design of technologies to optimize the management of multi-stage information processes, in particular, machine and instrument engineering, in the printing industry, as well as in education computerization systems. Today, the list of industries in which the operator makes decisions either at key stages of © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 197–205, 2023. https://doi.org/10.1007/978-3-031-36115-9_19
198
B. Durnyak et al.
the technological process or when coordinating the stages of the technological process among themselves is expanding. Significant progress in the modernization of the enterprise is achieved with the introduction of production complexes, which are integrated into the existing technological process with the gradual displacement of the outdated elemental base in the weakest places, in particular, allowing to carry out an operational analysis of the means of production to make an adequate management decision. Along with the introduction of such necessary and modern information technologies, a number of threats to cyber security are observed. Basically, the development of machine learning and artificial intelligence not only facilitates the mental activity of users within the Internet of Things industrial or Internet of Things but also floods these environments with growing. The most common cause is a malfunctioning machine learning model. An adversarial attack can involve presenting a model with inaccurate or false data during training or injecting maliciously crafted data to trick an already trained model. Thus, there is an urgent need to protect the provision of information systems with elements of artificial intelligence implemented at printing enterprises.
2 Analysis of Recent Research and Publications One of the important sources about the inadequacy of the protection of artificial intelligence systems can be considered the report of the National Security Commission of the USA [1], which notes the very small number of studies on the protection of artificial intelligence systems. And neglecting the basic principles of cyber security in conducting scientific research. Also, certain systems that have already been implemented in production are not 100% protected from attacks. In the study [2], the authors placed improvised road markings on the surface of a car with autopilot, which caused the car to cross into the oncoming lane. Work [3–5] proved that small and almost imperceptible changes in the images provided for medical diagnosis can lead to a misdiagnosis and harm to a person. The authors of the study [6–8] gave an example of how a road sign learned by the car control system during machine learning can easily lead to an accident by making corrections to it with improvised means. In the article [9–12], scientists from Google and the University of California proved that even the best forensic classifiers - artificial intelligence systems developed by the US government security departments, trained to distinguish and separate real and synthetic content are vulnerable to attacks. As a VentureBeat contributor points out [13–15], there has been a surge in research on adversarial attacks in recent years. Thus, from 2014 to 2022, the number of papers submitted to the preprint server Arxiv.org on adversarial machine learning increased from 2 to about 1,800. Competitive attacks on artificial intelligence systems received wide publicity at ICLR, Usenix, and Black Hat international conferences. Therefore, when designing and developing systems for IIoT, protection against adversarial attacks must be provided.
3 Presentation of the Main Research Material The paper examines the management of a printing enterprise as a whole structure based on elements of artificial intelligence and machine learning. Using the available elements of information technologies, a decision-making system was built based on the training of
Protection of a Printing Company with Elements of Artificial Intelligence
199
an artificial neural network and the formation of a knowledge bank. Two implementation options were considered and an analysis of the system’s reaction to external threats was carried out. 3.1 Building an Expert Simulator and Training ANNs by Conventional Means The main element of the enterprise management system is the analytical unit, which is responsible for understanding the problem and solving the tasks and localization of disturbances. When designing it, the parameters and main blocks of the knowledge base are first formed [16, 17]. An operator-expert survey was conducted, which determined the main stages and important points of the operation of the technological equipment (Fig. 1). Since the formed knowledge base is also used at the stages preceding the direct production of products, the survey also takes place among other experts. To form a knowledge base, a production goal is defined, after which a certain number of expert operators OEn are interviewed, who describes the situation St that arises during the work, and also give a description of the controlling influences Kv that must be implemented to obtain the desired result, the result Rz [18, 19]. These data form the
Fig. 1. A model of knowledge base formation and imitation of an expert
200
B. Durnyak et al.
properties of experts Fig. 1. OEn = {St, Kv, Rz}
(1)
The survey results of each of the expert operators are formed into a mathematical relationship: Stn + Kvn = Rzn
(2)
where at the output, numerical values of expert operator testing are obtained that correspond to the resulting value of the control action. The results of a mathematical operation are compared: Gn = | Rzn − Rzn+1 |
(3)
where G is the difference between the resulting scores of expert operators; Parameter G is sent to the unit for forming the value of knowledge competence, where the values are added up: n Gi (4) FVK = i=1
If the value of knowledge competence is less than 0.5n, then the corresponding entry is entered into the knowledge base as correct. If this condition is not fulfilled, then these statements are recognized as unreliable. In this regard, an expert simulation block was introduced into the model. To obtain a predicted result based on the received statements of expert operators, after analyzing the applied algorithms that would satisfy the needs of the system, the use of artificial neural networks (ANN) was chosen as the optimal option. After that, a set of knowledge data about the object is formed in the form of a training sample for an artificial neural network from the data received from experts about production goal 1 [20–22]. The data are tabulated in the form of normalized values, and a test and training sample is formed for training an artificial neural network. Training and testing of an artificial neural network is carried out, as a result of which a situational expert is formed based on the formed neural network apparatus, which, depending on the situation, provides predicted results. The resulting situational expert is used if the interviewed expert operators give different answers about the situation. The situation is considered stochastic and may depend on many parameters. In this regard, it is proposed to use a situational expert for the effective operation of an intelligent system with a knowledge base [11, 12, 23, 24]. For its construction, a neural network of direct propagation and the method of the fastest descent, which is implemented by the Levenberg–Marquardt algorithm, i.e. using the nonlinear method of least squares, are used. The training sample is formed from normalized data on the production situation, obtained during the survey of expert operators. The input vector of the expert’s opinion values (3.5) is applied to the input of the neural network, which reflects the parameters necessary for training and further functioning of the system. X = [St1]
(5)
Protection of a Printing Company with Elements of Artificial Intelligence
201
After processing the data according to the selected algorithm, the output vector of values (6) is obtained at the output of the artificial neural network: Rz1 Y = (6) Kv1 In this case, the neural network is represented as the expression of its output Y = Y(X, Q) after adjustment by the weights of Q neurons. The calculation of the network error for one epoch is represented by expression 7: F(Y ) =
1 (St1 − Rz1)2 + (Kv1 − Rz1)2 2
(7)
For the selected algorithm, the input sample must be divided into three subsamples (Fig. 3) in the percentage ratio. The largest number of examples required for training the network is 70%, the other two subsamples require a smaller number of examples, then the validation and testing samples each receive 15% of the records of the total sample [13, 14]. 3.2 Learning ANNs for Recognition from Adversarial Attacks After importing the necessary normalized data, we build a CNN and conduct training on clean data without involving additional ones for training. To prevent external interference, the construction model of the autoencoder, TrustScore, was built and trained on the obtained data set. The essence of this procedure is to calculate indicators of confidence in the existing knowledge base and data set for training. First, we preprocess the training data to find the α-high density sample of each class, which is defined as the training samples within that class after filtering out the α-fraction of samples with the lowest density. Let 0 ≤ α < 1 and let f be a continuous density function with a compact support X ⊆ RD. After that, we define Hα(f), an α-set f with high density, as a set of level λα f defined as the expression: {x ∈ X : f (x) ≥ λα}∂e λα := inf{λ ≥ 0 : X 1[f (x) ≤ λ]f (x)dx ≥ α} (8) To approximate the set of α-high density, the α fraction of points is filtered by the smallest empirical density, based on k-nearest neighbors. This data filtering step is independent of the given classifier h. The next stage is to provide a test sample, we determine the confidence estimate as the ratio between the linguistic data from the sample under study and the high-density α-set of the closest class, different from the predicted class. It is assumed that if the classifier h predicts a label that is significantly further than the closest label, then this is a warning that the simulated expert may be wrong. Thus, our procedure can be seen as a comparison with a modified nearest neighbor classifier, where the modification is to initially filter out linguistic variables that are not included in the high-density α-set for each class [16, 26, 27].
202
B. Durnyak et al.
3.3 Comparison of System Performance In the course of testing the traditional model and the autoencoder with TrusctScore on clean data and on disturbed samples with different disturbance indicators, the following results were obtained, shown in Fig. 2 and Fig. 3, which indicate that the expert imitation model with protection against competitive attacks has been developed, using an autoencoder and TsustScore, is significantly more accurate in recognition and resistant to adversarial FGM attacks compared to the traditional model. Despite the rather high efficiency, with powerful disturbances with a high delta, the information is severely damaged, which makes it virtually indeterminate, although even in this case we observe that the accuracy underlying the agreement with TsustScore is higher.
Fig. 2. Dependence of precession on delta.
where precision_config-accuracy is based on confidence at the output of a regular classifier without a corrector in the form of trust score; precision_trust-precision based on agreement with trust score.An analysis of the obtained graphs was performed, which demonstrates that the accuracy of the classification in agreement with the trust score (Precision_trust) is significantly higher than the accuracy of the base classifier at all delta values, even when the considered samples contain external disturbance and/or are strongly distorted. Where accuracy valuation is the accuracy of the classifier in its classical task without a corrector in the form of a trust score. Analyzing the resulting graphs, it is obvious that with an increase in delta, which causes an increase in the distortion of samples and their loss of important information, the number of samples for which a decision is made decreases, and the accuracy of recognition of disturbance signals obtained from polygraphic possession decreases somewhat. Given that the samples that have not passed the inspection are not allowed to the decision-making stage about them and do not affect the accuracy of the control process on the production task.
Protection of a Printing Company with Elements of Artificial Intelligence
203
Fig. 3. The dependence of precession on delta
4 Conclusion Artificial intelligence systems and Industrial Internet of Things tools, like any software tool, require protection against external unauthorized access, reliability, and safe operation in all respects and conditions. As a result of the study, modern and effective technologies and approaches for the implementation of the analytical block of the management system of a printing company with the ability to recognize various types of cyber-attacks on artificial intelligence systems and protection against them were considered and analyzed.
References 1. Dr. Eric Schmidt and others National security commissionon artificial intelligence interim report november 2019 2. Experimental Security Research of Tesla Autopilot /Tencent Keen Security Lab 2019–03 3. Finlayson, S.G., Bowers, J.D., Ito, J., Zittrain, J.L., Beamand, A.L., Kohane, I.S.: Adversarial attacks on medical machine learning 2019, vol. 363, Issue 6433, pp. 1287–1289 (2019) 4. Eykholt, K., et al.: Robust physical-world attacks on deep learning visual classification. Paper presented at the Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 1625–1634 (2018). https://doi.org/10.1109/CVPR.2018.00175 5. Carlini, N., Farid, H.: Evading deepfake-image detectors with white-and black-box attacks. Paper presented at the IEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshops, 2020-June 2804–2813 (2020). https://doi.org/10.1109/CVPRW5 0498.2020.00337 6. Metzen, J.H., Kumar, M.C., Brox, T., Fischer, V.: Universal adversarial perturbations against semantic image segmentation. arXiv preprint arXiv:1704.05712 (2017) 7. Wang, J., Liang, J., Zhang, L., Ding, X.: Non-fragile dynamic output-feedback control for-gain performance of positive FM-II model with PDT switching: an event-triggered mechanism. Int. J. Robust Nonlinear Control 8. Lopes, I.O., Zou, D., Abdulqadder, I.H., Ruambo, F.A., Yuan, B., Jin, H.: Effective network intrusion detection via representation learning: a denoising AutoEncoder approach. Comput. Commun. 194, 55–65 (2022). https://doi.org/10.1016/j.comcom.2022.07.027 9. Pan, Y., Du, P., Xue, H., Lam, H.K.: Singularity-free fixed-time fuzzy control for robotic systems with user-defined performance. IEEE Trans. Fuzzy Syst. 29(8), 2388–2398 (2020)
204
B. Durnyak et al.
10. Cre¸tu, A., Monti, F., Marrone, S., Dong, X., Bronstein, M., de Montjoye, Y.: Interaction data are identifiable even across long periods of time. Nature Commun. 13(1). https://doi.org/10. 1038/s41467-021-27714-6 11. Durnyak B., Lutskiv M., Shepita P., Nechepurenko V.: Simulation of a Combined Robust System with a P-Fuzzy Controller. Intellectual Systems of Decision Making and Problems of Computational Intelligence: Proceedings of the XV International Scientific Conference, vol. 1020, pp. 570–580 (2019) 12. Imamovi´c, B., Halilˇcevi´c, S.S., Georgilakis, P.S. Comprehensive fuzzy logic coefficient of performance of absorption cooling system. Expert Syst. Appl. 190, art. no. 116185 (2022). https://doi.org/10.1016/j.eswa.2021.116185 13. Salem, A.A., ElDesouky, A.A., Alaboudy, A.H.K.: New analytical assessment for fast and complete pre-fault restoration of grid-connected FSWTs with fuzzy-logic pitch-angle controller. Int. J. Electr. Power Energy Syst. 136, art. no. 107745 (2022). https://doi.org/10.1016/ j.ijepes.2021.107745 14. Tang, P., Ma, Y.: Exponential stabilization and non-fragile sampled-date dissipative control for uncertain time-varying delay TS fuzzy systems with state quantization. Inf. Sci. 545, 513–536 (2021) 15. Lin, J., Njilla, L.L., Xiong, K.: Secure machine learning against adversarial samples at test time. EURASIP J. Inf. Secur. 2022(1), 1–15 (2022). https://doi.org/10.1186/s13635-021-001 25-2 16. Mahmoodabadi, M.J., Andalib Sahnehsaraei, M.: Parametric uncertainty handling of underactuated nonlinear systems using an online optimal input–output feedback linearization controller. Syst. Sci. Control Eng. 9(1), 209–218 (2021) 17. Yoo, J., et al.: RaScaNet: learning tiny models by raster-scanning images. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 13673–13682 (2021) 18. Durnyak, B., Lutskiv, M., Shepita, P., Karpyn, R., Savina, N.: Determination of the optical density of two-parameter tone transfer for a short printing system of the sixth dimension. In: IntelITSIS, pp. 134–140 (2021) 19. Pasieka, N., Sheketa, V., Romanyshyn, Y., Pasieka, M., Domska, U., Struk, A.: Models, methods and algorithms of web system architecture optimization. In: IEEE International ScientificPractical Conference: Problems of Infocommunications Science and Technology, PIC S&T 2019 - Proceedings, pp. 147–152. https://doi.org/10.1109/PICST47496.2019.9061539 20. Pasieka, M., Sheketa, V., Pasieka, N., Chupakhina, S., Dronyuk I.: System analysis of caching requests on network computing nodes. In: 3rd International Conference on Advanced Information and Communications Technologies, AICT 2019-pp. 216–222 (2019) 21. Pasyeka, M., Sheketa, V., Pasieka, N., Chupakhina, S., Dronyuk, I.: System analysis of caching requests on network computing nodes. In: 3rd International Conference on Advanced Information and Communications Technologies, AICT2019 - Proceedings, pp. 216–222. https:// doi.org/10.1109/AIACT.2019.8847909 22. Pasyeka, M., Sviridova, T., Kozak, I.: Mathematical model of adaptive knowledge testing, 5th International Conference on Perspective Technologies and Methods in MEMS Design, MEMSTECH 2009, pp. 96–97 (2009) 23. Sun, K., Mou, S., Qiu, J., Wang, T., Gao, H.: Adaptive fuzzy control for nontriangular structural stochastic switched nonlinear systems with full state constraints. IEEE Trans. Fuzzy Syst. 27(8), 1587–1601 (2018) 24. Pasieka, N., Sheketa, V., Romanyshyn, Y., Pasieka, M., Domska, U., Struk, A.: Models, methods and algorithms of web system architecture optimization. In: IEEE International Scientific-Practical Conference: Problems of Infocommunications Science and Technology, PIC S&T 2019 - Proceedings, pp. 147–152. Doi:https://doi.org/10.1109/PICST47496.2019. 9061539
Protection of a Printing Company with Elements of Artificial Intelligence
205
25. Bandyra, V., Malitchuk, A., Pasieka, M., Khrabatyn, R.: Evaluation of quality of backup copy systems data in telecommunication systems. In: IEEE International Scientific and Practical Conference Problems of Infocommunications Science and Technology.-PIC S&T 2019.-08– 11 October 2019, Ukraine, pp. 329–325, Doi:https://doi.org/10.1109/PICST47496.2019.906 1379 26. Yoo, J., et al.: RaScaNet: learning tiny models by raster-scanning images. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 13673–13682 (2021) 27. Romanyshyn, Y., Sheketa, V., Poteriailo, L., Pikh, V., Pasieka, N., Kalambet, Y.: Socialcommunication web technologies in the higher education as means of knowledge transfer. In: IEEE 14th International Scientific and Technical Conference on Computer Sciences and Information Technologies (CSIT), vol. 3.-2019.-Lviv, Ukraine, pp. 35–39 (2019)
AMGSRAD Optimization Method in Multilayer Neural Networks S. Sveleba, I. Katerynchuk(B) , I. Kuno, O. Semotiuk, Ya. Shmyhelskyy, S. Velgosh, N. Sveleba, and A. Kopych Department of Optoelectronics and Information Technologies, Ivan Franko National University of Lviv, Lviv 79017, Ukraine {serhiy.sveleba,ivan.katerynchuk,ivan.kuno}@lnu.edu.ua
Abstract. The AMSGrad optimization learning method for multilayer neural network (MNN) was tested using the logistic function and the Fourier spectra of the error function. The logistic function describes the process of doubling the number of local minima. It was established that due to retraining of MNN, the learning error function of each neuron is characterized by a set of wave vectors with different periodicities. The average value of the learning error for all neurons can be considered as the average value for all existing periodicities. At the same time, the wave vector of the total oscillation can take both commensurate and incommensurate values. The optimization method of AMSGrad training leads to a change in the frequency spectrum of the existing periodicities of the error function on each neuron. That is, the speed of learning of each neuron is corrected, which removes the degeneracy of this system by preventing relearning processes of each neuron. Keywords: Multilayer neural network (MNN) · AMSGrad optimization method · Block structure
1 Introduction 1.1 AMSGrad Optimization Method The most common way to optimize neural networks is the gradient descent method. Gradient descent is an optimization algorithm that follows the negative gradient of an objective function to find the minimum of the error function. In most cases, gradient methods are based on an iterative procedure implemented according to the formula: wk+1 = wk + ηk p(wk ) where wk , wk+1 the current and new approximation of the values of controlled variables to the optimal solution, respectively, ηk is the convergence step, p(wk ) is the search direction in the N-dimensional space of controlled variables. The method of determining p(wk ) and αk at each iteration depends on the specifics of the method. The most common Cauchy method (the method of fastest descent) consists in implementing the rule: wk+1 = wk − α
∂J (w, b, x, y) ∂w
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 206–216, 2023. https://doi.org/10.1007/978-3-031-36115-9_20
AMGSRAD Optimization Method in Multilayer Neural Networks
207
∂J Denoting the current gradient g = ∂w , we get: wk+1 = wk − αk gk , where αk is the learning rate. The complete form of the derivative of the evaluation function ∂J (w,b,x,y) = αk δi , where δi − (yi − hi )f (zi ), zi = ni wi xi , xi - input variables, yi ∂wik expected output values, hi - calculated output values, and node number in the output layer. A limitation of the gradient descent method is that this method applies a single learning rate for all input variables. An extension of the gradient descent method, such as the Adaptive Movement Estimation (Adam) algorithm, uses a different learning rate for each input variable, but as a result, the learning rate can quickly decrease to very small values [1]. The AMSGrad method is an extended version of the Adam method, which attempts to improve the convergence properties of the algorithm by avoiding large abrupt changes in the learning rate for each input variable. Technically, gradient descent is called a firstorder optimization algorithm because it explicitly uses the first-order derivative of the objective function. It is known [2, 3] that the AMSGrad algorithm updates the exponential moving averages of the gradient (mt ) and the square of the gradient (vt ), where the hyperparameters β1 , β2 (the value of which varies in the interval [0,1)) control the exponential decay rates of these moving averages. The moving averages themselves are estimates of the 1st moment (mean) and 2nd moment (uncentered variance) of the gradient [4]. Thus, AMSGrad results in no step size increase, which avoids the problems experienced by Adam. For simplicity, the authors of [2] also remove the offset step used in Adam. A full AMSGrad update without corrected grades can be submitted as follows: mt = β1 N +1 mt−1 + 1 − β1 N +1 δt
vt = β2 vt−1 + 1 − β2 δt 2 vt = max v t−1 , vt
wt+1 = wt − αht mt / v t
where: mt – is the moving average of the gradient of the error function, vt – is the moving average of the square of the gradient of the error function, α – is the learning speed, N – is the number of epochs. Our preliminary studies have shown that, unlike MNN in which no optimization methods were used, in the considered MNN, when the AMSGrad optimization method is applied, with an increase in the learning speed, cascades of transition to a chaotic state and exit from it are traced. Their number increases as learning speed increases. Therefore, the purpose of our work is to conduct a study of chaotic states and the processes of entering and exiting these states, that is, the mechanisms of the cascade learning process when applying the AMSGrad optimization method. The goal of present work is to analyze the appearance of local minima of the error function and the mechanism of influence of the AMSGrad optimization method on the process of blocking their appearance.
208
S. Sveleba et al.
2 Methodology 2.1 Software Implementation A program for a MNN with hidden layers for recognizing printed digits was written in the Python programming environment. The array of each number consisted of a set of “0” and “1” of size 4x7. The sample of each digit contained a set of 4 possible distortions of the digit and a set of 3 arrays that did not correspond to any of the digits. For example, for the number “0” an array of x values: Numt1=[0,0,0,0,1,1,1,1,1,1,1,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] Numt2=[1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1] Numt3=[0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] Num01=[1,1,1,1,1,0,0,1,1,0,0,1,1,0,0,1,1,0,0,1,1,0,0,1,1,1,1,1] Num02=[1,1,0,1,1,0,0,1,1,0,0,1,1,0,0,1,1,0,0,1,1,0,0,1,1,1,1,1] Num03=[1,1,1,1,1,0,0,1,0,0,0,1,1,0,0,1,1,0,0,1,1,0,0,1,1,1,1,1] Num04=[1,1,1,1,1,0,0,1,1,0,0,1,1,0,0,1,1,0,0,1,0,0,0,1,1,1,1,1] Num05=[1,1,1,1,1,0,0,1,1,0,1,1,1,0,0,1,1,0,0,1,1,0,0,1,1,1,1,1] an array of y values: Num0Y= [[0],[0],[0],[1],[1],[1]] The neural network described with the help of this program contained 3 hidden layers with 28 neurons in each layer. The choice of the number of hidden layers and neurons in each of them was determined by the smallest error of learning and recognizing numbers. According to [5], this is a three-layer neural network with 28 neurons in each layer. The β1 = 0.9 and β2 = 0.999 parameter values were chosen as in [4]. According to [6], the sigmoidal function was chosen as the activation function. The implementation of this optimization method was carried out with the help of the following code:
AMGSRAD Optimization Method in Multilayer Neural Networks
209
for i in range (num -1): layer_errors.append(layer_deltas[i].dot(synapse[num - 1 - i].T)) layer_deltas.append(layer_errors[i + 1] * sigmoid_output_to_derivative(layers[num - 1 - i])) layer_deltass=layer_errors[i + 1] * sigmoid_output_to_derivative(layers[num - 1 - i]) # m(t) = beta1(t) * m(t-1) + (1 - beta1(t)) * g(t) m = beta1**(age+1) * m[i-1] + (1.0 - beta1**(age+1)) * layer_deltass # v(t) = beta2 * v(t-1) + (1 - beta2) * g(t)^2 v = (beta2 * v[i-1]) + (1.0 - beta2) * layer_deltass**2 # vhat(t) = max(vhat(t-1), v(t)) vhat = max(max(vhat.reshape(-1,1)), max(v.reshape(-1,1))) dd= m / np.sqrt(vhat+ 1e-8) d.append(dd) for i in range (num): synapse[num -1 -i] -= alpha * (layers[num -1 -i].T.dot(d[i]))
where num = 3 is number of layers of MNN. where δ- layer_deltass, hi -layers, age is the number of epochs, num is the number of hidden layers. 2.2 Logistics Function To analyze the error function, a logistic function of the following type was used: xn+1 = α − xn − xn 2 where n is a step, alpha is a parameter that determines the learning rate. Its fixed points: x1,2 = −1 ± (α + 1)1/2 eigenvalues, which can be calculated as follows: ρ1,2 = 1 ± 2(α + 1)1/2 The choice of this logistic mapping is due to the fact that it describes the process of doubling the frequency of oscillations [7]. In our case, this process is due to the occurrence of local minima when approaching the global minimum. For one-dimensional mappings, there are 2 ways to change the stability of a fixed point, when the point multiplier is ρ = +1 and ρ = −1. However, the number of related bifurcations (doublings) is noticeably larger. This is because they often involve more than one fixed point. 4 variants of bifurcations correspond to such a situation: tangential bifurcation (fold, saddle-nodal); transcritical bifurcation; fork-shaped bifurcation (bifurcation of loss of symmetry); doubling bifurcation.
210
S. Sveleba et al.
3 Application of the AMSgrad Optimization Method 3.1 AMSGrad and Different Number of Epochs Figure 1 shows the result of the developed program. Provided that beta1 = 0.9, beta2 = 0.999 at, Fig. 1 shows the dependence of the value of the logistic error function on the parameter α and Fourier spectra for 100 epochs and β1 = 0.9, β2 = 0.999. . The resulting branching diagram proves that the entire investigated range of α changes (0.000001 ÷ 0.008) can be divided into 4 parts: 1) the range of a sharp decrease in the value of the error function (α = 0.000001 ÷ 0.00002) – no retraining; 2) the range of slightly variable, monotonous behavior of the error function (α = 0.00002 ÷ 0.00025) is a satisfactory learning process; 3) the bifurcation range on the branching diagram (α = 0.00025 ÷ 0.00047) - the process of relearning (Fig. 1, b); 4) the range of chaotic non-monotonic behavior of the error function from alpha (α = 0.00047 ÷ 0.008) the appearance of chaos.
Fig. 1. Branching diagrams (a, b) in the range of α = 0 ÷ 0.00055. Fourier spectra (c) - during retraining, and (d) - in conditions of chaos, 100 epochs, β1 = 0.9, β2 = 0.999, digit “0”, when using the AMSGrad optimization method.
The process of retraining is accompanied by a transition through the global minimum and a doubling of the number of local minima. This process is especially well manifested at small values of the number of epochs (Fig. 2). Although here, too, the process of blocking the doubling of the number of local minima begins to follow, which makes it impossible for the system to transition to a chaotic state (Fig. 2). With an increase in the number of epochs (N ), a decrease in the gradient of the error function and an increase in the value of the expression 1 − β 1 N , at small values of 1 − β 2 , lead to the fact that the gradients become rarefied, since the ratio of the
AMGSRAD Optimization Method in Multilayer Neural Networks
211
vectors of the first and second moments forms block structure that specialize in certain patterns. This is clearly manifested at small epoch values (N = 5 at α > 0.0017; (Fig. 2)). Therefore, the optimization process especially manifests itself when the number of epochs increases, leading to a decrease in the gradient when approaching the global minimum.
Fig. 2. Branching diagrams for learning speed in the enlarged version, (a, b, c), 5 epochs, β 1 = 0.9, β 2 = 0.999, digit ‘0’, when using the AMSGrad optimization method.
3.2 Analysis of the Block Structure in the AMSGrad To analyze the appearance of the block structure in the branching diagrams, consider the behavior of the learning error function from the learning rate at 10 epochs. Figure 3a shows the branching diagram under the condition of 10 epochs, β1 = 0.9, β2 = 0.999, , digit “0”, and in an enlarged version of the intervals of existence of the block structure (Fig. 3b - m). The entire range of retraining, on the branching diagram, is characterized by the division into blocks (Fig. 3b). Let us consider in more detail each interval of existence of the block structure and conduct an analysis of the Fourier spectra in these ranges. It should be noted that the AMSGrad optimization method is based on the analysis of the objective function based on the values calculated in the previous step, so the Fourier spectra of the error function were calculated for a small interval of the learning rate change in each block. The first interval is the interval α = 0.0002 ÷ 0.00075 of a satisfactory learning process. At the same time, the Fourier spectra show the absence of harmonics. On the interval α = 0.00175 ÷ 0.018, the process of retraining of the neural network with the appearance of a bifurcation on the branching diagram (Fig. 3c) and the second harmonic on the Fourier spectra (Fig. 4a) can be traced. Moving further along the branching diagram as a result of the doubling process, we fall into the interval of the appearance of a chaotic state (α = 0.001825 ÷ 0.01914) (Fig. 3d). The Fourier spectra of this chaotic state, according to Fig. 4c, prove that a chaotic state arises as a result of doubling the number of local minima and not passing through the global minimum once. This interval of the existence of a chaotic state is characterized by areas of “transparency”, where there is no chaotic state of the neural network. These areas prove the absence of a chaotic state, and therefore a sharp decrease in the number of local minima. That is, the emerging chaotic state is characterized by the existence of a block structure, which, according to the authors, consists of areas of constant values of
212
S. Sveleba et al.
the number of local minima, and the averaged value over this block gives a chaotic value. This assumption is supported by the fact that the considered system is dynamic, since the considered optimization method works only if the next value depends on the previous one. Entry into the chaotic state goes through the doubling process, and the appearance
Fig. 3. Branching diagrams (a) from the learning speed in the enlarged version, (b - m)), 10 epochs, β1 = 0.9, β2 = 0.999, digit ‘0’, when using the AMSGrad optimization method.
AMGSRAD Optimization Method in Multilayer Neural Networks
213
of intervals of the absence of chaotic solutions in the chaos domain is accompanied by a sharp transition from chaos to symmetry and vice versa. Moving on to the next area of change α = 0.001925 ÷ 0.001975 where the neural network is characterized by the absence of a chaotic state (Fig. 3d, f and Fig. 4c). In this region of α change, the Fourier spectra of the target function are characterized by the presence of harmonics (Fig. 4c), and therefore the learning process of the neural network is accompanied by retraining. According to the branching diagram, this state of the neural network in this area is not chaotic, but is characterized by the processes of branching initiation and its disappearance (Fig. 3d, f). Therefore, the Fourier spectrum
Fig. 4. Fourier spectra in different ranges of learning speed changes, 10 epochs, β1 = 0.9, β2 = 0.999, digit ‘0’, when using the AMSGrad optimization method.
214
S. Sveleba et al.
of the objective function in this area of change in learning speed is characterized by the appearance of harmonics (Fig. 4c), which in terms of signal power is much smaller than in the area of α change, where only the relearning of this network can be traced (Fig. 3c). The next area of alpha change is the area of cascade transformations (α = 0.001990– 002000) shown in Fig. 3e. The branching diagram shows several transitions to and from the chaotic state. Under these conditions, the Fourier spectra are characterized by the appearance of higher harmonics and the absence of a chaotic state (Fig. 4d). Moving to the next interval of change α = 0.00210 ÷ 0.00234, the branching diagram shows a transition to a chaotic state due to the process of doubling the number of local minima and repeatedly passing through the global minimum (Fig. 3g). Under the given conditions, the Fourier spectra prove the existence of a chaotic state, but this chaotic state is different from the chaotic state that exists at α = 0.001825 ÷ 0.01914. The difference is that at α = 0.001825 ÷ 0.01914 Fourier spectra show the coexistence of a chaotic state with a state characterized by the existence of only a few harmonics. Therefore, the interval of change of α = 0.001825 ÷ 0.01914 may be characterized by the existence of a block structure. The next interval α = 0.00235 ÷ 0.00241 (Fig. 3g-k) is characterized by the absence of doubling on the branching diagram. Fourier spectra in this range also show the almost absence of harmonics (Fig. 4e). Therefore, in this interval of α changes (α = 0.00235 ÷ 0.00241), neural network learning is not characterized by retraining, that is, a satisfactory learning process takes place. Approaching the next interval of existence of a chaotic state (α = 0.00241 ÷ 0.002419), branching is observed on the branching diagram (Fig. 3k) with the appearance of harmonics in the Fourier spectra (Fig. 4g). A further increase in α leads to the appearance of a chaotic state on the branching diagram (Fig. 3k, l) in the range α = 0.00242 ÷ 0.00338. Fourier spectra at these values of α also confirm the existence of a chaotic state (Fig. 4h, α = 0.00242–00245). This chaotic state is no different from other chaotic states that occur at α = 0.001825–0019, α = 0.0021–00215 (Fig. 4b, f). With a further increase in the learning speed, as a result of a decrease in the number of doublings (according to the branching diagram in Fig. 3l, α = 0.003385–003395), possibly due to a decrease in the number of local minima, the absence of a chaotic state can be traced. Fourier spectra (Fig. 4i) of this interval of α change are characterized by a small number of harmonics, which are several times smaller in magnitude than in other intervals of the absence of chaos. A further increase in the learning speed is accompanied by a transition to a chaotic state (Fig. 3m), which is inherent in a larger range of changes in α > 0.004 (Fig. 3h). An interesting feature of this chaos is that its Fourier spectra are not characterized by the existence of n-fold harmonics. According to Fig. 4k at α = 0.0034–003405, the Fourier spectrum of this chaotic state is characterized by a smaller number of harmonics, which proves the existence of a mechanism for blocking the occurrence of harmonics.
4 Summary and Conclusion The learning error on a single neuron is described by its functional dependence, therefore the learning error at a given speed and a given step is a symbiosis of all neurons involved in the learning process. Since the learning error is calculated as the root-mean-square
AMGSRAD Optimization Method in Multilayer Neural Networks
215
deviation, the given functional dependences of the learning error on each neuron are, in the first approximation, periodic functions with different periods. That is, these functional dependencies are characterized by a spectrum of wave vector values. These vectors are characterized by rational values (that is, proportional fluctuations determined by the number of neurons in a layer and the number of hidden layers). The total value of the wave vector for such an ensemble of periodicities can take both commensurate and incommensurate values. It is known that in MNN, the correction of the weights of one neuron is influenced by all the neurons of the previous layer. Since this influence from the learning speed becomes non-uniform when approaching the global minimum, it can lead to its weakening, and therefore to non-uniform learning. This heterogeneity should manifest itself more when using optimization methods that are based on the use of an algorithm that updates the exponential sliding mean gradients (mt ) and the square of the gradients (vt ) based on previous values. That is, the appearance of a block structure, which is characterized by the coexistence of both a chaotic state and a state characterized by several harmonics, should be inherent to such multilayer neural networks. Based on the above results (Fig. 3), this state may exist in the interval α = 0.001982–0.001990. Therefore, when applying the AMSGrad method to the learning process of MNN, a block structure may arise. This block is characterized by the stochastic learning process of the neural network. Namely, in a certain range of changes in the learning speed, retraining of the neural network is observed, which is accompanied by the appearance of local minima. An increase in the number of local minima leads to the appearance of higher order harmonics. An increase in the number of doublings of existing harmonics causes the system to transition to a chaotic state. This occurs under the condition that the considered neural network training method is related to the correction of the weights of each neuron at a given epoch, and this weight correction is affected by all neurons from the previous layer. So, in this case, the learning error function for each neuron should be considered as a functional dependence described by a set of wave vectors of different periodicities. In this case, the average value of the learning error for all neurons can be considered as the average value for all existing periodicities. And the value of the wave vector of the total oscillation can take both commensurate and incommensurate values. Therefore, AMSGrad algorithm leads to a change frequency spectrum of existing periodicities of functional dependencies of each neuron. That is, the speed of learning of each neuron is corrected, which removes the degeneration of this system by preventing the processes of relearning the neural system. The results obtained above prove that the appearance of local minima is caused by non-uniform learning of the neural network, which is associated with the retraining of individual neurons. An increase in the number of local minima with an increase in the learning rate indicates an increase in the number of such neurons. The AMSGrad optimization method, due to the control of the exponential rate of decline of the average gradients and the square of the gradient of the target error function, causes a decrease in the number of retraining neurons.
216
S. Sveleba et al.
References 1. Kingma, D.P., Ba, J.L.: Adam: a method for stochastic optimization. Published as a conference paper at the 3rd International Conference for Learning Representations, San Diego (2015). https://doi.org/10.48550/arXiv.1412.6980 2. Phuong, T.T., Phong, L.T.: On the Convergence Proof of AMSGrad and a New Version IEEE Access (Volume: 7), 2019, 61706–61716. https://ieeexplore.ieee.org/document/8713445 3. Reddi, S.J., Kale, S., Kumar, S.: On the convergence of adam and beyond computer science. Published as a conference paper at ICLR 2018 [Submitted on 19 Apr 2019] arXiv:1904.092 37v1 https://doi.org/10.48550/arXiv.1904.09237 4. Brownlee, J.: Gradient Descent Optimization with AMSGrad From Scratch, 9 June 2021 https:// machinelearningmastery.com/gradient-descent-with-adagrad-from-scratch/ 5. Sveleba, S., et al.: Chaotic states of multilayer neural network. Electronics and information technologies. Issue 13, pp. 20–35 (2021). https://doi.org/10.30970/eli.16.3 6. Sveleba, S., Katerynchuk, I., Kuno, I., Semotiuk, O., Shmyhelskyy, Ya., Sveleba, N.: Peculiarities of the dependence of the learning error of multilayer neural networks on the activation function in the process of recognizing of printed digits. Electron. Inf. Technol. 17, 36–53 (2022). https://doi.org/10.30970/eli.17.4 7. Taranenko, Yu.: Information entropy of chaos. https://habr.com/ru/post/447874
Regional Economic Development Indicators Analysis and Forecasting: Panel Data Evidence from Ukraine Larysa Zomchak(B) , Mariana Vdovyn, and Olha Deresh Ivan Franko National University of Lviv, Lviv 79000, Ukraine [email protected]
Abstract. The development of Ukraine’s economy as a socio-economic system is determined by the development of its constituent subsystems - regions that function successfully if there are resources necessary for development and their economic evaluation. In this regard, it is important to investigate, due to which factors this economic growth is achieved, what is the contribution of each of these factors to the overall economic success of the country or region. The analysis of the economic development of the regions highlighted the important factors of the regional economic growth: the gross regional product, turnover of retail trade; volume of sold industrial products; capital investments; export volumes. The panel data approach is used in the investigation for regional economic development of Ukraine modelling. Gross regional product as the main economic indicator in the level of region is a dependent variable, independent variables are gross regional product in two previous periods, export, capital investment in previous periods, turnover of retail trade and volume of sold industrial products with two lags. So, with econometric modeling methods the main determinants of the regional economic development of Ukraine were revealed and the level of impact for each of them was estimated. The value of the gross regional product in the regions of Ukraine for the next period is forecasted. Keywords: Regional development · Economic development · Macroeconomic modelling · Panel model
1 Introduction Considering the peculiarities of regional development, the strengths and weaknesses of the country’s regions functioning, negative and positive trends investigations allows making effective decisions at both the mezzo and macro levels. At the level of the regions socio-economic development projects are implemented, and the indicators of the regions determine the level of country development. It can be argued that the study of the development of the regions of Ukraine is a relevant and timely issue, especially in the conditions of the full-scale invasion of russia on the territory of Ukraine, which will certainly have and will have a negative impact on the development of the regions and Ukraine as a whole, will cause great disparities in the development of the regions, © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 217–228, 2023. https://doi.org/10.1007/978-3-031-36115-9_21
218
L. Zomchak et al.
will lead to reduction of all economic indicators due to large expenditures by the state to ensure a peaceful sky over the heads of all Ukrainians, due to the occupation of part of the territory, due to the cessation of business activities, etc. The decentralization reform, which began in 2014 in Ukraine, was aimed at creating active and capable territorial communities, and not just transferring powers from state bodies to local self-government bodies. In the conditions of war, it is obvious that the emphasis shifts again to a strong central government, while it is important to understand the role, place and prospects of each of the individual regions in the defense of the country. Due to the special structure of the panel data allow to build more flexible and meaningful models and obtain answers to questions not available within models based, for example, only on spatial data for individual national economies. It becomes possible to take into account and analyze individual differences between economic units, which cannot be done within the framework of standard regression models. The purpose of the investigation is modeling and analysis of the specifics of economic growth, factors that influence it, in terms of individual regions of Ukraine at a certain point in time, development of such a model that would be suitable for further forecasting. The application of models on panel data makes it possible to obtain regression equations for each region of Ukraine in particular, and thus to investigate the influence and weight of economic impact factors in each region, to compare the economic development of different regions of Ukraine, to identify the strengths and weaknesses of the regional economy.
2 Literature Review The investigation of the economic development at the level of the region is considered in the articles of many scientists. A review of the main theories of regional economic development was conducted by Coccia M. [1], a review of the latest research on regional development can be founded in the article by Malecki E. [2], Gibbs D. and O’Neill K. [3] reviewed the scientific problems of regional development in the context of the green economy, Smol M. and co-authors[4] – in the circular economy, Aliyev, A. [5] – in the digital economy, and Amin E. [6] – in the context of institutional perspectives. An overview of models of regional economic development can be found in the study of Treys G. [7], and the implementation of models of regional economic development based on US statistical data in the article by Buchholz M., Bathelt H. [8]. More about methods and models, used for investigation on the regional level, in the Table 1. For modeling regional development, it is advisable to use models based on panel data, because they allow simultaneous consideration of several variables collected over time for the same objects. The purpose of the study is to model and analyze the specifics of economic growth, factors that influence it, in a section of regions in Ukraine at a certain point in time. On the basis of the selected factors of regional economic growth, the implementation of such a model, which would be suitable for further forecasting.
Regional Economic Development Indicators Analysis and Forecasting
219
Table 1. Methods and models used for regional economic development investigation Authors
Method and model
Xin, J., & Lai, W. (2021) [9]
multiple statistic analysis
Guangyu, T. (2020) [10]
ARIMA
Hu, Y., & Wang, Z. (2022) [11]
support vector machine
Yu, W., & Huafeng, W. (2019) [12], Matviychuk, A., Lukianenko, O., & Miroshnychenko, I. (2019) [13]
fuzzy logic
Xu, X., & Zeng, Z. (2021) [14], Chagovets, L., Chahovets, V., & Chernova, N. (2020) [15]
machine learning
Li, Z., Cheng, J., & Wu, Q. (2016) [16]
principal component analysis
Zhong, Z., & He, L. (2022) [17]
agent-based model
Emmanuel Halkos, G., & Tzeremes, N. G. (2010) [18]
data envelopment analysis
Santos, L. D., & Vieira, A. C. (2020) [19]
spatial econometrics
Shen, Y. (2020, October) [20]
panel vector autoregressive model
Márquez, M. A., Ramajo, J., & Hewings, G. J. (2013) [21] spatial vector autoregressive model Zomchak, L., & Lapinkova, A. (2023) [22]
ARDL
Jiang, X., He, X., Zhang, L., Qin, H., & Shao, F. (2017) [23]
structural equation model
Allo, A. G., Dwiputri, I. N., & Maspaitella, M. (2022) [24]
input-output model
Chen, L., Yu, N. N., & Su, Y. S. (2014, August) [25]
game theory
Dutta, P. K., Mishra, O. P., & Naskar, M. K. (2013) [26], topography analysis Alhosani, N. (2017) [27], Xu, F., Chen, Z., Li, C., Xu, C., Lu, J., & Ou, Y. (2011) [28] Koroliuk, Y., & Hryhorenko, V. (2019) [29], Ivan Izonin, et. al.[30], Tkachenko R., et. al.[31]
neural network
Lytvynenko, V., Kryvoruchko, O., Lurie, I., Savina, N., Naumov, O., & Voronenko, M. (2020). [32]
self-organizing algorithm
3 Method of Panel Data Modelling in Economic Research First of all, the application of models on panel data makes it possible to detect and analyze changes at the individual level, which is impossible neither within the framework of individual time series models nor within the framework of variation series models. Panel data models can be used not only to explain the behavior of different research objects, they can also explain why individual research objects behave differently in different time periods. The panel data models require compliance with the following assumptions: panel data should be balanced, panels are characterized by short time series; in order to take
220
L. Zomchak et al.
into account the time effect, additive dummy variables can be used; the possibility of taking into account specific individual effects (disturbances). In general, the longitudinal data model can be presented in the form: Yit = α + Xit βit + εit where Yit is the value of the indicator for the object i (individual, region, firm, etc.) in the time period t (i = 1,2, 3,…, N; t = 1, 2, 3,…, T); Xit is a vector of independent variables; εit —disturbance for the object i (individual, region, firm, etc.) in the time period t (i = 1,2, 3,…, N; t = 1, 2, 3,…, T); α is a scalar; βit are model parameters that measure the partial effects of a change in Xit . As we can see, such a model is similar to the multiple regression, so the methods of unknown parameters estimating are also similar. In addition, models of longitudinal data make it possible to additionally divide disturbances into several components, and accordingly, according to this feature, all models can be divided into one- (one-way error component model) and two-way error component (two-way error component) errors. In the paper, we consider models with a one-dimensional error component, because they are the most common in practice.
4 Results: Economic Development of the Regions of Ukraine Model Specification The number of determinants of the regional economic development is rather large, so it is quite difficult to take all of them into account, and even more so to accurately forecast them, especially when the entire economy of the country works in wartime mode. Therefore it is necessary to analyze a certain set of factors and select those that have the most significant influence. The following determinants of the regional economic development were identified: – – – – – –
Retail turnover (million UAH); Volume of sold industrial products (million UAH); Capital investments (million UAH); Volume of manufactured construction products (million UAH); Export (million USD); Employed population (thousands of people).
Gross regional product is the dependent variable as one of the main indicators of the regional economic development. The model for forecasting the gross regional product, as an indicator of the economic growth of the region, can be written in the form (1): VRP = f (ORT ; OR Pr; KI ; OVBP; EXPORT ; ZN )
(1)
where VRP is gross regional product (million UAH); ORT – turnover of retail trade (million UAH); ORPr – volume of sold industrial products (million UAH); KI – capital investments (million UAH); OVBP – volume of manufactured construction products
Regional Economic Development Indicators Analysis and Forecasting
221
(million UAH); EXPORT – export volumes (million USD); ZN – employed population (thousands of people). The construction of the model will take place across 24 regions of Ukraine based on retrospective data from 2010 to 2020. The first important step in the specification of the model is to check the series for stationarity - Panel Unit Root Test. For example, export as a factor of economic growth is not stationary because the probability value for the tests is greater than 5%, that is why we reject the null hypothesis and can say that the original series is not stationary. Next, we have to check whether this series is stationary in the first differences, etc. So, after checking all the series, we conclude that they are all stationary in the first differences. The next important step in the study of such a phenomenon as economic development is conducting a test for the causal relationship between the dependent and independent variables - the Granger test (Table 2). Table 2. Granger ‘s test for causality Variable
X2 - statistics
ZN
12.60959
OVBP
9.092023
p- value 0.126 0.3346
ORT
26.19425
0.001
ORPR
22.84009
0.0036
KI
16.36029
0.0375
EXPORT
18.80705
0.0159
Variables for which the p- value is larger than 5% have no or rather weak effect on the gross regional product. So, we can say that the employed population and the volume of manufactured construction products do not have a significant impact and these variables can be excluded from the model. Panel autoregressive model in which the previous values of the variables can influence the current values with a certain lag, that is, with a certain “lag” in time. The definition of lags also occupies a very important place in the specification of models. For this, it is necessary to conduct a test - VAR Lag Order Selection test, which is based on different criteria, namely Akaike, Schwartz, Hannan-Quinn. For conclusions, it is necessary to take the smallest informational criterion. In our case, the value of the number of lags for all criteria coincides and is two (Table 3). Also confirmed by another test - VAR Lag Exclusion Wald Test. The assumption that four lag can be applied on quarterly data was not confirmed. For choosing the type of model, it is necessary to conduct the Durbin -Vue- Hausman test, which will answer the question of which model to choose. This test also checks whether random effects estimation produces unbiased estimates or not. After calculating this value, we obtained the following result (Table 4).
222
L. Zomchak et al.
Table 3. Lag determination according to Akaike ‘s criteria, Schwartz and Hannan-Quinn criterious Lag
AIC ( Akaike)
SC ( Schwartz)
HQ ( Hannah-Queen)
0
19.9785
20.16548
20.05408
1
19.83488
20.04857
19.92125
2
19.78521*
20.02562*
19.88239*
3
19.80579
20.07291
19.91376
4
19.815
20.10883
19.93377
5
19.8157
20.13624
19.94527
6
19.81178
20.15903
19.95214
Table 4. The Darbin -Vue- Hausman test results Summary test
X 2 - statistics
Prob
Random period
53.373701
0.0000
As one can see, the calculated value of the X2 statistic is less than the critical value calculated according to the table, and the p - value is less than 5%. That is, the null hypothesis is not rejected. Therefore, we can conclude that the data will be better described by a model with fixed effects. The next important test for model specification is the Wald test (Table 5). It gives a clear understanding of whether we really need to use cross-sectional effects. Table 5. The Wald test results Summary test F-statistics X 2 - statistics
Value
Prob
2.359498
0.0013
58.230242
0.0001
According to the value of the F-statistic, we cannot reject the null hypothesis (p-value less than 0.05) that the model with the common cross-section is worse than the model with fixed effects. So, as a result of the conducted tests, we obtained a panel autoregression model with fixed cross-sectional effects of the economic development of the regions of Ukraine.
Regional Economic Development Indicators Analysis and Forecasting
223
5 Discussion: Panel Data Model of Ukraine Regions Economic Development Application In order to estimate model parameters, it is necessary to use the generalized least squares method (panel EGLS), because this approach will allow us to avoid problems related to autocorrelation and heteroskedasticity (assumption of constancy of variance) in the future. That is, in contrast to the least squares method, this method takes into account information about the inequality of variance and therefore makes it possible to obtain the best linear estimates. The model specified above will look like this (2): VRPit = αi +
p
ϕij VRPi,t−j +
j=1
+
q
σij EXPORTi,t−j +
j=1 q
γij ORTi,t−j +
j=1
q
βij KIi,t−j
j=1 q
θij ORPRi,t−j
(2)
j=1
where i is the index of regions; t – time lag; p, q – number of lag values; α— scalar; VRP – gross regional product (million UAH); ORT – turnover of retail trade (million UAH); ORPr – volume of sold industrial products (million UAH); KI – capital investments (million UAH); EXPORT – export (million USD). After estimating the parameters of the model, we get the following results, where not all variables are statistically significant. This is evidenced by the p- value. Therefore, we can get rid of the model EXPORT(-1), EXPORT(-2), Kl, Kl (-1), ORT and ORT(-2). Estimates of the new model are presented in the Table 6. It is obvious that all variables are statistically significant because the p - value is less than 0.05. Table 6. Results of estimation of model parameters Variable
Coefficients
Standard error
t-statistics
Prob
C
-5625.650
1256.985
-4.475511
0.0000
VRP ( - 1)
0.884471
0.048593
18.20159
0.0000
VRP (- 2)
0.229284
0.028394
8.075005
0.0000
EXPORT
2.656496
0.493614
5.381723
0.0000
Cl ( - 2)
-0.767422
0.216006
-3.552778
0.0005
LOCATION ( - 1)
-0.239889
0.049652
-4.831429
0.0000
ORPR
0.212883
0.014747
14.43527
0.0000
ORPR ( - 1)
-0.237092
0.024662
-9.613807
0.0000
ORPR ( - 2)
0.082794
0.023641
3.502198
0.0006
224
L. Zomchak et al.
As in the case of paired regressions, the standard error characterizes the dispersion of the actual values of the resulting variable around the theoretical ones. We can claim that for each variable from Table 7, this value is quite small and close to 0. Then the model will look like this (3): VRPit = −5625.65 + 0.884471∗VRPi,t−1 + 0.229284∗VRPi,t−2 + 2.656496∗ EXPORTit − 0.767422∗KIi,t−2 − 0.239889∗ORTi,t−1 + 0.212883∗ORPRit − 0.237092∗ORPRi,t−1 + 0.082794∗ORPRi,t−2
(3)
Figure 1 presents actual, calculated values of gross regional product and confidence intervals. If comparing the results of the values calculated by the panel autoregression model and the actual values, for example, for 2020, one can see that the model is quite accurate, and the difference in values is insignificant (Fig. 2). If we talk about the evaluation of quality criteria of models of longitudinal data, it is somewhat different from classical multivariate analysis. At the same time, the values of the coefficient of determination are in the interval [0, 1] regardless of which evaluation method was used to obtain the calculated (theoretical) values. After the calculation, get the following values (Table 7). Table 7. Evaluation of the quality criterion of logit data models Mean square deviation
0.997531
d-statistics
2.230461
The relation of determination
0.996968
F-statistics
1772.28
Prob
0.0000
So, according to the results from Table 7, we can conclude about the adequacy of the proposed model. The root mean square deviation becomes rather small. This means that the random variables are normally distributed. According to the Durbin-Watson test (d-statistic), it follows that the autocorrelation is uncertain, because the d-statistic is greater than 2 and falls into the interval [2.1;3,07]. That is, we cannot assert the presence or absence of autocorrelation. The determination ratio becomes almost unity. This gives an understanding that such a model almost completely describes the variation of the variables. So, on average across the regions of Ukraine, 99.69% of changes in the gross regional product are explained by changes in the gross product itself, the volume of capital investments, realized industrial products and exports, as well as retail trade turnover.
Regional Economic Development Indicators Analysis and Forecasting
225
Fig. 1. Actual and calculated GRP values
500000 400000 300000 200000 0
Cherkasy Chernihiv Chernivtsi Dnipropetrovsk Donetsk Ivano-Frankivsk Kharkiv Kherson Khmelnytsky Kirovohrad Kyiv Luhansk Lviv Mykolayiv Odesa Poltava Rivne Sumy Ternopil Transcarpathian Vinnytsia Volyn Zaporizhzhia Zhytomyr
100000
VRP
VRPpr
Fig. 2. Comparison of actual and calculated GRP for 2020
6 Summary and Conclusion Therefore, with the help of the analysis of the economic development of the regions, a number of important factors of influence on the regional economic growth were singled out, namely on such an indicator as the gross regional product, in particular: turnover of retail trade; volume of sold industrial products; capital Investments; export volumes. After conducting a number of tests (causality test and series stationarity test, Darbin -Wue- Hausman test and Wald test), the model was specified as a panel autoregression
226
L. Zomchak et al.
model with fixed cross-sectional effects. For such a model, it was necessary to apply the generalized least squares method (panel EGLS), which is based on the assumed inequality of variance and therefore provides the opportunity to obtain the best linear estimates. After estimating the unknown parameters of the model, namely the coefficients of the scalar and the coefficients of the factor variables (VRP(-1), VRP(-2), EXPORT, Kl (-2), ORT(-1), ORPR, ORPR(-1), ORPR (-2)), get the following values, respectively – 5625.65, 0.0885, 0.0229, 2.657, -0.767, -0.239, 0.213, -0.237, 0.083. That is, if on the assumption that one parameter is variable and the others are equal to zero, then a change in the corresponding factor by 1% will lead to a change in the gross regional product by 0.0885, 0.0229, 2.657, -0.767, -0.239, 0.213, -0.237, 0.083, respectively. In general, the quality assessment of the built model gives good results. We can definitely state that the variation of the gross regional product is almost completely explained by the model, because the determination ratio is close to unity. Based on this, it is obvious that on average in the regions of Ukraine, 99.69% of changes in the gross regional product are explained by changes in the values of the gross product itself, the volumes of capital investments, realized industrial products and exports, as well as the turnover of retail trade, with a corresponding lag. And the calculated value of the F-statistic exceeds the empirical value (found according to the Fisher distribution table), that is, we can claim that the model adequately describes such a dependence. In this study, where the object is the regions of Ukraine, with the help of models on panel data, not only the behavior of the gross regional product in different regions of Ukraine, but also its dynamics over time was investigated. In general, the proposed model of economic development of regions is suitable and can be used for forecasting. It is worth remembering that the concept of “economic growth” is very complex and multifaceted, it requires a deeper study, coverage and analysis of a greater number of factors affecting it, which will result in a better quality of forecasting at the regional level. Further improvements to the approach to modeling the economic development of the regions of Ukraine on panel data can also be in the direction of using other types of models, in particular, structural equations on panel data instead of theoretical vector autoregressive equations and justification dependencies between the resulting and factor variables. Of course, Russian armed aggression against Ukraine will make adjustments to the development of the economy of the country and its regions in the future. Therefore, the model characterizes the pre-war situation and may inadequately describe the development of economic processes after Ukraine’s victory in the war.
References 1. Coccia, M.: An introduction to theories of national and regional economic development. Turkish Economic Review 5(4), 350–358 (2019)
Regional Economic Development Indicators Analysis and Forecasting
227
2. Malecki, E.J.: Entrepreneurs, networks, and economic development: A review of recent research. Reflections and extensions on key papers of the first twenty-five years of advances (Advances in Entrepreneurship, Firm Emergence and Growth) [M]. Emerald Publishing Limited, Bingley, 71–116 (2018) 3. Gibbs, D., O’Neill, K.: Future green economies and regional development: a research agenda. Transitions in Regional Economic Development, 287–309 (2018) 4. Smol, M., Kulczycka, J., Avdiushchenko, A.: Circular economy indicators in relation to eco-innovation in European regions. Clean Technol. Environ. Policy 19(3), 669–678 (2017) 5. Aliyev, A.G.: Technologies ensuring the sustainability of information security of the formation of the digital economy and their perspective development directions. Int J Inf Eng Electron Business (IJIEEB) 14(5), 1–14 (2022) 6. Amin, A.: An institutionalist perspective on regional economic development. Economy, 59–72 (2017) 7. Treyz, G.I.: Regional economic modeling: a systematic approach to economic forecasting and policy analysis. Springer Science & Business Media (2013) 8. Buchholz, M., Bathelt, H.: Models of regional economic development: Illustrations using US data. Zeitschrift für Wirtschaftsgeographie 65(1), 28–42 (2021) 9. Xin, J., Lai, W.: Research on regional economic development of guangxi based on multiple statistic analysis. Annals of Mathematical Modeling 13(1), 12–21 (2022) 10. Guangyu, T.: Analysis and Decision-making of Regional Economic Vitality and Its Influencing Factors. The Frontiers of Society, Science and Technology 2(14) (2020) 11. Hu, Y., Wang, Z.: Support vector machine-based nonlinear model and its application in regional economic forecasting. Wireless Communications and Mobile Computing (2022) 12. Yu, W., Huafeng, W.: Quantitative analysis of regional economic indicators prediction based on grey relevance degree and fuzzy mathematical model. Journal of Intelligent & Fuzzy Systems 37(1), 467–480 (2019) 13. Matviychuk, A., Lukianenko, O., Miroshnychenko, I.: Neuro-fuzzy model of country’s investment potential assessment. Fuzzy economic review 24(2), 65–88 (2019) 14. Xu, X., Zeng, Z.: Analysis of regional economic evaluation based on machine learning. J. Intell. Fuzzy Sys. 40(4), 7543–7553 (2022) 15. Chagovets, L., Chahovets, V., Chernova, N.: Machine learning methods applications for estimating unevenness level of regional development. Data-Centric Business and Applications. Springer, Cham., 115–139 (2020) 16. Li, Z., Cheng, J., Wu, Q.: Analyzing regional economic development patterns in a fast developing province of China through geographically weighted principal component analysis. Lett. Spat. Resour. Sci. 9(3), 233–245 (2016) 17. Zhong, Z., He, L.: Macro-regional economic structural change driven by micro-founded technological innovation diffusion: an agent-based computational economic modeling approach. Comput. Econ. 59(2), 471–525 (2022) 18. Emmanuel, H.G., Tzeremes, N.G.: Measuring regional economic efficiency: the case of Greek prefectures. Ann. Reg. Sci. 45(3), 603–632 (2010) 19. Santos, L.D., Vieira, A.C.: Tourism and regional development: a spatial econometric model for Portugal at municipal level. Port. Econ. J. 19(3), 285–299 (2020) 20. Shen, Y.: The interactive effect of science and technology finance development and regional economic growth in the yangtze river economic belt: analysis based on panel vector autoregressive (PVAR) model of interprovincial data. Journal of Physics: Conference Series 1624(2), 022026 (2020, October) 21. Márquez, M.A., Ramajo, J., Hewings, G.J.: Assessing regional economic performance: regional competition in Spain under a spatial vector autoregressive approach. Geography, institutions and regional economic performance. Springer, Berlin, Heidelberg, pp. 305–330 (2013)
228
L. Zomchak et al.
22. Zomchak, L., Lapinkova, A.: Key interest rate as a central banks tool of the monetary policy influence on inflation: the case of Ukraine. Advances in Intelligent Systems, Computer Science and Digital Economics IV, pp. 369–379. Springer Nature Switzerland, Cham (2023) 23. Jiang, X., He, X., Zhang, L., Qin, H., Shao, F.: Multimodal transportation infrastructure investment and regional economic development: a structural equation modeling empirical analysis in China from 1986 to 2011. Transp. Policy 54, 43–52 (2017) 24. Allo, A.G., Dwiputri, I.N., Maspaitella, M.: The impact of electricity investment on inter-regional economic development in Indonesia: An Inter-Regional Input-Output (IRIO) approach. Journal of Socioeconomics and Development 5(1), 1–12 (2022) 25. Chen, L., Yu, N.N., Su, Y.S.: The application of game theory in the Hercynian economic development. International Conference on Management Science & Engineering 21th Annual Conference Proceedings, pp. 794–799 (2014) 26. Dutta, P.K., Mishra, O.P., Naskar, M.K.: A method for post-hazard assessment through topography analysis using regional segmentation for multi-temporal satellite imagery: a case study of 2011 Tohuku earthquake region. Int. J. Ima. Graph. Sig. Proc. 5(10), 63 (2013) 27. Alhosani, N.: Mapping urban expansion due to special economic zones in the united arab emirates using landsat archival data (Case Study Dubai). Int. J. Ima. Graph. Sign. Proc. 9(4), 22 (2017) 28. Xu, F., Chen, Z., Li, C., Xu, C., Lu, J., Ou, Y.: A Study on sustainable development of grain production coping with regional drought in China. Int. J. Modern Edu. Comp. Sci. 12, 80–86 (2011) 29. Koroliuk, Y., Hryhorenko, V.: ANN model of border regions development: approach of closed systems. Int. J. Intell. Sys. Appli. 11(9), 1 (2019) 30. Izonin, I., et al.: Stacking-based GRNN-SGTM ensemble model for prediction tasks. Proceedings of the DASA, pp. 60–66 (2020) 31. Tkachenko, R., et al.: Piecewise-linear approach for medical insurance costs prediction using SGTM neural-like structure. CEUR Workshop Proceedings 2255, 170–179 (2018) 32. Lytvynenko, V., Kryvoruchko, O., Lurie, I., Savina, N., Naumov, O., Voronenko, M.: Comparative studies of self-organizing algorithms for forecasting economic parameters. Int. J. Modern Edu. Comp. Sci. 12(6), 1–15 (2020)
An Enhanced Performance of Minimum Variance Distortionless Response Beamformer Based on Spectral Mask Quan Trong The1(B)
and Sergey Perelygin2
1 Digital Agriculture Cooperative, Hanoi, Vietnam 2 Faculty of Media Technologies, State University of Film and Television St Petersburg, Saint,
Petersburg, Russian Federation
Abstract. Minimum Variance Distortionless Response (MVDR) is one of the most attractive techniques commonly used in microphone array beamforming for speech enhancement applications. The conventional MVDR beamformer is very sensitive to the direction-of-arrival of useful signal and microphone gains. Several research has successfully made MVDR beamformer more robust in practice scenario, where exist several types of noise, transport vehicle, third-party speaker, teleconference, communication. MVDR-based equipment is already installed in numerous audio devices: hearing aid, surveillance, telephone, dialogue customer. Nevertheless, MVDR performance is often corrupted in complex surrounding environments, due to imperfect propagation of sound source, phase error or imprecise of the direction-of-arrival of desired speaker. In this paper, the author introduced a solution for enhancing performance of MVDR beamformer in noisy scenario. The method uses a spectral mask, which is based on Differential Microphone Array and an additive function of a priori SNR. Experimental results on real recoding microphone array received data show an increase of the signal-to-noise ratio (SNR) from 5.5 (dB) to 8.0 (dB), and reduce speech distortion to 3.5 (dB) in comparison of conventional MVDR and the suggested method. The effectiveness of the proposed technique shows us that it can be integrated into multi-microphone system for complicated further speech processing. Keywords: Microphone array · Minimum variance distortionless response · Speech enhancement · Noise reduction · Spectral mask
1 Instruction Speech corrupted or degraded performance is a complex task in a speech-based system due to interferences and surrounding noise. In a few decades, the demand of speech enhancement for clean - speech use in teleconference system, audio device, hearing aid, communication system has become a vital part. The effectiveness or capability of these systems depends on the quality and intelligibility of processed signal and the amount of suppressed noise from mixture of noisy signal. The purpose of speech enhancement is obtaining desired original signal and alleviating all negative effects caused by unwanted © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 229–238, 2023. https://doi.org/10.1007/978-3-031-36115-9_22
230
Q. T. The and S. Perelygin
noise in hardy noisy environments. Facing the difficulty for people to understand the speech and words in the situation, where the existence of various types of noise corrupt speech quality, noise reduction is demanded to help avoid this drawback (Fig. 1).
Fig. 1. Separation of speech source is always the most complex task in speech enhancement.
Many single-channel algorithms usually use frequency-domain to process signal. Boll presented spectral subtraction technique, which concerned subtracting spectral noise component from mixture of noisy spectrum. The approach, which is based on statistical model, was introduced in Ephraim and Malah have been evaluated to suppress background at high SNR environment. Method OM-LSA was confirmed its effectiveness in Cohen and Berdugo’s work. However, this approach is too sensitive with dependencies on the type of background noise, an imprecise estimation of spectral noise component often leads to distortion of desired speech signal and introduce musical noise at the output. Theoretically, multi-microphone [1–5, 12–14] speech enhancement direction, which allows implementing more less distortion when compared to single-channel methods, has been harshly developed (Fig. 2).
Fig. 2. The use of microphone array beamforming.
Microphone array (MA) beamforming - based method, which has been intended to be the most widely used tool for study the spatial characteristics of acoustic environment.
An Enhanced Performance of Minimum Variance
231
Beamforming algorithm steer MA’s response to a preferred direction of desired source on the interesting region toward target speech, and they provide a spatial distributed beampattern with a certain level pressure. The amount of spectral speech component can be computed by using a priori information of direction of arrival of useful signal, distributed geometry of MA to capture desired signal, an appropriate signal processing algorithm. Spatial filtering has been vital component in this approach, which helps to boot the useful target speech at certain direction while attenuating annoyed interference from other directions. In real-life situations, the MA’s working can be adjusted by using an appropriate beamforming. The first type of beamformer, which has a static frequency response and static constant coefficients, is named fixed beamformer. The simplest one is a delay-and-sum beamformer. The most popular widely used adaptive beamformer is MVDR [6–10]. Adaptive beamformer allow utilize a certain number of microphones, while efficiently mitigating surrounding noise and saving desired speech. Based on these factors, dual-microphone system (MA2) is considered a famous solution to implement numerous signal processing algorithms for improving the quality and intelligibility. There are several methods to improve the performance of MVDR beamformer (Fig. 3).
Fig. 3. An implementation of microphone array in frequency domain
Relative transfer function (RTFs) is one of the most essential quantities required for accomplishing speech enhancement by using MVDR beamformer. This vector RTF can be determined a priori or based on assumptions about talker location, acoustic environment, microphone characteristics, position. A precise RTF makes performing MVDR beamformer more robustness to achieve high diversity, high noise reduction. Diagonal loading becomes a commonly used technique to enhance the robustness of MVDR beamformer. This approach is derived from criteria of the Euclidean norm of the coefficients vector and an additional quadratic constraint. Diagonal loading also alleviates some complicated tasks of using the covariance matrix of microphone array signals and better adjust the peak of sidelobe. However, it is not exactly how to choose an appropriate parameter based on the uncertainty steering vector. The subspace-based adaptive beamforming technology need the information of the initial noise covariance matrix. Hence, the performance is often degraded in presence of imprecise knowledge of steering vector and matrix.
232
Q. T. The and S. Perelygin
In low environments with low SNR, a real-valued unitary transformation MVDR algorithm is used to improve the performance of beamformer. Otherwise, the scholar proposed root MVDR algorithm, which overcomes spectral search and decreases the computational complexity. The traditional adaptive beamformer tends to work on determined acoustical scenario and cannot outperform good performance in the non-stationary or complex environment. In this paper, the author suggested the use of a spectral mask, which based on the principle of Differential Microphone Array (DMA) and a function of a priori SNR for enhancing the performance of MVDR beamformer. Many unwanted reasons: the microphone mismatch, the difference of received data, phase error, the influence of interference or imperfect environment that significantly attenuates target directional useful signal. Illustrated experiments were verified and show the effectiveness of suggested method for speech enhancement while mitigating all non-target directional signals and keeping desired speech source. This contribution is organized as follows: The next section introduces the model working of MVDR beamformer. In section III, the spectral mask, which is based on DMA and an additional function of a priori SNR, is determined and used as pre-processing step to block the target directional speech component. Experiment with DMA2 in real environment was shown in Section IV, a comparison between the conventional MVDR beamformer (conMVDR) and suggested method (spmMVDR) show the improvement of suggested method in term of the signal-to-noise ratio (SNR). Concluding remark and future work are described in Section V.
2 Mvdr Beamformer MVDR beamformer is one of the most efficient microphone array beamforming to obtain the desired target speaker by using the priori information of the direction of arrival (DOA) and observed matrix of microphone array signal without speech distortion. In this contribution, the author will illustrate the scheme of MVDR beamformer in dualmicrophone system (DMA2), due to its simplicity and convenience to implement almost digital signal processing.
Fig. 4. The scheme of MVDR filter on dual-microphone system.
An Enhanced Performance of Minimum Variance
233
Consider DMA2 depicted in Fig. 4. The received microphone array signals in the frequency domain X1 (f , k), X2 (f , k) are written as: X1 (f , k) = S(f , k)ejs + V1 (f , k)
(1)
X2 (f , k) = S(f , k)e−js + V2 (f , k)
(2)
where f , k: the index of frequency and frame respectively. S(f , k): the original target direction speech source, V1 (f , k), V2 (f , k) are the obtained noise at DMA2. s = π f τ0 cosθs , θs : : the direction-of-arrival (DOA) of interest signal relative to MA, τ0 = d /c, d : the distance between two microphones, c = 343(m/s): the speed of sound in the air, τ0 : the time delay. T T If we denote, X(f , k) = X1 (f , k) X2 (f , k) Ds (f , θs ) = ejs e−js , T V(f , k) = V1 (f , k) V2 (f , k) so the Eq. (1–2) can be presented as following: X(f , k) = S(f , k)Ds (f , θs ) + V(f , k)
(3)
where Ds (f , θs ): steering vector has the information of DOA, which plays a significant role in almost beamforming algorithm. The optimum constraint of minimum of the power of noise while preserving the original of target speech, leads to the optimum problem, which can be described the following equation: min H (4) W (f , k)Φ VV (f , k)W(f , k)st W H (f , k)Ds (f , θs ) = 1 W where Φ VV (f , k) = E V H (f , k)V(f , k) is the covariance matrix of noise. Using Lagrange method, (4) gives us the optimum weight vector as: W(f , k) =
Φ −1 VV (f , k)Ds (f , θs )
−1 DH s (f , θs )Φ VV (f , k)Ds (f , θs )
(5)
In almost speech applications, the information of noise is not always available, so the covariance matrix of observed microphone array signals Φ XX (f , k) is used instead of. Therefore, the final solution for MVDR beamformer can be yielded that: W(f , k) =
Φ −1 XX (f , k)Ds (f , θs )
−1 DH s (f , θs )Φ XX (f , k)Ds (f , θs )
(6)
where:
∗ E X1 (f , k)X1 (f , k) E X1∗ (f , k)X2 (f , k) Φ XX (f , k) = E X2∗ (f , k)X1 (f , k) E X2∗ (f , k)X2 (f , k) The power spectral density (PSD), E Xi (f , k)Xj∗ (f , k) = PXi Xj (f , k) is determined
as: PXi Xj (f , k) = αPXi Xj (f , k − 1) + (1 − α)Xi∗ (f , k)Xj (f , k)∀i, j ∈ {1, 2} where α is the smoothing parameter and often in range {0..1}.
(7)
234
Q. T. The and S. Perelygin
3 The Proposed Method The author’s ideal of improving the speech enhancement is using a spectral mask for preprocessing alleviate the speech component in the microphone signals X1 (f , k), X2 (f , k) as: Xˆ 1 (f , k) = SM (f , k) × X1 (f , k)
(8)
Xˆ 2 (f , k) = SM (f , k) × X2 (f , k)
(9)
where SM (f , k) is the spectral mask. Xˆ 1 (f , k), Xˆ 2 (f , k) is then processed by MVDR beamformer to extract desired speech signal. The SM (f , k) is determined with the information of beampattern, which null-steering towards the direction of signal, and a function of the signal-to-noise ratio (SNR). The differential microphone array (DIF) allows forming the null-beampattern toward the direction of speaker, θs . X1 (f , k) − X2 (f , k)e−jωτ 2
ωτ0 τ = jS(f , k)sin cosθ + 2 τ0
YDIF (f , k) =
where τ = τ0 cosθs . The obtained null-beampattern is calculated by:
YDIF (f , k) = sin ωτ0 cosθ + τ B(f , θ ) = S(f , k) 2 τ0
(10) (11)
(12)
A method for calculating the covariance of speech σs2 (f , k) is expressed as: σs2 (f , k) =
1 −1 DH s (f , θs )Φ XX (f , k)Ds (f , θs )
(13)
Therefore, noise covariance σn2 (f , k) is determined by: σn2 (f , k) =
PX1 X1 (f , k) + PX2 X2 (f , k) − σs2 (f , k) 2
(14)
σs2 (f , k) σn2 (f , k)
(15)
The temporal SNR(f , k): SNR(f , k) =
The proposed author’s spectral mask is derived as: YDIF (f , k) 1 × SM (f , k) = S(f , k) 1 + SNR(f , k)
(16)
The next will perform an experiment in a real acoustic environment with a speaker, a DMA2 in presence of background noise to verify the effectiveness of suggested spectral mask.
An Enhanced Performance of Minimum Variance
235
4 Experiments The scheme of the experiment is illustrated in Fig. 5. A target speaker stands at the direction θs = 900 (deg) relative to the axis DMA2, which located in a room conference in presence of background noise, interference and third-party speaker. For further signal processing, the author use NFFT = 512, overlap 50% , the smoothing parameter α = 0.1, the sampling frequency Fs = 16kHz, Hamming window, the distance between two microphones d = 5(cm). The experiment is aiming to compare performance of the suggested method (spmMVDR) and the conventional MVDR (conMVDR). An objective measurement SNR [11] is used for estimating the speech quality.
Fig. 5. The scheme of illustrated experiment
Fig. 6. The waveform of the received microphone array signal
The original microphone signal is presented in Fig. 6. Figure 7 shows the received output signal by using the conventional MVDR beamformer. With the proposed spectral mask, the obtained processed signal as illustrated in Fig. 8. From Figs. 7 and 8, the suggested method has the ability of preserving the original speech component of target speaker and reducing the speech distortion to 3.5 (dB) as in Fig. 9. In Table 1, the author compared the speech quality in terms of the signal-to-noise ratio (SNR). SpmMVDR increases the SNR from 5.5 to 8.0 (dB). From the experimental results, the proposed technique was verified in a real acoustic experiment to reduce speech distortion and enhance speech quality. The proposed spectral mask has removed the desired target speech component in microphone array signals
236
Q. T. The and S. Perelygin
Fig. 7. The output signal by conMVDR
Fig. 8. The obtained signal by the proposed method (spmMVDR)
Fig. 9. The energy of microphone array signal, processed signal by conMVDR and spmMVDR
Table 1. The signal - to - noise ratio (dB) Method Estimation
Microphone array signal
conMVDR
spmMVDR
NIST STNR
5.2
21.0
29.0
WADA SNR
3.7
18.4
24.9
and improved the performance of MVDR beamformer. Consequently, this technique can be applied into other MA digital signal processing, such as: Generalized Sidelobe Canceller, Differential Microphone Array, Linear Constrained Minimum Variance.
An Enhanced Performance of Minimum Variance
237
5 Conclusion The evaluation of original MVDR beamformer seriously degraded in the presence of microphone mismatches, phase error, the different microphone sensitivities, the error between the assumed and actual DOA. So, in the speech enhancement applications, robustness of the beamformer is the most necessary task. In this article, the author proposed a spectral mask as a pre-processing step for signal processing to enhance the quality of output signal. The experiments have confirmed the improvement of this approach in real scenarios in the presence of noise and interference. The performance showed better reduce speech distortion and the proposed method is suitable for acoustic equipment, mobile device, teleconference, hearing aids or other compact acquisition distant in adverse noise environments. In the future, the author plans to combine the properties of environment to integrate into spectral mask for achieving more noise reduction and speech quality.
References 1. Benesty, J., Chen, J., Pan, C.: Fundaments of Differential Beamforming. p. 122. Springer (2016). https://doi.org/10.1007/978-981-10-1046-0 2. Microphone Arrays. In: Brandstein, M., Ward, D. (eds.) Springer-Verlag, Heidelberg, Germany. XVIII, p. 398 (2001). https://doi.org/10.1007/978-3-66204619-7 3. Hoshuyama, O., Sugiyama, A., Hirano, A.: A robust adaptive beamformer for microphone arrays with a blocking matrix using constrained adaptive filters. IEEE Trans. Signal Process. 47, 2677–2684 (1999) 4. Benesty, J., Cohen, I., Chen, J.: Fundamentals of Signal Enhancement and Array Signal Processing, p. 440. Wiley, IEEE Press (2017) 5. Benesty, J., Chen, J., Huang, Y.: Microphone Array Signal Processing, p. 240. Springer-Verlag, Berlin, Germany (2008). https://doi.org/10.1007/978-3-54078612-2 6. Lockwood, M.E., et al.: Performance of time- and frequency domain binaural beamformers based on recorded signals from real rooms. The Journal of the Acoustical Society of America 115, 379 (2004). https://doi.org/10.1121/1.1624064 7. Zhu, Y., Fu, J., Xu, X., Ye, Z.: Modified complementary joint sparse representations: a novel post-filtering to MVDR beamforming. In: Proc. 2019 IEEE International Workshop on Signal Processing Systems (SiPS), pp. 1-6. Nanjing, China (2019). https://doi.org/10.1109/SiPS47 522.2019.9020522 8. Fischer, D., Doclo, S., Habets, E.A.P., Gerkmann, T.: Combined Single-Microphone Wiener and MVDR Filtering based on Speech Interframe Correlations and Speech Presence Probability. Speech Communication; 12. ITG Symposium, pp. 1–5. Paderborn, Germany (2016) 9. Sun, Z., Li, Y., Jiang, H., Chen, F., Wang, Z.: A MVDR-MWF combined algorithm for binaural hearing aid system. Proc. 2018 IEEE Biomedical Circuits and Systems Conference (BioCAS), pp. 1-4. Cleveland, OH (2018). https://doi.org/10.1109/BIOCAS.2018.8584798 10. Fischer, D., Doclo, S.: Subspace-based speech correlation vector estimation for singlemicrophone multi-frame MVDR filtering. In: Proc. ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 856–860. Barcelona, Spain (2020). https://doi.org/10.1109/ICASSP40776.2020.9052934 11. https://labrosa.ee.columbia.edu/projects/snreval/
238
Q. T. The and S. Perelygin
12. Bhatia, V., Whig, P.: Performance analysis of multi-functional bot system design using microcontroller. I.J. Intelligent Systems and Applications 02, 69–75 (2014). https://doi.org/10.5815/ ijisa.2014.02.09 13. Shcherbyna, O., Zaliskyi, M., Kozhokhina, O., Yanovsky, F.: Prospect for using lowelement adaptive antenna systems for radio monitoring stations. I. J. Computer Network and Information Security 5, 1–17 (2021). https://doi.org/10.5815/ijcnis.2021.05.01 14. Mandal, S.: Linear antenna array pattern synthesis using elephant swarm water search algorithm. I.J. Information Engineering and Electronic Business 2, 10–20 (2019). https://doi.org/ 10.5815/ijieeb.2019.02.02
Modelling Smart Grid Instability Against Cyber Attacks in SCADA System Networks John E. Efiong1(B) , Bodunde O. Akinyemi1 , Emmanuel A. Olajubu1 , Isa A. Ibrahim2 , Ganiyu A. Aderounmu1 , and Jules Degila3 1 Department of Computer Science and Engineering, Obafemi Awolowo University, Ile-Ife,
Nigeria {bakinyemi,emmolajubu,gaderoun}@oauife.edu.ng 2 Department of Cybersecurity, Federal University of Technology, Owerri, Nigeria 3 Department of Computer Science, Institute of Mathematics and Physics, University of Abomey, Calavi, Benin [email protected]
Abstract. The SCADA systems in the Smart Grid Network (SGN) are increasingly facing cyber threats and divers attacks due to their known proprietary vulnerabilities, most often leading to power instability and cascading failures in the Grid. This paper associates the SGN’s instability with cyberattacks and models intrusions into SCADA systems using the Decentral Smart Grid Control (DSGC) model on the Stability dataset. The Time-Synchronization Attack was modelled with its effects on Transmission Line Fault Detection, Voltage Stability Monitoring and Event Locating. Classification and prediction were done on WEKA using a number of Machine Learning Algorithms and performances evaluated. Random Forest (RF) classifier was found to be more promising in all the benefit metrics yielding 92.0%, 91.9%, 92.0%, and 91.9% for Accuracy, Precision, Recall and F-measure, respectively. An ensemble classifier – Bagging, performed significantly well after the RF with 90.0%, 89.9%, 90.0% and 89.9%. The ROC Area under Curve also showed RF having 97% and Bagging 96.7%. 100 trees were constructed for the RF while considering 4 random features, and the out-of-bag error was 0.0084. RF had a TPR of 92.0%, which is the highest and an FPR of 0.11. These results provide insights into identifying appropriate ML techniques and models for the Grid’s instability problem vis-à-vis modelling attacks. This is expected to help in effective security planning, threats detection, attack mitigation and risk management. Keywords: SCADA Systems · Cyberattacks · Smart Grid Stability · Decentral Smart Grid Control · Cyber-Physical Power Systems
1 Introduction Stability is one of the essential characteristics of the Cyber-Physical Power System (CPPS), also identified as the Smart Grid, as a heterogeneous, multi-faceted system [1]. This is in addition to the robustness, efficiency, reliability, resource management © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 239–250, 2023. https://doi.org/10.1007/978-3-031-36115-9_23
240
J. E. Efiong et al.
and load-balancing that the CPPS is expected to provide. Smart Grid stability entails maintaining a balance between the energy generated and the energy consumed [2]. It is a state of equilibrium in the business chain. The Grid ensures this by responding to volatility in voltage and frequency disturbances. In the Smart Grid, the Phasor Measurement Unit (PMU) is a popular tool for monitoring voltage stability and analysing oscillation stability. The PMU also allows power state estimation, monitoring and control of the Wide Area Network, Volt-Amps Reactive (VAR) optimisation, blackout analysis, real-time electricity costing, and fault detection in the transmission lines. Targeted cyberattacks may result in putting the PMU off its primary functions, thereby resulting in power instability in the electric Grid. Additionally, one of the main goals of a cyberattack on the Supervisory Control and Data Acquisition (SCADA) systems managing the Smart Grids is to interfere with the control process of the circuit breakers, thereby forcing the circuit breakers to respond to external commands other than what the control engineer would be issuing. This could be achieved by the attacker through a command injection attack, data injection attack and time-synchronisation [29] attack using primarily the interception and injection attack strategies. When circuit breakers are forced open, it results in a denial of service (instability), which may have cascading effects on several processes depending on which circuit is broken. Power stability is equivalent to the availability of resources in the grid; thus, an attack tends to cause instability or a total blackout. For instance, an attack can cause a delay in the release of the status information of the power generator to the control centre for a prompt response and necessary actions [3, 4]. Furthermore, it can trip circuit breakers at will and cripple the grid’s load-balancing capabilities. Modelling Smart Grid instabilities against cyberattacks help to examine the risks associated with cyberattacks and their propensity to affect not only power stability but the safety of personnel and security of the components of the SCADA network. According to [1], such modelling benefits power system engineers in four ways: First, they will respond to attacks based on their ability to point out the problem with specific components. Second, mitigation strategies can be developed to prevent future occurrences. Third, it will help in regular evaluation of the security posture of the organisation. Fourth, it will help the engineer to engage in a more formidable design and development of resilient Cyber-physical Systems for the Power Sector. It is important to note that power control engineers do not necessarily provide security for power assets. This role is expected to be delivered by a trained industrial cybersecurity professional in the Smart Grid. Thus, such individuals and researchers would appreciate the threat-modelling results in the grid. Existing body of knowledge that model Smart Grid instabilities focus rather on the mechanical prediction of instability using a plethora of techniques. There is an acute shortage of works that associates instability with the possibility of cyberattacks. This paper understudies that subtle operations of attacks and their propensity to distort the stability of the Grid. Typically, in this paper, we model the Time-synchronization Attack (TSA) and implement Machine Learning algorithms for attack identification as a precursor for instability in the Smart Grid Network. The primary objective of this paper is to associate power instability with cyberattacks and provide an ML-based model for intrusion detection in the Smart Grid Network. The rest of the paper is organised
Modelling Smart Grid Instability Against Cyber Attacks
241
as follows: Section II discusses related works. The Threat/Attack modeling technique, dataset and methods used are covered in Section III. The experimental setup is presented in Section IV, and the presentation and discussion of results are done in Section V. The last Section concludes the paper.
2 Related Works Efforts have been made to model attacks on the Smart Grid network. Studies by [5] modelled Grid Stability with Decentral Smart Grid Control (DSGC) based on frequency by measuring the cycles per second of the alternate current. However, this was only applicable at the power generation phase of the Grid process. Reference [5] adopted the Decision Tree (CART) algorithm and space-filling designs to implement the DSGC model. Reference [6] investigated how the PMU could be used to model a Time-Synchronised attack in the Smart Grid network with the aim of revealing tampered data, possible effects on voltage stability, Automatic Generation Control (AGC) [7] and power system frequency control [8]. Instability was modelled using the delay-dependent stability analysis technique by [9] and [10] for the eigenvalue computation of the Grid Network. A directed-graph method with a dynamic system equation was utilised by [11]. The technique was used to compute the status of an individual node in the Grid network. Reference [12] experimented with a sort of distributed denial of service as a strategy to cause transient instability in the grid using the Variable Structure system model. The theory supported a coordinated multi-switch attack capable of causing cascading failures in the Smart Grid through intrusion into the network from a single circuit breaker and extending to other nodes. Recently, an optimal partial feedback attack modelling method that relies on convex relaxation and pontryagin’s maximum principle for instability was proposed by [13] to model data injection and command injection attacks. Here, the adversary modifies control signals to tweak the performance of the master unit and changes its location to avoid being traced. The proposed method revealed the attack locations [13]. Some of the works that model instability against threats and attacks in the CyberPhysical systems of the Power sector have indicated that a multi-agent-based technique may be appropriate in detecting and spotting threats with high affinity for causing instability in the Smart Grid. For instance, in applying the multi-agent method, [14] leveraged the unique characteristics of the power system’s separate physical and cyber components, which helped classify cyberattacks appropriately and separated them from the material faults of the system. Similarly, [15] proposed a control framework based on the multi-agent perspective to reduce transient instability in the grid. Also, [16] proposed a distributed averaging-based integral (DAI) controller to solve the grid’s instability attack-facing problem based on time-delay and dynamic communication topology. To leverage control of the DER and reduce transient instability in the Grid, [17] proposed a differential game-theoretic technique designed to model the attacker’s behaviour in the Grid network. Other methods used in literature include linear matrix inequalities and the Lyapunov stability technique [18], Lyapunov-based time-varying multiple delayed systems method [6], time-delayed power system stability analysis by the integral quadratic constraints method [19], realistic delay modelling method [20], etc.
242
J. E. Efiong et al.
The traditional methods of modeling instability on the Smart Grid discussed above do not present formidable tools for identifying, detecting, and mitigating sophisticated threats and attacks. The availability of advanced attack-launching tools and techniques has made the adversary’s job relatively more accessible and penetrative. To address these flaws, [21] proposed an optimized Deep Learning model for predicting Smart Grid instability. Although the DL model yielded an accuracy of 99.6%, the primary focus was on solving the fixed inputs (variables of the equations) of the Decentral Smart Grid Control (DSGC) system. Reference [22] proposed an optimized data-matching learning network-based model for predicting stability of a decentralized Power Grid linking electricity price formulation to Grid frequency. The model was designed for forecasting other causes of instability, excluding cyber threats. A recent study by [23] leveraged Genetic Algorithm to predict instability in the Grid. Using a 14-feature-based dataset extracted from UCI-ML repository, the model achieved an accuracy of 98%. This dataset did not depict the DSGC system and the proposed model was not associated with cyberattack mechanisms in the Grid. Our study employs machine learning techniques to model instability, associates same with cyberattacks and provides insights that can help develop intrusion detection systems and possible prevention in the Smart Grid Network.
3 Datasets and Methods This Section presents the attack-instability modeling, dataset, experimental setup and methods. 3.1 Attack-Instability Modelling Attack-instability modelling can follow a dynamic system-based technique where the physical components of the CPS follow differential equations utilising energy flow, and the cyber components follow differential equations using information flow. It is expected that the cyber components would disturb the physical systems; as such, the stimulant of the states of the generator, typically frequency and angle in the rotor swing equation of the generator, is used to model the perturbations [1]. This study adopted the technique leveraged by [7]. 3.1.1 Time-Synchronization Attack In this study the Time-Synchronization Attack (TSA) is modelled. The TSA is an attack that exploits the wide area monitoring systems (WAMSs). The TSA targets vulnerable networks with dynamic data exchanges at the nodes [24]. The technique is to modify the time synchronization at the endpoints, resulting in cascading failures. In the electric Grid, a TSA attacker can build a vector to control the synchronisation generator to remotely control the Distrito control Distributed Energy Resources (DERs) remotely [30]. This enables the attacker to have lateral access to the system and enforce the instability of the Smart Grid.
Modelling Smart Grid Instability Against Cyber Attacks
243
We modelled the effect of TSA in Transmission Line Fault Detection was using Eq. (1): (A + B)(C + D. exp(jθ ) 1 ln( (1) D = 2γ L (C + D)(A + B. exp(jθ ) where: D is the fault location error concerning TSA; θ is asynchronism of phase angle measurement between the transmitting and receiving endpoints concerning TSA; L is the length of the transmission line, γ is the attenuation constant, while A, B, C, and D are the formulas of the transmitting and receiving voltage and current respectively. We further modelled the effects of the TSA in Voltage Stability Monitoring using Eqs. (2), (3) and (4): ZT = 2 =− Zsh
VS exp(jθS ) − VR exp(jθR ) IS exp(jθS ) − IR exp(jθR )
Vs IR + VR IS 2 IR exp(j2θR ) − IS2 exp(j2θS ) ZL =
∗ (exp j(θS + θR ))
VR exp(jθR ) IR exp(jθR )
(2) (3) (4)
andZ are the T-equivalent parameters for computing Voltage Stability where ZT , Zsh L Monitoring as affected by the TSA [31]. We then modelled the effect of the TSA on Event Locating using Eq. (5):
(xi − xe )2 + (yi − ye )2 − Ve2 (ti − te )2 = 0
(5)
where: ti , i = 1, 2, 3, 4 is the disturbing event arrival time to the ith PMU; (xi , yi ) and (xe , ye ) are the coordinates of the ith PMU and the disturbance event locations. Ve represents the event propagation velocity in the Smart Grid network. 3.2 Decentral Smart Grid Control (DSGC) Model The Decentral Smart Grid Control (DSGC) model is a system that monitors frequency property of the Grid. This implies that an essential component of the Grid that the DSGC model takes due cognizance of is the frequency. This property ensures that there is a strict observance of the state of equilibrium between the power generated and consumed. The model expresses the physical dynamics of electric power generation and how it relates to the consumption of loads [5, 25, 26]. The DSGC model can be expressed mathematically as shown in Eq. 6: d θj d 2 θj + = P − α Kjk sin(θk − θj ) j j dt 2 dt
(6)
k
where: j = index representing the number of grid participant (producer(s) generating power to transmit through the grid and consumers take load from the Grid);
244
J. E. Efiong et al. d 2 θj dt 2
= grid stability indicator (negative indicates grid is unstable; positive indicates grid is stable); Pj = mechanical power produced (e.g. by generator P1 ) or consumed (e.g. by one of several consumers P2 to P4 ) (s−2 ); αj = damping constant related to the power dynamics of the grid; d θj dt Kjk
= change in rotor angle for participant j relative to grid frequency ω; = Coupling strength between grid participant j and k, which is proportional to line capacity (s−2 ); and. θk and θj are rotor angles for grid participants j and k at a specific point in time t. The DSGC model takes into consideration the power Grid synchronization to ensure power stability [27, 28]. This becomes a critical factor which a Time Synchronization attacker leverages to exploit the Grid. 3.3 Dataset The dataset used for experiment was the Electrical Grid Stability Simulated Dataset created by [5] to support the Decentral Smart Grid Control (DSGC) project. The dataset was eventually donated to the University of California (UCI) Machine Learning Repository. The smart grid stability (DSGC) project aimed to examine and make arrangements for both energy generation and potential utilisation by preparing for unsettling influences and vacillations that may occur in the electric business chain. The dataset was generated using a Grid stability reference 4-node star network simulation shown in Fig. 1. The dataset contains 10,000 samples. The original dataset contained 12 features and 2 target classes for classification and prediction. The features were classified into reaction times of the participants (tau1 – tau4), nominal power generated or consumed (p1 – p4) and price elasticity coefficient for each participant on the network (g1 - g4). The reaction time takes real values between 0.5 and 10, where tau1 was assigned the supplier node, and the rest were assigned the consumer nodes. The nominal power was a real value between -2.0 to -0.5 (for p2 – p4); when generated, it was positive, and when consumed, it became negative. For the total power consumed to be equal to the total generated, Eq. (7) must hold: p1(suppliernode) = −(p2 + p3 + p4)
(7)
The elastic price coefficient must also assume real values between 0.05 to 1.00, where g1 was assigned the supplier node and g2 – g4 the consumer nodes, with g representing ‘gamma’. For the dependent variables, 2 were identified as ‘stab’ and ‘stabf’. The former is the maximum real part of the root of the differential equation. If it is positive, the system is said to be linearly unstable; if it is negative, it is linearly stable. The stabf is a binary categorical target class with labels (stable or unstable). Our models were implemented based on the stabf variable. We applied the Principal Component Analysis (PCA) to extract the most important features from the ranked attributes based on the computed eigenvectors. We also used the Ranker Search Method with Chi-Squared Attribute Evaluator (Chi2) and Info Gain Attribute Evaluator (InfoGain). The PCA selected tau1, g1,
Modelling Smart Grid Instability Against Cyber Attacks
245
Fig. 1. Decentral Smart Grid Control 4-node Network (Wood, 2020)
tau3, tau4, g2, g4 and tau2 while both Information Gain and Ch2 selected tau1, tau2, tau4 and tau3 as the best averagely ranked attributes. Although the dataset does not explicitly denote attack patterns, there is a strong association between the instability experienced in Grids during power generation, transmission and distribution leading to a denial of service attacks. The chance of this occurring increases as the Smart Grid is controlled and supervised by SCADA systems which lack security mechanisms to deter intrusions. Hence, the need to look at the scenarios from an intrusion point of view.
4 Experimental Setup The experiment was carried out on the Waikato Environment for Knowledge Analysis (WEKA), a machine learning data mining tool for knowledge synthesis. The tool allows attributes selection, data preprocessing, classification, clustering, association, linking, visualisation and cross-validation. We split the dataset into 75% for training and 25% for testing and validation. We projected the “unstable” class into “Anomaly” and the “stable” into “normal” states of the system. For training and evaluating the dataset for prediction, the following classifiers were used; Five (5) Trees-based classifiers – Random Forest (RF), Random Tree (RT), LADTree, ADTree and J48; Two (2) Rules-based – PART and JRip; One (1) Functionsbased – Logistic; One (1) Bayes-based – Naïve Bayes (NB) and One (1) Meta-based – Bagging. Figure 2 shows the implemented ML classifiers. The classifiers were evaluated using Accuracy, Precision, Recall and F1-Measure. We also examined their respective performances regarding False Positive Rate (FPR) and True Positive Rate (TPR), which
246
J. E. Efiong et al.
are helpful for computing Detection Rate. The Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) were computed and evaluated.
Fig. 2. The Implemented ML Classifiers
5 Results and Discussion A total of 10 Machine Learning algorithms were implemented, and results are shown in Tables 1 and 2. All the models performed reasonably well, outputting more than 80% in all metrics. However, the Random Forest (RF) classifier demonstrated the best results in all the benefit metrics yielding 92.0%, 91.9%, 92.0%, and 91.9% for Accuracy, Precision, Recall and F-measure, respectively. The ensemble classifier closely followed this – Bagging with 90.0%, 89.9%, 90.0% and 89.9%. The last column of Table 1 shows the ROC Area under Curve, with RF having 97% and Bagging 96.7%. The PART and Naïve Bayes had an equal result of 92%, while LADTree and ADTree had 91.0% and 91.9%, respectively. A Random Forest of 100 trees was constructed while considering 4 random features and the out-of-bag error was 0.0084. This shows the efficiency of the RF classifier in modelling instability against Smart Grid attacks. PART generated 65 rules, and JRip generated 51 rules. The J48 tree classifiers had a total of 813 trees with 407 leaves, while RandomTree had a size of 1873 trees. The bagging classifier of the Meta class had 393 trees. The LADTree classifier generated a total of 31 trees, 21 predictor nodes for the Tree size, 16 predictor nodes for Leaves, and expanded nodes of 100 with 517157 instances. ADTree had 31 nodes, and 21 Leaves used predictor nodes. The Confusion Matrices of the classifiers, whose results are represented in Table 2, showed that RF had the True Positive Rate (TPR) of 92.0%, which is the highest and the False Positive Rate (FPR) of 0.11, which is the lowest. The MAE and RMSE of the RF were 0.18 and 0.25, respectively. The Bagging technique produced a TPR of 90.0% and an FPR of 0.13. The MAE and RMSE of Bagging were 0.17 and 0.27, respectively. It should be noted that J48 returned the lowest MAE, 0.14 but with a significantly high RMSE, 0.35.
Modelling Smart Grid Instability Against Cyber Attacks
247
Table 1. Model Classification and Prediction Results Classifiers
Metrics Accuracy
Precision
Recall
F-Measure
ROC Area
PART
85.9%
86.1%
85.9%
85.5%
92.0%
JRip
87.4%
87.3%
87.4%
87.3%
87.5%
Logistic
82.3%
82.1%
82.3%
82.1%
89.9%
NB
83.5%
83.5%
83.5%
83.1%
92.3%
J48
86.7%
86.7%
86.7%
86.7%
86.5%
RF
92.0%
91.9%
92.0%
91.9%
97.9%
RT
83.2%
83.3%
83.2%
83.3%
81.9%
Bagging
90.0%
89.9%
90.0
89.9%
96.7%
LADTree
84.1%
84.0%
84.1%
84.0%
91.0%
ADTree
83.8%
83.8%
83.8%
83.4%
91.9%
Table 2. Model Evaluation Results Classifiers PART
Metrics TPR
FPR
MAE
RMSE
85.9%
0.21
0.16
0.32
JRip
87.4%
0.15
0.18
0.32
Logistic
82.3%
0.22
0.24
0.35
NB
83.5%
0.24
0.29
0.34
J48
86.7%
0.16
0.14
0.35
RF
92.0%
0.11
0.18
0.25
RT
83.2%
0.19
0.17
0.40
Bagging
90.0%
0.13
0.17
0.27
LADTree
84.1%
0.20
0.25
0.34
ADTree
83.8%
0.23
0.31
0.36
6 Conclusion The utility service-providing domain is delicate, and requires adequate care and attention to all eventualities. This study proposes that not all power failure or instability results from mechanical faults or natural occurrences. Targeted attacks at specific components of the Grid, especially the time synchronization system of the PMU can result in serious dangers. Cyberattacks on the Smart Grid can prompt communication delays or service denials and puncture data exchanges between the PMU and the Control Centre or even
248
J. E. Efiong et al.
take control of the Control Systems. This ultimately results in Power instability, technically referred to as ‘denial of service’. Several kinds of attacks exist that target the Smart Grid. We have modelled the Time-Synchronization Attack (TSA), which is a stealthy attack that establishes an attack vector or increases an attack surface such that the attacker takes control of the synchronisation generator, thus remotely controlling the Distributed Energy Resources (DERs) [32]. This attack technique can be catastrophic due to the several other components and processes the DER usually manages. In addition, the TSA has the propensity to open the network for distributed denial of service (DDoS). Modelling attacks in the Smart Grid network helps identify threats and develop mitigation mechanisms. In this paper, we have demonstrated 10 ML algorithms that can help model Smart Grid Instability against Cyberattacks on Cyber-Physical Power Systems and other critical infrastructure. Random Forest classifier of the Trees Methods and Bagging classifier of the Ensemble techniques were found to be more promising and can hence be tuned to provide higher accuracy, precision, recall and F-measure. These results have implications for effective threats-modelling in the Smart Grid network. The idea portrayed in this paper would help in attacks identification, threats classification, prediction, intrusion detection and developing mitigation process. Future studies may consider modelling other kinds of attacks in the network by leveraging deep learning algorithms. Acknowledgment. The Research was supported by funding from the Digital Science and Technology Network (DSTN), France, through the Research Partnership programme with the OAUKnowledge Park (OAK-park), African Centre of Excellence, Obafemi Awolowo University, Ile-Ife, Nigeria.
References 1. Yohanandhan, R.V., Elavarasan, R.M., Manoharan, P., Mihet-Popa, L.: Cyber-Physical Power System (CPPS): A Review on Modeling, Simulation, and Analysis With Cyber Security Applications. IEEE Acesss 8, 151019–151064 (2020) 2. Gopakumar, P., Jaya bharata Reddy, M., Mohanta, D.K.: Letter to the Editor: Stability Concerns in Smart Grid with Emerging Renewable Energy Technologies. Electric Power Components and Systems 42(3–4), 418–425 (2014). https://doi.org/10.1080/15325008.2013. 866182 3. Dong, C., et al.: Effective method to determine time-delay stability margin and its application to power systems. IET Gener. Transmiss. Distrib. 11(7), 1661–1670 (2017) 4. Bevrani, H., Watanabe, M., Mitani, Y.: Power System Monitoring and Control, p. 2014. Wiley, Hoboken, NJ, USA (2014) 5. Arzamasov, V., Böhm, K., Jochem, P.: Towards Concise Models of Grid Stability. IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm), Aalborg, 1–6 (2018) 6. Zhang, F., Cheng, L., Gao, W.: Prediction based hierarchical compensation for delays in wide-area control systems. IEEE Trans. Smart Grid 9(4), 3897–3899 (2018) 7. Sridhar, S., Govindarasu, M.: Model-based attack detection and mitigation for automatic generation control. IEEE Transactions. Smart Grid 5(2), 580–591 (2014)
Modelling Smart Grid Instability Against Cyber Attacks
249
8. Anwar, A., Mahmood, A.N., Tari, Z.: Identification of vulnerable node clusters against false data injection attack in an AMI based Smart Grid. Inf. Syst. 53(201–212), 2015 (2015) 9. Mary, T.J., Rangarajan, P.: Delay-dependent stability analysis of Microgrid with constant and time-varying communication delays. Electrical Power Components Systems 44(13), 1441– 1452 (Aug., 2016) 10. Ye, H., Gao, W., Mou, Q., Liu, Y.: Iterative infinitesimal generator discretization-based method for eigen-analysis of large delayed cyber-physical power system. Electr. Power Syst. Res. 143, 389–399 (Feb. 2017) 11. Kundur, D., Feng, X., Mashayekh, S., Liu, S., Zourntos, T., Purry, K.L.B.: Towards modelling the impact of cyberattacks on a Smart Grid. Int. J. Secur. Netw. 6(1), 2 (2011) 12. Liu, S., Chen, B., Zourntos, T., Kundur, D., Butler-Purry, K.: A coordinated multi-switch attack for cascading failures in smart grid. IEEE Trans. Smart Grid 5(3), 1183–1195 (May 2014) 13. Wu, G., Wang, G., Sun, J., Chen, J.: Optimal partial feedback attacks in cyber-physical power systems. IEEE Trans. Autom. Control, early access (Mar. 19, 2020). https://doi.org/10.1109/ TAC.2020.2981915 14. Rahman, M.S., Mahmud, M.A., Oo, A.M.T., Pota, H.R.: Multi-agent approach for enhancing security of protection schemes in cyber-physical energy systems. IEEE Trans. Ind. Informat. 13(2), 436–447 (2017) 15. Farraj, A., Hammad, E., Kundur, D.: A cyber-physical control framework for transient stability in smart grids. IEEE Trans. Smart Grid 9(2), 1205–1215 (Mar, 2018) 16. Schiffer, J., Dörfler, F., Fridman, E.: Robustness of distributed averaging control in power systems: Time delays & dynamic communication topology. Automatica 80(1), 261–271 (Jun. 2017) 17. Srikantha, P., Kundur, D.: A DER attack-mitigation differential game for smart grid security analysis. IEEE Transactions. Smart Grid 7(3), 1476–1485 (2016) 18. Li, J., Chen, Z., Cai, D., Zhen, W., Huang, Q.: Delay-dependent stability control for power system with multiple time-delays. IEEE Trans. Power Syst. 31(3), 2316–2326 (2016) 19. Cai, D., Huang, Q., Li, J., Zhang, Z., Teng, Y., Hu, W.: Stabilization of time-delayed power system with combined frequency-domain IQC and time-domain dissipation inequality. IEEE Trans. Power Syst. 33(5), 5531–5541 (2018) 20. Liu, M., Dassios, I., Tzounas, G., Milano, F.: Stability analysis of power systems with inclusion of realistic-modeling WAMS delays. IEEE Trans. Power Systems 34(1), 627–636 (Jan. 2019) 21. Breviglieri, P., Erdem, T., Eken, S.: Predicting smart grid stability with optimized deep models. SN Computer Science 2(2), 1–12 (2021). https://doi.org/10.1007/s42979-021-00463-5 22. Wood, D.A.: Predicting stability of a decentralized power grid linking electricity price formulation to grid frequency applying an optimized data-matching learning network to simulated data. Technology and Economics of Smart Grids and Sustainable Energy 5(1), 1–21 (2020). https://doi.org/10.1007/s40866-019-0074-0 23. Dewangan, F., Biswal, M., Patnaik, B., Hasan, S., Mishra, M.: Chapter Five - Smart grid stability prediction using genetic algorithm-based extreme learning machine. In: Bansal, R.C., Mishra, M., Sood, Y.R. (eds.) Electric Power Systems Resiliency, pp. 149–163. Academic Press (2022). https://doi.org/10.1016/B978-0-323-85536-5.00011-4 24. Delcourt, M., Shereen, E., Dan, G., Le Boudec, J.Y., Paolone, M.: Time-synchronization attack detection in unbalanced three-phase systems. IEEE Transactions on Smart Grid 12(5), 4460–4470 (2021). https://doi.org/10.1109/TSG.2021.3078104 25. Schäfer, B., Matthiae, M., Timme, M., Witthaut, D.: Decentral smart grid control. New J Phys 17(1), 15002 (2015) 26. Schäfer, B., Grabow, C., Auer, S., Kurths, J., Witthaut, D., Timme, M.: Taming instabilities in power grid networks by decentralized control. The European Physical Journal Special Topics 225(3), 569–582 (2016). https://doi.org/10.1140/epjst/e2015-50136-y
250
J. E. Efiong et al.
27. Nishikawa, T., Motter, A.E.: Comparative analysis of existing models for power-grid synchronization. New J Phys 17(1), 15012 (2015) 28. Schmietendorf, K., Peinke, J., Friedrich, R., Kamps, O.: Self-organized synchronization and voltage stability in networks of synchronous machines. The European Physical Journal Special Topics 223(12), 2577–2592 (2014). https://doi.org/10.1140/epjst/e2014-02209-8 29. Delcourt, M.M.N.: Time-Synchronization Attacks against Critical Infrastructures and their Mitigation. Lausanne, EPFL (2021). https://doi.org/10.5075/epfl-thesis-8577. https://infosc ience.epfl.ch/record/288441 30. Marie, M., Delcourt, N.: Time-Synchronization Attacks against Critical Infrastructures and their Mitigation (2021) 31. Moussa, B.: Detection and Mitigation of Cyber Attacks on Time Synchronization Protocols for the Smart Grid. September (2018) 32. Zhang, Z., Gong, S., Dimitrovski, A.D., Li, H.: Time synchronization attack in smart grid: Impact and analysis. IEEE Transactions. Smart Grid 4(1), 87–98 (Mar. 2013)
Program Implementation of Educational Electronic Resource for Inclusive Education of People with Visual Impairment Yurii Tulashvili1(B) , Iurii Lukianchuk1 , Valerii Lishchyna1 , and Nataliia Lishchyna2 1 Department of Computer Science, Lutsk National Technical University, Lutsk 550000,
Ukraine [email protected] 2 Department of Software Engineering, Lutsk National Technical University, Lutsk 550000, Ukraine
Abstract. It is substantiated that the educational software used in the educational process during the computer training of visually impaired people contributes to the formation of their correctional and compensatory devices for the use of computer technology. The peculiarities of the inclusive educational process of professional training of persons with visual impairments with the use of computer technology are pointed out. The architecture and features of creating educational software for the visually impaired are revealed. A mathematical model and algorithm for its implementation of educational software for support of inclusive educational process of professional training of persons with visual impairments have been developed. The developed training software presents the theoretical content of training in an adapted form. The developed practical tasks contain an exhaustive list for the formation of correctional and compensatory devices and the acquisition of skills and abilities to use computer technology for people with visual impairments. Keywords: People with visual impairments · Software · Vocational rehabilitation
1 Introduction Process of integrating people with disabilities into public relations involves a set of actions aimed at helping subjects with mental and physical disabilities to master the values of modern civilization, to be involved in active social activities through rehabilitation measures. Vocational rehabilitation of people with disabilities is defined as the process of providing rehabilitation measures for people with disabilities, taking in to account the limited scope of maintaining the ability to work, contraindications to certain types of work due to health for vocational education and rational employment. Training of visually impaired people with the use of information technology in professional activities is an extremely important issue of socialization, is one of the main tasks of their professional rehabilitation, the solution of which will include them © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 251–260, 2023. https://doi.org/10.1007/978-3-031-36115-9_24
252
Y. Tulashvili et al.
in society not only as socially full, but also as creatively active members [1]. This is especially with the use of ICT [2]. The development of new and the development of existing methods of using information technology for the training of the blind is especially important. Today, inclusive education is faced with the urgent task of introducing special software and hardware in the process of training future professionals.
2 Literature Analysis The urgency of the problem of vocational rehabilitation, as the main means of social and labor adaptation of people with visual impairments, is determined primarily by the place of blind people in the social environment, the fact that all visually impaired can be included in society as socially full and creatively active its members [3]. In his work “Information technologies for the visually impaired in the system of additional education of Kabardino-Balkari” L. Shautsukova noted the possibility and necessity of using new information technologies in the educational process of people with visual impairments. She believes that computer technology, in this case, should be considered not so much as a subject of study, but as a new means of creating corrective and compensatory devices, and computer technology as a tool to empower people with visual impairments to their successful rehabilitation. Modern society [4]. Research by Ermakov V. and Yakunin G. possible areas of employment of people with visual impairments was investigated in [5]. It is determined that the ideal computer workplace for a blind programmer should have the following hardware and software: an individual computer with sufficient resources for the operation of tools and adaptive software; several “screen access” programs that have different capabilities and allow you to work in different operating systems; braille display; speech synthesizer; scanner; Internet connection. The use of information technology in the educational process for blind people is not only wide access to new educational technologies, but also access to the global information space [6]. With the help of information technology, people with visual impairments have the opportunity to exercise their professional self-determination, as well as the opportunity to communicate freely with the world. Research proves that success of mastering information technologies, telecommunication networks for the visually impaired creates conditions for their professional rehabilitation [7]. Digital content related to engineering, adapted using adaptation methods, allows blind students to effectively use technical means, communication channels in combination with information technology [12]. The formation of the structure of the education system on the basis of expert evaluation plays a significant role in the analysis and selection of learning content. Methods of structuring the content of learning during the mastery of information technology and the development of compensatory devices in people with visual impairments have been disclosed in previous studies [8].
Program Implementation of Educational Electronic Resource
253
3 Object, and Research Methods In the process of forming and presenting the content of education should take into account the psycho-physiological characteristics of subjects of education with visual deprivation, such as fragmentation, verbalism and low rate of learning. Therefore, one of the conditions for the effective solution of the problem of forming the content of vocational computer training is to take into account the natural characteristics of people with visual impairments. Research on modeling the content of learning technologies determines that the implementation of operational management of cognitive activity of educational entities should be carried out both at the didactic macro level and at the didactic micro level [10]. Therefore, among the main tasks to be solved at each level of the structural and organizational model of forming the content of professional computer training of visually impaired people, first of all is the task of adapting the content at the didactic macro level and concretizing and shaping the content of blind and partially sighted students work with the use of computers at the didactic micro level [10]. In conducting research, we emphasize the special importance of the principle of formation of compensatory devices in the teaching of information technology and its combination with traditional principles in the process of forming the educational content of professional computer training for visually impaired people. The matrix method of selection and structuring of educational material proposed in researches is widely used for formation of the maintenance of training [13]. However, this method is not fully suitable for small learning tasks for computer training of people with visual impairments. In conducting research, we emphasize the special importance of the principle of formation of compensatory devices in the teaching of information technology and its combination with traditional principles in the process of forming the educational content of professional computer training for visually impaired people. The implementation of these two key principles in combination with traditional principles is to conduct a special analysis of the constituent elements of the content of education, in accordance with the defining criterion, which is characterized by the following indicators, those we have studied before [8]: 1) Perception of techniques and methods of activity through tactile and auditory analyzers, which allows you to fully determine the properties of the object or object to which the action is directed or which can be studied. 2) The formation of a complete picture of the object or subject to which the action is directed or which can be studied by repeated performance of perceptual actions and, as a consequence, the emergence of a stable information base of activity in a person with visual impairments. 3) Possibility of clear presentation of concepts, thoughts, theories and disclosure of properties of objects and objects at the verbal and tactile level of perception. The combination of these indicators, in turn, determines the level of typical information and communication actions using modern information technology adaptation, which can be performed by subjects with visual deprivation, according to their psychophysiological indicators (blind or partially sighted).
254
Y. Tulashvili et al.
Based on this, the main feature of the process of forming the educational content of professional computer training for visually impaired people is to ensure the effectiveness of the implementation of traditional and specific principles by adapting national requirements at the didactic macro level and specifying the content of didactic microlevel in a certain specialty. A feature of the educational activities of people with visual impairments is the specification of the content of education at the didactic micro level to the possibilities of professional training of people with visual impairments. Therefore, in the process of determining the content of professional computer training for people with visual impairments, we focus on the specifics of building the content of the learning process. The formation of the content of the educational task in this case is provided by the method of multiple repetition of educational objects, which is to use in the process of compiling the content of practical tasks repeatedly used selection of educational material. Thus, in the process of forming the content of professional computer training of persons with visual impairments in a particular specialty provides structured adaptation of all disciplines and specification of their content in accordance with the invariant and variable parts of the training curriculum.
4 Results The study of the scientific problem involves the disclosure of the features of the process of vocational training of persons with visual impairments using software training, the formation of quality knowledge of future professionals who are visually impaired, the use of computer technology in an inclusive educational environment. The compensatory devices as the main factor in the process of forming the competence of the future specialist. In developing the training software, we relied on the aspect that information technology for people with visual impairments is not only a set of acquired levels of knowledge, skills and abilities, but, above all, the ability to access the global information space. With the help of information technology, people with visual impairments get the opportunity for their professional self-determination, the opportunity to communicate freely with the world. In such centers, practical experience of professional training of the blind with the use of information technology has been gained [11]. Based on many years of experience, it was concluded that the most common problems faced by teachers in the training of visually impaired people in teaching them to work with computer technology are the following: – overcoming the psychological barrier associated with uncertainty about computer technology, with great doubts about the ability to master it; – the question of mastering the techniques of working on the keyboard. Getting to know the computer keyboard and learning the location of the keys on it and setting hands for people who have experience with a typewriter comes down to finding out the differences between typing text and entering text from the keyboard into a computer. For others, such skills are formed in the process of hard work, which is quite sufficient for mastering the keyboard;
Program Implementation of Educational Electronic Resource
255
– the need for orientation in the electronic text document, both in the self-created and in that typed by someone; – work with operating systems is to learn the techniques of moving between computer disks, directories, files; – awareness that the same keyboard shortcuts in different software can perform different functions; – mastering the concept of file name and extension, their purpose and ways to create and change them, etc. The need to solve these and other problems that arise in the process of teaching the blind information computer technology, necessitates the development of appropriate teaching methods, manuals, recommendations for working with relevant software products and more. The professional development of people with profound visual impairments in the field of engineering and pedagogical training is associated with the study of special disciplines in the field of computer technology. First of all, in the process of professional rehabilitation, special attention was paid to the study of computer technology, information and communication technologies in classes in subjects closely related to computer science. Studies of the problem of inclusion in higher education on the example of teaching visually impaired students through the use of special programs, educational Internet environments using so-called assistive technologies (Assistive Technologies) show the effectiveness of the use of computer tools in the educational process. Assistive technologies help the blind to obtain information for learning and work by receiving voice support when using modern processor technology [9, 14]. The basis of the success of its application is to learn how to work with a regular keyboard. The study of new educational material should be based on the principle of consistency and gradual complication. The material should be presented interconnected, in a logical sequence of its study. For example, when studying the Word text editor, you should first get acquainted with the combinations of those “hot keys” that will open the editor, save the document to your computer disk, and then move on to the combinations associated with document formatting. Armed with the “ten-finger method” of working on the keyboard, blind students begin to feel more confident. They form a positive motivation for further study, they understand that they have ample opportunities to access educational material, which today in many educational institutions is available on electronic media. The next stage of rehabilitation work is extensive cooperation with teachers. The material presented in lectures, practical and laboratory classes should be transferred as much as possible to electronic media, such as Word documents. The most painstaking task is to “voice” mathematical formulas, drawings, diagrams and graphs. This requires additional time for their comprehension and correct reproduction of the text. The study materials prepared in this way are provided to blind students before the classes for preliminary acquaintance with them on their own. In seminars, practical and laboratory classes, classrooms where students with visual impairments will study must be equipped with at least one computer workstation. This will ensure the employment of the blind student during the lesson and will not allow him to sit passively when the
256
Y. Tulashvili et al.
whole group is working on a task, which consists in calculating the results, compiling an analysis, generating a report. Analysis of the experience of professional training of persons with visual impairments by the Laboratory of Assistive Learning Technologies and other rehabilitation centers shows that today professional and higher education is facing a humanistic problem determined by global trends in informatization of all spheres of human activity and full involvement of human potential development of the information society. Qualitative changes in public life have led to the restructuring of technologies for collecting, storing, transmitting and presenting information. Without understanding these processes, no modern person can lead an active life today. The intensive development of computer hardware and software has led to the emergence and further spread of the concept of using information technology as a means of developing compensatory devices for the visually impaired. Adaptive assistive computer technology provides greater access to blind and partially sighted people to information provided in electronic form, opens new opportunities for its socially useful activities. In terms of the use of information technology as an adaptation technology for people with visual impairments, the concept of “information technology” acquires a new specific meaning that reflects the impact of computer technology on modern living conditions of people with visual impairments through the adaptive direction of modern information technology. The software provides basic knowledge of computer science and techniques of working with computer technology. This creates the conditions for the formation of skills and abilities that people with visual impairments need to work on a personal computer, to successfully master the techniques of working with information systems. The structure of the educational material is built in such a way that each educational module is divided into a content module of theoretical training and a content module of practical training. Theoretical training will provide visually impaired people with knowledge of the basics of computer science and the principles of computer technology, and practical training, through exercises, will allow you to acquire skills and abilities to work on a personal computer. Practical exercises should be performed with the monitor turned off using the JAWS for Windows screen reader, performing actions only on the keyboard console. This will contribute to the successful development of compensatory devices in visually impaired and blind education, which will allow them to use computer technology even when the defect has led to complete loss of vision. The Fig. 1 shows software is designed for people with visual impairments and will be useful when studying in educational institutions in disciplines designed to develop in subjects with visual impairments knowledge, skills, and abilities to use computer technology to process information. As a result of this approach in the system of content of professional computer training of people with visual impairments we have identified: set A, consisting of all professional industry competencies; many professional competencies of a specialist in the application of computer technology – B; set C, which corresponds to a set of professional competencies in the use of computer technology, which are available for mastery of persons with visual impairments with the use of computer adaptation tools.
Program Implementation of Educational Electronic Resource
257
Fig. 1. The interface of the developed software product (Informatyka_for_blind).
The ratio of the relationship between the many professional competencies, the content of which can be revealed through educational facilities (denote s), is available for people with visual impairments through computer means of adaptation: C ∈ B ∈ A ≡ ∀s(s ∈ C → ∀s(s ∈ B → s ∈ A)).
(1)
This means that any learning object x is accessible to people with visual impairments through computer means of adaptation and that is part of the professional competence, which is an element of the set C, is an element of the set B of professional competencies, which, in its in turn, determines its inclusion in the set of professional industry competencies A. The formation of the content of the educational task in this case is carried out using the proposed method. The method consists in the selection in the process of compiling the content of practical tasks of the selection of educational material by increasing the level of their complexity (level). Any educational task consists of a set of elements in the form of their combinations of Si. The combination of educational objects takes the form of a tuple – a mathematically ordered and finite set of elements S. In the process of defining the set S = {S1, S2, …, Sn}, as a set of various objects of intellectual learning, we consider such a set of educational objects, which forms a certain amount of knowledge, skills and abilities that correspond to the competence to be mastered. a person with visual impairments, studying the thematic section of a particular content module. The main condition for the formation of the content of practical work is that the number of training units should not exceed the dimensions of the tuple of data. To form the structure of connections in each educational task we use a polynomial generating function: n n (2) xk = (1 + x)n . k=1 k
258
Y. Tulashvili et al.
According to the results of calculations, binomial coefficients will give possible connection structures. We will reveal the principles of this approach on the example of determining possible structures of educational tasks formed from three educational objects S1, S2, S3 provided they are combined to the number n = 5, when the following characteristics are identified: S1 occurs no more than once, S2 – no more than two, and S3 – once or twice. Then the generating function takes the following form: F(si, x) = (1 + S1 x)(1 + S2 x + S22 x2 )(S3 x + S32 x2 )
(3)
According to the results of calculations we get: S3 x + (S1 S3 + S2 S3 + S3 S3 )x2 + (S1 S2 S3 + S2 S2 S3 + S1 S3 S3 + S2 S3 S3 )x3 +(S1 S2 S3 S3 + S2 S2 S3 S3 + S1 S2 S2 S3 )x4 + (S1 S2 S2 S3 S3 )x5 .
(4)
From the solution we have the following possible values of the combinations educational objects. Learning tasks can be formed in the process of compiling learning tasks under such a condition that is determined by the presence of all initial actions in different variations. Therefore, learning tasks need to be formed from a set of tuples: n = 3 : [S1 S2 S3 ], [S2 S2 S3 ], [S2 S3 S3 ], [S1 S3 S3 ]; n = 4 : [S1 S2 S3 S3 ], [S2 S2 S3 S3 ], [S1 S2 S2 S3 ]; This approach allows you to automate the process of content formation using the algorithm below (Fig. 2).
Fig. 2. Algorithm for forming the content of the educational task
Program Implementation of Educational Electronic Resource
259
Figure 3 reveals the formation and interaction of data streams that form the content of the educational task using the proposed algorithm of the software training tool designed for people with visual impairments. The stage of software implementation of content management software provided by the software tutorial is designed using web on platform Node.js as web-site on React.js. We are currently updating the software product, as well as developing an application for a smartphone. The software is designed for people with visual impairments and will be useful when studying in educational institutions in disciplines designed to develop in subjects with visual impairments knowledge, skills and abilities to use computer technology to process information.
Fig. 3. Simplified data model
5 Conclusion In the process of professional training of visually impaired people it is necessary to apply such pedagogical methods that use new achievements in the field of technical means and assistive technologies, which are constantly updated and improved, adapting the needs of the blind to the needs of modern society. The proposed formalized approach to the formation of the content of education in the system of teaching visually impaired people to use computer tools as a means of adaptation is more effective because it does not require complex mathematical apparatus to implement simple algorithms for structuring the content of educational activities of people with visual impairments. Mathematical model and algorithm for its implementation to determine the content of educational tasks of vocational training of people with visual impairments can be easily implemented through software. The software implementation of content management software provided as the software tutorial and is realized web on platform Node.js as web-site on React.js. This creates an opportunity to implement in teaching technology the principles of individual and differentiated approaches, which are the basis of special pedagogical management of educational and developmental process of professional computer training of
260
Y. Tulashvili et al.
people with visual impairments, which is the subject of further research. Currently, educational materials are being prepared to improve learning outcomes by applying neural network algorithms.
References 1. Jennifer, K., Delia, F.: Enabling people with disabilities through effective accessible technology policies. In: Rune, H., Bjørn, H., Jerome, B., Delia, F., Ana, M., Guillén, R. (eds.) The Changing Disability Policy System, pp. 127–143. Routledge (2017) 2. Lewthwaite, S.: Web accessibility standards and disability: developing critical perspectives on accessibility. Disabil. Rehabil. 36(16), 1375–1383 (2014) 3. Ferri, D., Favalli, S.: Web accessibility for people with disabilities in the european union: paving the road to social inclusion. Societies 8(2), 40 (2018). https://doi.org/10.3390/soc802 0040 4. Shautsukova, L.: Information technologies for the visually impaired in the system of additional education of Kabardino-Balkaria. In: Proceedings of the 2nd All-Russian meeting of the NMS on informatics. Actual problems of informatics in modern education, pp. 524–528. MSU Press, Moscow (2005) 5. Ermakov, V., Yakunin, G.: Development, Training and Education of Children with Visual Impairments, p. 240. Enlightenment, Moscow (2000) 6. Senjam, S.S., Foster, A., Bascaran, C., Vashist, P., Gupta, V.: Assistive technology for students with visual disability in schools for the blind in Delhi. Disabil. Rehabili.: Assistive Technol. 15(6), 663–669 (2020). https://doi.org/10.1080/17483107.2019.1604829 7. Laabidi, M., Jemni, M., Ben Ayed, L.J., Brahim, H.B., Jemaa, A.B.: Learning technologies for people with disabilities. J. King Saud Univ. – Comput. Inform. Sci. 26(1), 29–45 (2014). https://doi.org/10.1016/j.jksuci.2013.10.005 8. Tulashvili, Y.: Theoretical and methodical principles of professional computer training of persons with visual impairment: dis. of Dr. Ped. Sciences: [spec.] 13.00.04 “Theory and methods of professional education”, p. 528. Vinnytsia State Pedagogical University Mykhailo Kotsyubynsky University. Vinnytsia (2012) 9. Viner, M., Singh, A., Shaughnessy, M.F.: Assistive technology to help students with disabilities. In: Singh, A., Viner, M., Yeh, C.J. (eds.) Special Education Design and Development Tools for School Rehabilitation Professionals, pp. 240–267. IGI Global (2019). https://doi. org/10.4018/978-1-7998-1431-3.ch012 10. Sineva, E.P.: Tiflopsychology, vol. 144. Textbook. Kyiv (2006) 11. Burgstahler, S.: 20 Tips for Teaching an Accessible Online Course. Disabilities, Opportunities, Internetworking and Technology, University of Washington (2022). https://www.washington. edu/doit/20-tips-teaching-accessible-online-course 12. Batanero, C., de Marcos, L., Holvikivi, J., Hilera, J.R., Oton, S.: Effects of new supportive technologies for blind and deaf engineering students in online learning. IEEE Trans. Educ. 62(4), 270–277 (2019). https://doi.org/10.1109/TE.2019.2899545 13. Alekseeva, A., Antonenko, O., Zhadan, K., Lyfenko, M.: Experience in using e-learning tools in inclusive educational space of higher school. Phys. Math. Educ. 18(4), 17–24 (2018) 14. Jones, V.L., Hinesmon-Matthews, L.J.: Effective assistive technology consideration and implications for diverse students. Comput. Schools 31(3), 220–232 (2014). https://doi.org/10.1080/ 07380569.2014.932682
A Version of the Ternary Description Language with an Interpretation for Comparing the Systems Described in it with Categorical Systems G. K. Tolokonnikov(B) VIM RAS, Moscow, Russia [email protected]
Abstract. The article continues the development of a version of the language of ternary description (TDL), in which it is possible to compare the concepts of the system given by A.I. Uyemov on TDL and in the theory of categorical systems developed by the author. Categorical systems theory is an integral part of algebraic biology that predicts the properties of organisms based on the genome, including intellectual properties, which leads to artificial intelligence models. A wide range of systems approaches is formalized in TDL, so it is very important to find relationships between TDL systems and category systems theory. This paper proposes a version of the TDL, in which it is possible to present the definition of the system in the TDL in the required form, in which, in particular, the subjective interpretation of the symbol “arbitrary” in the LTO is excluded, which is unacceptable in classical logic, without which it is impossible to establish the relationship of this type of systems with the system block of algebraic biology. Keywords: Categorical systems · Algebraic biology · Predicates · Properties · Relations · Artificial intelligence
1 Introduction In the previous work “Ternary description language and categorical systems theory”, reported at the conference, we began to develop a version of the ternary description language (TDL), in which a number of inaccuracies of the original TDL were corrected, which is necessary to clarify the general theory of systems proposed by A.I. Uyemov [1–8], and its relationship with the categorical systems theory that we are developing [9–12]. We note that the definitions introduced below agree with our definitions in that paper. In this work, we take the next step. A.I. Uyemov proceeds from the objectivity of systems: “The objectivity of connections between phenomena, their independence from our consciousness means the objectivity of systems. Material systems exist independently of human consciousness” [6]. For now, the recognition of objectivity for material systems is enough for us, although A.I. Uyomov also speaks about ideal systems. Let’s draw an analogy with the concept of an atom. Atoms, like systems, exist objectively. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 261–270, 2023. https://doi.org/10.1007/978-3-031-36115-9_25
262
G. K. Tolokonnikov
The science of atoms consists in the search for true statements about atoms. We must not forget the fundamental role of the concept of truth for science. At first, we do not know the properties of atoms, but having discovered atoms with instruments, we first give a clear name, as G. Frege says, to these particles, then physicists study them, it may turn out that not all particles detected as atoms are atoms (for, example, they can sometimes be confused with molecules), this deficiency is corrected in the process of studying atoms. So, the first step in the study is to discover atoms and introduce the designation for these objects “atom”, the further steps of building a theory are to search for true statements about atoms and their properties. The result is a more and more complete definition of the concept of an atom (which includes atoms of various elements, and so on), due to its inexhaustibility, like any real object, it is not possible to achieve a final definition covering all its properties. This is how a number of definitions of the atom arise: the Thomson atom, the Bohr atom, the atom of non-relativistic quantum mechanics, and the atom in quantum field theory… The search for true statements about the atom is not limited by direct experiments. As G. Frege writes: “The grounds that justify the recognition of some truth are often contained in other, already recognized truths…. Logic deals only with those grounds of the process of judgment which are truths. The process of judging, in which other truths are recognized as the grounds supporting it… is called the process of inference. There are laws related to this kind of confirmation, and the establishment of the laws of correct inference is the goal of logic” [14]. For reasoning, a language is needed, for the role of which natural language is not suitable, due to the fact that its purpose is much broader than the task of finding and substantiating the truth in the process of reasoning. Let’s move on similarly (scientifically) to systems. Many researchers (in accordance with A.I. Uyemov’s requirement for the objectivity of systems) discovered what they called systems. They have been given a clear name “system”. Further, in the process of studying this object of reality, researchers, to the best of their abilities, put forward options for defining the concept of a system, at the moment when A.I. Uyemov gave his definition, several dozen definitions of the concept of a system accumulated, in particular, the concept of a functional system according to P.K. Anokhin, the formalization of which is carried out within the framework of categorical systems. In his approach, however, A.I. Uyemov breaks with the natural scientific process of studying a phenomenon (system), in order to improve or replace the existing definitions with a more adequate and accurate one. Instead of studying the phenomenon, he took up the study of a set of definitions obtained by other researchers, proclaimed that it was necessary to look for something in common in their definitions, found in his opinion this something in common, and presented it as his definition of the system. This approach for the case of the atom would consist in revealing something in common (which undoubtedly exists) in the definitions of the atom by Thomson, Bohr, and others, and declaring this common definition to be more adequate to the real atom. The obvious absurdity of such an approach in the case of an atom, in our opinion, means the absurdity of such an approach in the case of any other existing object, including the system recognized by A.I. Uyemov as an objective phenomenon. It can only be argued that the definition of the system given by A.I. Uyemov, if it makes sense to consider it, is only in isolation from the method of its derivation.
A Version of the Ternary Description Language
263
For the study, inference, as G. Frege points out, and presentation of the results of study, in particular, systems, a formal language based on suitable logic is required. A.I. Uyemov does not believe that: “…Formalization based on the currently available logical apparatus can serve as the basis for a general systems theory.“ Moreover, in his opinion: “For the general theory of systems, a “mathematical suit” has not yet been prepared… There are no mathematical means adequate to the structural representations of systems of objects.” The language of ternary description (TDL) proposed by him is such a “logical and mathematical apparatus adequate to the problems of general systems theory” [6]. Unfortunately, in many respects, the TDL is unfinished and, as indicated in [17], has incomplete and contradictory moments. We also emphasize the opinion expressed by P. Materna: “The tendency to build “non-classical” and “non-standard” logics is justified only when it is shown that “classical or “standard” logic cannot solve some logical problems [14]. I don’t think anyone showed it. One of my main arguments is that the “standard” logic can easily cope with the solution to those problems, because of the solution on which the TDL was built”. Unfortunately, P. Materna did not provide solutions using standard logic for those logical problems that A.I. Uyemov put forward, to justify the need to develop TDL. A similar problem of standard logic A.I. Uyomov saw in the apparent impossibility of an unambiguous understanding of predicates from several variables. For this case, we further find (according to P. Materna) a solution to the problem of A.I. Uyemov by means of ordinary logic. One of the main theses of A.I. Uyemov, by which he substantiates the need to construct a TDL, consists in the difficulties of ordinary logic indicated by him, in particular, which, in his opinion, exist in the definition of many-place predicates. Let us consider these problems of “different understanding of the simultaneous prediction of many things” [6]. He writes: “One or another property is attributed to a set of things S1 ,…, Sn – the predicate P…. In what sense can the predicate P be attributed (produced) to a set of things? Various understandings of this connection are possible” [5]. It seems to us that the understanding of the formula (S 1 ,…, S n )P (A.I. Uyemov writes the name of the predicate in the left entry) is unambiguously given by the definition, and in the text of A.I. Uyemov there are, in our opinion, various misunderstandings of the indicated formula. Since A.I. Uyemov operates with the term set, we will define of an n-place predicate from set theory. “Let A be an arbitrary set. An n-ary function f defined on A with values in the set {I, L} is called an n-ary predicate on A. The set of those sequences (a1 ,…, an ) from An (the Cartesian product of n factors) for which is called an n-ary relation on A corresponding to the predicate refer to [16, 18]. Within the framework of axiomatic set theory (for example, NGB from [16]), this definition is unambiguous, and, we emphasize, has no alternative understandings, including those proposed by A.I. Uemov and, by the way, in which he sees logic problems. Consider the understanding offered by A.I. Uyemov. Obviously, it would not make sense to consider the same understanding of the predicate P, which is given in the above definition from mathematical logic, so one should expect some other understanding given by A.I. Uyemov. “With one understanding, the inherentness of a predicate in a set means its inherent in each element of this set” [4]. A.I. Uyemov does not define the terms “assignment of a predicate”, “intrinsicity of a predicate”, unknown in mathematical logic. Let us
264
G. K. Tolokonnikov
assume that they are synonymous with “the presence of a relation (predicate)”, but, here, the meaning of the “intrinsicity of its < predicate >… Element” (that is, a separate element), which is also not defined by A.I. Uyemov, is already completely unclear. Nevertheless, A.I. Uyemov gives a name to something unknown (according to G. Frege, this is already a way out of the logic in which each name has a single meaning in the form of a corresponding object). Apparently, A.I. Uyemov has an intuitive understanding of “inherence”, but he did not find the opportunity to take the trouble to convey it to the reader. A.I. Uyemov proposes to guess what “inherence” is by considering an example: “For example, “all the large planets of the solar system move approximately in the plane of the ecliptic” [6]. Probably, the elements in this example correspond to the “major planets of the solar system” from the list: Mercury, Venus, Earth, Mars,…, Neptune, and the predicate P is = “Mercury, Venus, Earth, Mars,…, Neptune moving approximately in the plane of the ecliptic. In this example, if you enter a unary predicate (property) = “___ moves approximately in the plane of the ecliptic”, the original predicate is the same as the predicate. This formula clearly expresses the intuitive feeling of A.I. Uyemov. Thus, instead of the expected other understanding of the predicate P promised by A.I. Uyemov, we see a clear misunderstanding of the original predicate P, for some reason reduced to a very special case, expressed by the last formula. The predicate P does not satisfy the understanding given by A.I. Uyemov, the simplest example is the case for which there is no decomposition into unary predicates. Let’s move on to the following understanding of the predicate P, proposed by A.I. Uyemov: “In a different understanding, the predicate denotes a property inherent in the aggregates of most elements of the system, in a particular case – all elements taken together.“ A.I. Uyemov again does not define what a “property inherent in the aggregates of most elements” is, he thinks, for example, that the reader understands the word “majority” in his text… The work of searching for the meaning of A.I. Uyemov’s text is again shifted by him on the reader who is given the example “the forest burned down” as a clue. An example that is unsuccessful, since it is easy to carry out a construction similar to the first (mis)understanding of the formula for the predicate: we introduce the predicate = “___ burned down”, here, obviously, trees (bushes, etc.), the original predicate P is now represented by conjunction and negations (for unburned trees). From these conjunctions it is possible to assemble predicates on the Boolean of the forest, most likely the representation of the original predicate P by such predicates corresponds to the intuition of A.I. Uyemov in this second understanding of the predicate P. Further, we read: “In a certain respect, the reverse will be such an understanding of predication, when the predicate P is inherent in at least some of the elements of the system, for example, in the sentence “the ancient Greeks were outstanding philosophers” [6]…. are not differentiated, although the difference between them is significant. Of course, there is an ambiguity in the intuitive part of “some of the elements of the system”, one should rely on the boolean. We cannot find the meaning of the phrase “reverse < to the previous >”, considered in the book [6] and [14], inverse relations make sense for binary relations, but not for arbitrary…
A Version of the Ternary Description Language
265
A direct contradiction in the phrase “they are not differentiated, although the difference between them is significant” directs A.I. Uyemov to the already mentioned term “majority” to save the text, but we will not waste time on correcting the text. The non-differentiability of the third understanding from the second gives the same representation of the initial predicate P as in the previous understanding, so here we do not have another understanding of the predicate promised by A.I. Uyemov. The fourth understanding is described by the words: “Finally, such an understanding of predication is possible, in which the predicate characterizes the entire set by itself (as one whole)… In this case, the predicate of the system is not a predicate of a simple collection of all its elements… In some cases, such a predicate (let’s call it PIV) cannot be decomposed into a set of predicates related to individual elements of the system…” [6]. The fourth understanding, apparently, despite the mysterious words “the whole set in itself” (A.I. Uyemov does not notice that the essence is one element of the Cartesian product, for example, one point of three-dimensional space, and not three points x, y, z…) coincides with the definition of a predicate from mathematical logic. We also note the direct error of A.I. Uyemov “In some cases, all Pk are equal… to the predicate of the entire system” [6] (a function of several variables that explicitly depends on them cannot be equal to the function one of these variables). A.I. Uyemov did not offer any other understanding of the predicate P (except for the one given in the definition from mathematical logic), that is, the “problems of logic” declared here by A.I. Uyemov (in particular, in the form of “differences in predication types”) are absent. There are several dozen definitions of the concept of a system, the deepest of which is the functional system. These approaches largely fall under the language of ternary description (TDL) developed for systems [1–8], where the definition of functional systems was also considered. However, this consideration turns out to be incomplete, since the role of the systemforming factor, which also underlies categorical systems, remains unclear. This article continues the development of the TDL version, in which it is possible to present the definition of the system in TDL in a form in which it is possible to establish the relationship of this type of system with the system block of algebraic biology.
2 Ternary Description Language Version with Explicit Interpretation The A.I. Uyemov introduces two symbols t and a into TDL, in fact, without an exact indication of their meanings, in violation of the key principle according to G. Frege “there should be no meaningless meaningless expressions in science” [18]. Indeed, he writes: “… The elementary cell of the formal apparatus of general systems theory… Consists of two “objects” – definite and indefinite (t, a)…, “t” and “a” are just things.” [6]. In TDL words in which the symbols t participate, and their interpretation as a thing, relation, or property is introduced depending on the place occupied by the symbols in the word, so (t)a is interpreted in natural language as the phrase “a certain thing has an indefinite property”. A.I. Uyemov has nothing else in mind, both in the quoted book and in other works, using these symbols. In mathematical logic, such statements are not
266
G. K. Tolokonnikov
considered, since the phrase “certain thing” does not have any meaning, and the phrase “indefinite property” does not define any predicate. We propose the following correction of the contradictory situation in the TDL regarding “t” and “a”. We will consider the roots of the words of a certain fragment of a natural language and denote their totality by R. In the fragment under consideration, we are primarily interested in the parts of speech noun, adjective (denoting a property of an object and answering the questions “what?”, “what?”, “what?”, “what?”, “whose?” and so on) and a verb. In a language fragment, we usually consider only declarative sentences, simple declarative sentences consisting of a subject (which is a noun), a predicate (which is a verb), and an object (which is an adjective), and complex sentences (consisting of several simple sentences, connected by unions, complex and other sentences). The main members of the sentence are the subject and the predicate; the secondary members of the sentence include the addition, usually in the form of an adjective. The root is associated with the meaning (lexical meaning) of the word, just like the thought, we do not define the concept of meaning, therefore we will restrict ourselves to a typical description: the root is a morpheme that carries the lexical meaning of the word, in Russian the root is present in all independent parts of speech and is absent in many service parts of speech (for example, it is not in the union “and”, the interjection “ah”). Compound words have multiple roots. Words can be formed from the root (point to words with the same root), which can be the subject in the sentence, the resulting word will define a certain concept, for definiteness we will understand it according to G. Frege within the framework of his universe. This concept has a set or class of things that fall under it. From the root, you can also form a verb and an adjective (more precisely, in the fragment of the language under consideration, we take into account only such word roots), which will serve as a predicate and an object in the sentence. The limitations of the fragment of the language under consideration include the requirement that the matching of the root is unambiguous: the root corresponds to only one noun noun(r), one adj(r) adjective and one verb(r). Thus, there is a set of things V = noun(R) and a set of properties S = adj(R). We will call a statement or judgment such declarative sentences for which it is possible to raise the question of their truth or falsity. For example, the phrase “a certain thing has an indefinite property” considered in the indicated fragment of the language is a sentence for which the truth value does not directly exist. However, for such sentences, we will offer an interpretation in which truth-values exist. By the symbol t we will mean a variable running through the set of roots of words R. Other possible variables will be denoted using primes: t, t’, t’‘… – different variables on the set of roots. Denote by a, a , a , … The variables running through the set of variables {t, t t …}, if a is one indefinite root, then a is another indefinite root. This is our understanding of the meaning of the signs “t” and “a”. So we have two kinds of variables. At the same time, there is a connection between the varieties in the form of interpretation of elements of the second kind (a, a , a , …) by elements of the first kind, which themselves are interpreted by the roots of words.
A Version of the Ternary Description Language
267
A.I. Uyemov, in addition to “t” and “a”, introduces the symbols of brackets), (,], [, an asterisk *, the symbol i. Some words in the alphabet from the letters t, a,), (,], [, * it compares declarative sentences, which we call T DL-statements, and phrases that can play the role of a subject in a sentence (the latter are obtained if the TDL-statement is enclosed in square brackets), natural language: (t)a – “a certain thing has an indefinite property”; (t)t – “a certain thing has a certain property”; (a)t – “an indefinite thing has a certain property”. (t*)a – “an indefinite property is attributed to a certain thing”; a(*t) – “an indefinite relation is attributed to a certain thing”; [(a)t] – “an indefinite thing with a certain property”; and so on. When comparing “t” and “a”, the words thing, property, and relation are replaced depending on the place in the formula (if the symbol is to the right of the symbol enclosed in parentheses, then the word property is selected, if on the left, then the relation, so that inside the parentheses, the word thing is selected). In our version of TDL, which we will call TDLI (TDL with interpretation), two-sorted predicate letters are used to interpret LTO statements. Let us compare in our version of TDL, that is, in TDLI, the two-sorted predicate P for (t)a: P(t, a) = “the definite thing noun(t) has the indefinite property adj(a)” = “noun(t) is adj(a)”. To evaluate the predicate, we select a specific pair (v, s), v is a fixed thing, for example, an explicitly specified “hand”, s is a fixed property, for example, “red of a given redness level”. We have fixed hand v ↔ class “hands” ↔ root t = “hands”, fixed level of red ↔ property class “red” ↔ root t’ = “red” ↔ variable a = t’: P(“hand”, t’) = “the selected particular hand is red of the selected particular level of redness”. If the selected hand is red of the selected redness level, then the predicate is 1, if not, then it is 0. Predicates for (ξ *)η and η(*ξ ), where ξ , η are t or a, are introduced and evaluated similarly. Let us consider an analogue in TDLI of the operation of enclosing TDL-statements in square brackets using the case [(a)t] as an example. An analogue would be the introduction of the operation µ of the translation of the judgment P(t, a) = “noun(t) is adj(a)” into the root µP(t, a) = µPta = “noun(t), which is adj(a)”, corresponding to given thing noun(t) having the property adj(a). Here, linguistically, at each evaluation, we have two roots with the second subordinate as an adjective to the first root. For now, we will proceed formally, similarly to how a free algebra is built on a fixed set of generators and under some operation (the application of an operation to generators is declared a new element). For new roots, we leave the requirement of unambiguous mappings noun(), adj(), and verb(). Let’s move on to the use of the concept of “arbitrary” in TDL, in addition to the concepts of “definite” (t) and “indefinite” (a). A.I. Uyemov in TDL for the concept of “arbitrary” introduces the symbol A into the alphabet.
268
G. K. Tolokonnikov
The definition of the system given by him has the form (≡ means equal by definition, the letter i indicates that the right and left of the identity under the letter A means the same arbitrary object). (1) (iA)S ≡ ([a(∗iA)])t, or t iA∗ a . “Object A can be understood as “any thing”. But this any thing is essentially different from a. Object a is an indefinite thing, object A is an arbitrary thing. As A, we can take any thing we want. Otherwise, we take any thing that comes across. This difference is well illustrated… by the choice of brides in a Russian fairy tale. If the brides lined up in front of the prince and he chose any one, then such a situation would be symbolized with the help of A. It is a different matter when he shoots an arrow and must marry any one that comes across…. The difference between A and a is not expressed in the syntax of the predicate calculus” [6]. Here A.I. Uyemov is mistaken, this difference can be easily expressed using limited quantifiers. Let’s bring this expression and we will use it in TDL, our corrected version of TDL. Recall what bounded quantifiers are (see, for example, [15]) and use them for the notion of “arbitrary”. Let the predicate R(x) be given on the set that the variables x run through; it selects a certain subset. If we want to consider variables only on this subset, then bounded quantifiers are used, and properties are satisfied for any predicate P(x). A.I. Uyemov in TDL does not speak about the usual choice from a set of elements, but about the choice of elements that he (or the prince, or someone else) wants to choose. He may like several elements (the prince may like several brides) and from them he will already make the usual choice. Thus, for him in the set of elements, there is a subset of elements he likes, which is described by the predicate R() = “I liked this ___ element”. Thus reasoning, variables, quantifiers are narrowed to the specified subset, but narrowed variables, formulas, and restricted quantifiers can be applied on the narrowed subset, this is a well-known section of the predicate calculus. So, let the predicate R() be given, then in our TDL, for example, (t)A has the following analogue: PR (t, aR ) = noun(t)is adj(aR ) , A = aR
(2)
Here, however, we are faced with a very serious question: do the predicates R corresponding in the choice of brides, for example, to A.I. Uyemov or to some prince differ? The answer is obviously yes. As you know, “like” is an extremely “subjective” verb. In TDL, subjective judgments and concepts are clearly introduced and developed in this way, which leads, in matters of truth, to the dependence of conclusions on one or another person and his feelings. As a result, introducing the symbol A, which is clearly associated with subjective sensations, A.I. Uyemov violates the most important principle of science in TDL, justified by G. Frege and, in particular, on the basis of which he built the predicate calculus, which presented classical logic in its modern form. The choice from what is desired (likes and the like) reflects the subjective idea of the chooser about the chosen object (thing). “The goal toward which science is striving is truth…. What is true is true regardless of our recognition (p. 287)…. The states of their inner world experienced by different people
A Version of the Ternary Description Language
269
cannot be combined in one consciousness and, thanks to this, compared (p. 288)…. From the meaning and meaning of a sign, the representation associated with it should be distinguished. If the meaning of a sign is a sensually perceived object, then my idea of it is an internal image… The idea is subjective: the idea of one person is not the idea of another… The idea differs significantly from the meaning of the sign in that the meaning of the sign can be the common property of many people… (p. 232)…. In order to more clearly present all the originality of our predicate true, let us compare it with the predicate beautiful…. The beautiful has gradations, but the truth does not… The truth does not depend on our recognition of its truth, and the beautiful is only beautiful for the one who perceives it as such. What’s great for one isn’t necessarily great for another…. How can one make the Negro in Central Africa give up the view that the narrow noses of Europeans are ugly, while the wide Negro noses, on the contrary, are beautiful?… Even if it were possible to give a definition of a normal person and, thereby, objectively define the beautiful, it would still happen on the basis of subjective beauty…”(p. 311 in [13]). A.I. Uyemov, introducing the symbol A, associated with the concept of “arbitrary” he defines, tries to introduce elements of subjective representations into the logic of the DTL created by him, which, being different for different people, do not allow to attach meaning to the symbol A and, thereby, to determine the above R() predicate for bounded variables and language quantifiers. Some way out, which we accept for the development of the DTL, can be the construction of the predicate R(), based on the version indicated by G. Frege with the definition of a normal person. It seems possible to develop a set of properties for such a definition, while still changing the idea of a normal person over time, if one pays attention, for example, to the solution in intuitionistic logic that is available in Kripke’s semantics.
3 Conclusion There are numerous systems theories, in particular, the general theory of systems by A.I. Uyemov, based on TDL and covering several dozens of system approaches. Categorical systems theory is an integral part of algebraic biology that predicts the properties of organisms based on the genome, including intellectual properties, which leads to models of artificial intelligence (AI) related to strong AI, in contrast to approaches based on neural networks [16–19]. In this paper, a version of the TDL fragment is proposed, in which the definition of the system given by A.I. Uyemov was successfully presented in the form necessary for comparison with the system block of algebraic biology. In an important special case of interpretation of many-place predicates, Materna’s idea that the problems of logic put forward by A.I. Uyemov to justify the need to construct a TDL can be solved within the framework of ordinary logic is confirmed.
References 1. Uyemov, A.: The Language of ternary description as a deviant logic: Part 1. Boletim da Sociedade Paranaense de Matematica 15(1), 25–35 (1995)
270
G. K. Tolokonnikov
2. Uyemov, A.: The Language of ternary description as a deviant logic: Part 2. Boletim da Sociedade Paranaense de Matematica. 17(1), 71–81 (1997) 3. Uyemov, A.: The Language of ternary description as a deviant logic: Part 3. Boletim da Sociedade Paranaense de Matematica. 18(1), 173–190 (1998) 4. Uyemov, A.I.: Some issues of the development of modern logic. Sci. Notes Tauride Natl. Univ. 21(1), 89–96 (2008) 5. Uyemov, A.I.: Things, properties and relations, M., Publishing house of the Academy of Sciences of the USSR (1963). (in Russian) 6. Uyemov, A.I.: System approach and general theory of systems, M., Thought (1978). (in Russian) 7. Uyemov, A.I., Saraeva, I.N., Tsofnas, A.Yu.: General systems theory for the humanities, W., Wydawnictwo “Uniwersitas Rediviva” (2001). (in Russian) 8. Uyemov, A.I.: Some questions of the development of modern logic. Sci. Notes Taurida Natl. Univ. 21(1), 45–58 (2008) 9. Tolokonnikov, G.K.: Informal category systems theory. Biomachsystems 2(4), 7–58 (2018) 10. Tolokonnikov, G.K.: Categorical gluings, categorical systems and their applications in algebraic biology. Biomachsystems 5(1), 148–235 (2021) 11. Tolokonnikov, G.K., Petoukhov, S.V.: From algebraic biology to artificial intelligence. Adv. Intell. Syst. Comput. 1126, 86–95 (2020) 12. Tolokonnikov, G.K.: Convolution polycategories and categorical splices for modeling neural networks. Adv. Intell. Syst. Comput. 938, 259–267 (2020) 13. Frege, G.: Logic and logical semantics, M., Aspkt Press (2000). (in Russian) 14. Leonenko, L.L.: Definitions in the language of ternary description. Sci. Notes Taurida Natl. Univ. 24(3), 397–404 (2011) 15. Leonenko, L.L., Tsofnas, A.Y.: On the adequacy of logical analysis to philosophical reasoning. Questions Philos. 5, 85–98 (2004) 16. Karande, A.M., Kalbande, D.R.: Weight assignment algorithms for designing fully connected neural network. IJISA 10(6), 68–76 (2018) 17. Dharmajee Rao, D.T.V., Ramana, K.V.: Winograd’s inequality: effectiveness for efficient training of deep neural networks. Int. J. Intell. Syst. Appl. 10(6), 49–58 (2018) 18. Hu, Z., Tereykovskiy, I.A., Tereykovska, L.O., Pogorelov, V.V.: Determination of structural parameters of multilayer perceptron designed to estimate parameters of technical systems. IJISA 9(10), 57–62 (2017) 19. Awadalla, H.A.: Spiking neural network and bull genetic algorithm for active vibration control. IJISA 10(2), 17–26 (2018)
A Hybrid Centralized-Peer Authentication System Inspired by Block-Chain Wasim Anabtawi, Ahmad Maqboul, and M. M. Othman Othman(B) Computer Science Department, An-Najah National University, Nablus, Palestine [email protected]
Abstract. The process of identifying and authenticating people is considered to be a crucial task nowadays. However, this ease of use comes with its own problems; like identity theft and identity fraud. Which, might be exploited by means like; the collusion of a corrupt authenticating organization, or internal fraud within the authenticating organization, or by a manipulative majority in case of multipleauthenticators. The proposed system will be an attempt to deflect such cases by establishing an unbiased organizational authentication system. This work illustrates a design of a hybrid authenticating system that does not have a single actor that carries out the whole authentication process. Instead, it will borrow the concept of authenticating peers from the block-chain technology and utilize it with the architecture of centralized authenticating system. Security analysis of the proposed system shows its effectiveness against various types of attacks and scenarios. Keywords: Authentication · Peer-Authentication · Centralized-Authentication · Block-Chain · Trust · Digital certificates
1 Introduction The use of the Internet and computers in the past years had eased the way transactions occur. A vast majority of these transactions include exchanging high valued assets, which are done over the Internet; according to [1]. And thus, the security of this process is considered to be a vital enabler for modern transactions. It is well-known that the Internet from security prospective has two main features; obscurity & openness [2, 3]. Which, provides an attractive circumstances for malicious users to commit frauds; by spoofing data or identity (Identity Theft). Recently there has been a continuous increase in the number of such frauds and identity thefts, as can be seen by the “Consumer Sentinel Network Data Book 2020 issued by the United States’ Federal Trade Commission (FTC) [4]. Which, tracks consumer fraud and identity theft complaints. It showed 45% increase over 2019 in fraud cases, which is correlated to the 113% increase in the identity theft cases. It is known that improper measures or improper protection of privacy by governments or organizations can be used directly or indirectly by malicious users in order to carry out identity thefts’ attacks. This can be more serious issue if we consider corruption in addition to the improper measures. According to © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 271–287, 2023. https://doi.org/10.1007/978-3-031-36115-9_26
272
W. Anabtawi et al.
the”Transparency International” in their”Corruption Perceptions Index 2020 [5]; more than two-thirds of the 180 scored countries scored 50/100 on the Corruption Perceptions Index (CPI), and the average score is 43/100. Given the previously mentioned reasons; the users/systems of the Internet will be left without enough guarantee on the honesty of the identity of the users. That’s why user identification and authentication have become an essential technology. User identification and authentication systems originally began as centralized systems; with digital certificates and keys being provided by centralized organization systems. In such systems of identification and authentication; digital certificates are used for users and systems in order to prove their identity. For example, [3] had declared that the whole purpose of an identity authentication system is to generate a digital certificate, which will be a proof for the identity of the subject. This has provided a type of protection to the user’s identity to some degree. However, systems faces various shortcomings. One of them is not guaranteeing the integrity or fairness of the centralized organizations. The users in the mentioned centralized systems rely solely on authority of these systems in order to carry out transactions. This means that the central authority has the ability to alter any data; because no distribution of authority is present. In addition to that, the centralized systems are bound to be the single point of failure. For the previously mentioned issues, it became more evident that there is a need for a decentralized system. One of the efforts to was the introduction of the block-chain technologies [6, 7]. Blok-chain is a decentralized system that distributes authority between the users of the system. Making it suitable for applications such as”Distributed Ledger Management for an Organization using Blockchains” [8]. Block-chain also implements the concept of”cooperative maintenance”, which makes all the nodes in the system share the same information regarding the identities of users and their authentication as mentioned by [1]. In addition, block-chain have an immutable database infrastructure; i.e. if any record in the block-chain is tampered with, the rest of the peer nodes would cancel the transaction related to that block-chain. Even though block-chain based systems provide a secure, transparent, un-breakable records of data according to [9], but it still has its disadvantages. The main disadvantage being the need of high energy consumption and big computational load in order to keep track of real-time ledger as mentioned by [10]. Another disadvantage of block-chain according to [12], that it’s vulnerable to the 51% attack; in which a group of miners exist where they control more than 50% of the network’s mining computing and hashing power. Which gives them the ability to prevent new transactions from being confirmed, and the ability to manipulate the block-chain records, and reverse transactions. Many efforts have been put to make use of the decentralized Block-Chain based authentication systems. Lin, He, and Huang et al. [13] made a secure mutual authentication system based on block-chain, in order to execute an access control policy. Cui, XUE, and Zhang et al. [14] Proposed a hybrid Block-Chain based identity authentication system regarding multi-wireless sensor networks, in which local block-chain was used to authenticate the identity of the nodes in the system. In 2017 [15] proposed an identity authentication system based on Bitcoin in block-chain, which suggested the identity management system to be built on the Bitcoin block-chain. Sen, Mukhopadhyay, et al.
A Hybrid Centralized-Peer Authentication System
273
[16] in 2021 proposed a Blockchain based Framework for Property Registration System. In which, the system relies on multiple endorsing peers that are different governmental and organizational authorities. As for the previously mentioned decentralized block-chain based systems, even though there is no single point of failure. However, those systems are vulnerable to the 51% attack, in addition to the burden of high computation requirements. The rest of this paper is organized as follows. Section 2 explains the motivation and goals sought in this work. Section 3 shows the detailed design of the proposed system. After that, Sect. 4 sheds more light about some special cases that might accrue during the use of the proposed system, and shows how the proposes system remedies those cases. Section 5 shows the evaluation of the proposed system in terms of applicability; by evaluating generated data, and the needed cryptographic operation. Furthermore, evaluation in terms of security is done in Sect. 6. Finally, Sect. 7 concludes the work done in this paper.
2 Motivation and Goals This paper proposes a user authentication system that borrows the concept of blockchain, while keeping the central governmental organization updated on what’s going on for legal purposes. The identification system is used to prove a Subject’s identity, without needing a certain central side that controls everything. The proposed system consists of three main actors, these actors all depend on each other. The first actor being the Subject asking for a certificate to prove their identity. Second actor will be a central government which will verify the Subject from its own side. Third actor will be a certain number of peers assigned to Subject which will verify the Subject’s identity for the Subject. In addition to the use of digital certificates, the distribution of authority mentioned in the paper will be an assurance that there won’t be any modifications from any side during a transaction. Moreover, the proposed system provides immunity to identity theft and identity fraud. And, preventing any exploits by means like; the collusion of a corrupt authenticating organization, or internal fraud within the authenticating organization, or by a manipulative majority in case of multiple-authenticators.
3 System Architecture and Design The proposed system design is based on a mobile application, which when the user asks for a certificate to be verified the application will produce an authenticated certificate having the user’s information. Which will be used for transactions and identity proof. 3.1 User Sign up For users to be added in the system, they need to sign up via the device, or they can go to a governmental organization which uses the proposed system. Subjects which are going to be identified will give their national ID, Full name, Date of birth, Address, Mobile number, Username, and password which will be used to log in the mobile application.
274
W. Anabtawi et al.
Adding users from a governmental side allows the government to do background checks and check for the preciseness of the given information, the user’s information will be added into the database of the central government and the proposed system as shown in Fig. 1, even though the users are added in the database, they won’t have the ability to request certificates until they’re validated from the governmental organization.
Fig. 1. User gives information to central
3.2 Initial User Validation When a new user logs in the application using their device, the application will generate a public and a private key to the Subject wanting to identified, which will both be stored in the Subject’s device. For an Subject to be validated in order to ask for certificates and participate in the system, after logging in the application, they will be asked to provide a password different from the one used to login the system, in addition to a bio-metric identifier, which will either be a finger ID, or a face ID. If the users don’t have a device with a touch ID, or a face ID scanner, they’ll go to a governmental center to provide them. After the user provides these private information, they will be hashed, and encrypted with their private key, and sent to the central governmental side. The user’s public key will be sent with these information as well, in which they’re saved in a table#2 containing them. Afterwards the central side will send a file to the user’s device containing the following: 1) Information of the peers that will verify the user. 2) Public keys of the peers that will verify the user.
A Hybrid Centralized-Peer Authentication System
275
3) Digital Signature #1 from the central side. Digital Signature #1 will contain: 1) Plain text which has ID of peers that will verify the users & Timestamp of when the peers got sent to the user. 2) Encrypted hash of the plain text using the central side private key. 3) Public key of the central side. This process is all illustrated in Fig. 2, and the contents of digital signature #1 are in Fig. 3.
Fig. 2. Initial User Validation Process.
When this process is finished, the user will exchange public keys with the central side successfully, the user is validated, he/she will be able to participate in the system, ask for certificates, and validate other peers. 3.3 Certificate from Central Server For users to obtain the authenticated certificate that verifies their identity, they firstly need to request a certificate from the central governmental organization, they could
276
W. Anabtawi et al.
Fig. 3. 1) and 2) are the contents of digital signature #1 from the central
either go to one of the governmental organizations located in their area, after that they’ll request the certificate in which they should give their bio-metric ID and the transaction password. The biometric ID and password will both get hashed then encrypted with the user’s private key and sent to the central server, the central server will then use the user’s public key to decrypt the password and the bio-metric ID. These information will be checked and compared with the data in table #2, if they match the user will have a certificate from the central server side and it will be stored on the user’s device. And if they don’t match, the user will not have the certificate. All of this process is illustrated in Fig. 4.
Fig. 4. Certificate from central side process.
A Hybrid Centralized-Peer Authentication System
277
3.4 Live Message #1 One of the essential parts in the design of the proposed system are the live messages, because it’s the way the user communicates with his/her generated peers, and the central. Side. Live Message #1 is sent from the user to the central side periodically each day, in order to check who will user verify while participating in the system. The message will contain the user’s ID, and after sending that message to the central side, the central side checks in table #3 for peers that are generated to the user. If there are peers that get verified by the user, then their public key, and ID are both sent to the user from the central side. This process is illustrated in Fig. 5. The purpose of this message is to generate peers that need verification from the user, which are known as the peers that will the user verify. When this process is done, the user will have both peers that verify him/her, and peers that needs verification by him/her.
Fig. 5. Live message #1 process.
3.5 Certificate from peers & Live message #2 This subsection has three parts, in which the process of getting the certificate from peers is explained, all three parts depend on each other, and for a user to get the certificate from the peers, he/she need to go through all three parts without any complications. 1) First Part of the process: The first part consists of the process of obtaining the authenticated certificate to get a certificate from peers, in which the user requests the certificate from the peers through the central side. The user starts by requesting a certificate from peers through the central side, in which the user firstly sends a digital signature that got sent to the user in the initial validation process. In addition to a message asking for a certificate, which is hashed and encrypted using the private key. Afterwards
278
W. Anabtawi et al.
the central side decrypts the signature and checks it, after checking for its authenticity, the central adds the ID of the user that wants the certificate, his/her peers, and the message the user sent into table #4.The message will contain a string with the following text: “I’m User A, and i need certificate from peers side”. The first half is illustrated in Fig. 6. The use of digital signature #1 in this process ensures that user really exists, and the use of the encrypted message is to request users without any complications.
Fig. 6. First part of obtaining certificate from peers.
2) Second part of the process: This is the second part of the process, which will talk about live message #2 and its connection to getting the certificate from the peers. Live Message #2 is a message that’s sent periodically over the day from peers, the purpose of this message is to ask the central side if there are any users that need verification. At this stage, the peers have users to verify, but these users didn’t ask for certificate, which makes it different than Live Message #1. When the central side receives the messages from the peers, the central side checks in table #4, when it finds users asking for verification in order to get certificate from peers, the central side will send the hashed & encrypted message from table #4 with the ID of the user asking to be verified to the peers in order to verify him/her. The application will check the files of the peers, and compare it to the received ID, if they match then the peers will decrypt the message using the user’s public key in their file. Afterwards the peers will send a message hashed & encrypted with their private keys, in addition to a timestamp to the central side. The central side will store the ID of the user who asked for a certificate, the ID of the peers that will verify him/her, and the confirmation message from the peers in table #5.This half is all illustrated in Fig. 7. The contents of the message that
A Hybrid Centralized-Peer Authentication System
279
the peers encrypt, hash and send to the central will consist of the following text:”The user who asked for peers certificate is verified by (Peer ID)”.
Fig. 7. Second part of obtaining certificate from peers.
3) Third Part of the process (Live Message #3): This sub-subsection is the third and final part of getting the certificate from the peers side. Live Message #3 is used by users in order to check if any peer had sent them any certificate. This message is sent periodically over the day to the central governmental side, in which the user asks the government to check for any new certificates. After the user asks the central to check for messages sent to him/her, the central will check table #5 mentioned before, and search it for messages. When the central takes the message from table #5, the message is firstly encrypted with the peer’s private key, then the central side private
280
W. Anabtawi et al.
key. Afterwards the message is sent to the user who needs the certificate from peers, the user decrypts it using the central public key then the peers’ public keys. This process is illustrated in Fig. 8.
Fig. 8. Third part of obtaining certificate from peers.
3.6 Confirmation with Central Side In this section, the user should have 2 certificates, the first one from the central side, and the second one being the one from the peers that verified him/her. After obtaining the certificates, the user will receive an automated confirmation message, in which it asks him/her to confirm the certificates. Upon hitting confirm in the message, an encrypted message will be sent to the central side which contains the user’s info, his/her certificates, and the timestamp of receiving them. The central side will then decrypt it using the user’s public key, afterwards the central side will save the message, the ID of the user, the beginning time of the certificate, and the end time of the certificate in table #6. Table #6 is used as a history for certificates, and can be checked in case of any possible complications. All of this process is illustrated in Fig. 9. Lastly a final message is sent to the user via the central side, that message will be later converted into an authenticated certificate which is explained in the next section, and the contents of the message are shown in Fig. 10. 3.7 Creation of the Authenticated Certificate This is the final process of obtaining the authenticated certificate, which contains: 1) Certificate from the central side.
A Hybrid Centralized-Peer Authentication System
281
Fig. 9. Confirmation with the central side process.
Fig. 10. Contents of the final message the user gets, which will be used in the making of the authenticated certificate.
282
W. Anabtawi et al.
2) Certificate from the peers side. 3) Digital Signature #1 which was sent to the user from the central side, as shown in Fig. 3. 4) The confirmation message that’s sent to user from table #5 via central side as shown in Fig. 8. 5) The final confirmation message from central side as shown in Fig. 9. 6) The message that’s sent from the user to the central side upon confirming the certificates, which is shown in Fig. 9 in the confirmation with central process. Those elements will all be converted into an authenticated certificate, which will have a life span, in which the user should use before it expires. The contents of the authenticated certificate are illustrated in Fig. 11, Each number represents the element corresponding to the list mentioned in this section..
Fig. 11. Contents of the authenticated certificate.
4 Scenarios Different scenarios occur while users are using the system, this section will mention and discuss some of them with their solution in the proposed system. 4.1 No Certificate from the Central Side When the Subject wanting to be verified requests a certificate from the central governmental side, and the governmental side doesn’t send them one, the information the
A Hybrid Centralized-Peer Authentication System
283
Subject provided could be wrong, or simply the central side doesn’t want to provide them with a certificate for any reason. That’s why governmental signatures exist, these signatures are generated to the user’s device as soon as they install the application on his device. These signatures are a definite proof that the user has peers and could ask for certificates from the central side and the peers’ side. The user can check what’s wrong with the certificate process from central side by showing them the signature on his device. The user might take legal action if the governmental side doesn’t want to give him/her the certificate for no reason. If the user is not authorized to do transactions because of his suspicious background, they can simply do a background check before he’s added to the system, which won’t allow him/her to ask for certificates and get an authenticated certificate in the first place. 4.2 Peers Might be Offline When users ask for a certificate from peers side and the users might be offline, that’s why the peers send different types of messages during the period of the day. So when a user requests a certificate from the peers and they don’t respond for a while or don’t respond at all, the user will check what’s going on by asking the central side. Because the user has their information, and the governmental signature that proves that the peers in fact exist and should send him the message back in order to get the certificate..
5 Evaluation and Discussion The evaluation process for the proposed system began by testing multiple scenarios using two machines, the first machine which was a mac-book pro 2013 using the macOS Catalina V 10.15.7, with 16GB of RAM and 2.7 GHz Quad- Core Intel Core i7 Processor. The Mac-book pro was used as the hosting server for the scenarios. The second machine used was a Dell Latitude E7250, with a 2.3 GHz Core i5 processor, with 8GB of ram, using the Windows 10 64-bit OS. The Dell laptop was used to log in to the user and ask for certificates from peers generated to him/he. 5.1 Message Size Before and After the Execution of Some Functions The second test was held by comparing the sizes of the messages when certain functions get executed to show the effect of the functions and services used in the proposed system. The functions tested were the following: 1) 2) 3) 4)
Validation function Get Certificate function Live MSG#2 Get Peer Certificate
The results with their differences are shown in Table 1. The results show the differences between the messages before and after the executions of functions, these results show the impact of using certain applications and services during the run of the proposed system. Because we used Flask application for the web server application, its effect is clear on the size of the message in Table 1.
284
W. Anabtawi et al. Table 1. Size Comparison
Function Name
MSG SIZE BEFORE
MSG SIZE AFTER
Difference
Validate
232
367
135
Get Certificate
232
1214
982
Live MSG #2
232
28
-204
Get Peer Certificate
232
1696
1464
5.2 Number of Hashing Functions and Encryption – Decryption Functions The final evaluation process was by counting the number of hashing functions and encryptions during the scenarios that require them. The results of this test are shown in Table 2. The results in Table 2 show number of hashing functions and encryptions which shows how many are needed for each scenario. For example, the initial user validation process requires the following: 1) 2 Hashing functions ( 1 from the user side, 1 from the central side, & none from the peers’ side). 2) 1 Encryption function from the central side only. 3) 1 Decryption function from the user side only. Table 2 also shows the sum of all operations for each side, for example in order for the user to all operations, the following amount of functions is required: 1) 6 Hashing functions. 2) 5 Encryption functions. 3) 6 Decryption functions. These results can be used to study the effect of the hashing algorithms used in the proposed system and how many are required.
6 System Security Analysis 6.1 Identity Fraud If a user’s device is stolen and the person that stole the device requests a certificate to get an authenticated certificate and do transactions based on a stolen identity, the proposed system requires him to log in using the username and password of the user, and if he knows them, he would still need the transaction password and the bio-metric ID in order to get any certificate. The proposed has multiple layers of security which provide a secure environment for the users to use.
A Hybrid Centralized-Peer Authentication System
285
Table 2. Number of Hash, Encryption & Decryption Function Operation
D: Decryption, E: Encryption, H: Hash Function
Total Sum on All Sides
User-Side
Central Server
Peer-Side
# Hashings
# Encryptions
# Decryptions
1D, 1H
1E, 1H
/
2
1
1
Certificate 1E, 1H from Central to User
1D
/
1
1
1
Certificate from Peers to User
2D, 1E, 1H
1D, 2E, 1D, 1E 1H
2
4
4
User wants Final Full Certificate
3D, 3E, 3H
3D, 3E, 1D, 2E, 1H 2H
5
8
7
Sum of All Operations
6D, 5E, 6H
5D, 6E, 2D, 3E, 1H 4H
Initial User Validation
6.2 Network Spoofing If someone decided to spoof the network while a request for any certificate was going on, if the person spoofing the network in a way gets their hand on any information. The information is all hashed and encrypted using the user’s private key. The encryption of messages and hashing them prevents any kind of tampering, and editing on the messages, and the use of timestamps in certificates and messages prevents any manipulation from happening. Which guarantees the inability for anyone to decrypt and unhash these information to use them against users and hack their accounts. 6.3 Replays in Certificates The proposed system uses timestamps with digital signatures from the governmental organizations side in order to prevent users from getting multiple certificates. For example, if a user finds a way to manipulate the proposed system in order to get more than one certificate from the central side for example, the proposed system will prevent him from doing that by checking through the timestamps and certificates the users get, in addition to the certificate history table mentioned in the design section.
7 Conclusion The whole purpose of founding an unbiased citizen authentication system is to find solutions for problems like identity theft fraud, and the central government knowing too much about its citizens. This paper started by explaining different types of authentication
286
W. Anabtawi et al.
systems, their types, and which type of methods they have used. Then it talked about the technology of block-chain, when not to use it, and why. Afterwards, the paper talked about the proposed system, which borrows the concept of block-chain, and utilized it with the use of centralized governmental systems. The paper then talked about the architecture of the system, which is based on three main actors, all helping each other in order to achieve an unbiased citizen authentication system. Which aims to assure that no one uses the system for their own advantage, The paper then talked about different scenarios which may occur when using the system. Following that, the paper then explained the security analysis for the proposed system by explaining different features, which allow the relationship between the main three actors to be unbiased towards anyone. In the end, the paper talks about the evaluation/discussion process, in addition to an analysis regarding the security of the proposed system. The proposed system grants the users the ability to get certificates in order to obtain an authenticated certificate to complete transactions and prove their identity using a mobile device.
References 1. Fan, P., Liu, Y., Zhu, J., Fan, X., Wen, L.: Identity management security authentication based on blockchain technologies. Int. J. Netw. Secur. 21, 912–917 (2019) 2. Pfleeger, C., Pfleeger, S., Margulies, J.: Security in Computing. Pearson Education (2015). https://books.google.ps/books?id=VjMqBgAAQBAJ 3. Lin, Q., Yan, H., Huang, Z., Chen, W., Shen, J., Tang, Y.: An id-based linearly homomorphic signature scheme and its application in blockchain. IEEE Access 6, 20632–20640 (2018) 4. F. T. C. (FTC): Consumer Sentinel Network Data Book 2020 (2020). https://www.ftc.gov/ system/files/documents/reports/consumer-sentinel-network-data-book-2020/csn_annual_ data_book_2020.pdf, Accessed 1 Sep-2021 5. T. International: Corruption Perceptions Index 2020 (2020). https://images.transparencycdn. org/images/CPI2020_Report_EN_0802-WEB-1_2021-02-08-103053.pdf. Accessed 1 Sep 2021 6. Nakamoto, S.: Bitcoin: A peer-to-peer electronic cash system. In: Decentralized business review, p. 21260 (2008) 7. Anwar, S., Anayat, S., Butt, S., Butt, S., Saad, M.: Generation analysis of blockchain technology: bitcoin and ethereum. Int. J. Inform. Eng. Electr. Bus. (IJIEEB) 12(4), 30–39 (2020) 8. Pawade, D., Jape, S., Balasubramanian, R., Kulkarni, M., Sakhapara, A.: Distributed ledger management for an organization using blockchains. Int. J. Educ. Manag. Eng. 8(3), 1 (2018) 9. Jacobovitz, O.: Blockchain for identity management. In: The Lynne and William Frankel Center for Computer Science Department of Computer Science. Ben-Gurion University, Beer Sheva (2016) 10. Chakraborty, R., Pandey, M., Rautaray, S.: Managing computation load on a blockchain based multi-layered internet of things network. Procedia Comput. Sci. 132, 469–476 (2018) 11. Ali Syed, T., Alzahrani, A., Jan, S., Siddiqui, M.S., Nadeem, A., Alghamdi, T.: A comparative analysis of blockchain architecture and its applications: problems and recommendations. IEEE Access 7, 176838–176869 (2019) 12. Ye, C., Li, G., Cai, H., Gu, Y., Fukuda, A.: Analysis of security in blockchain: Case study in 51percent-attack detecting. In: Proceedings – 2018 5th International Conference on Dependable Systems and Their Applications, DSA 2018, ser. Proceedings – 2018 5th International
A Hybrid Centralized-Peer Authentication System
13.
14.
15.
16.
287
Conference on Dependable Systems and Their Applications, DSA 2018. United States: Institute of Electrical and Electronics Engineers Inc., Dec. 2018, pp. 15–24, 5th International Conference on Dependable Systems and Their Applications, DSA 2018; Conference date: 22-09-2018 Through 23-09-2018 Lin, C., He, D., Huang, X., Choo, K.-K.R., Vasilakos, A.: Bsein: A blockchain-based secure mutual authentication with fine-grained access control system for industry 4.0. J. Netw. Comput. Appl. 116, 42–52 (2018) Cui, Z., Xue, F., Zhang, S., Cai, X., Cao, Y., Zhang, W., Chen, J.: A hybrid blockchain-based identity authentication scheme for multi-wsn. IEEE Trans. Serv. Comput. 13(2), 241–251 (2020) Augot, D., Chabanne, H., Chenevier, T., George, W., Lambert, L.: A user-centric system for verified identities on the bitcoin blockchain. In: Garcia-Alfaro, J., Navarro-Arribas, G., Hartenstein, H., Herrera-Joancomartí, J. (eds.) Data Privacy Management, Cryptocurrencies and Blockchain Technology: ESORICS 2017 International Workshops, DPM 2017 and CBT 2017, Oslo, Norway, September 14-15, 2017, Proceedings, pp. 390–407. Springer International Publishing, Cham (2017). https://doi.org/10.1007/978-3-319-67816-0_22 Sen, S., Mukhopadhyay, S., Karforma, S.: A blockchain based framework for property registration system in e-governance. Int. J. Inform. Eng. Electron. Bus. 13(4), 30–46 (2021)
Heuristic Search for Nonlinear Substitutions for Cryptographic Applications Oleksandr Kuznetsov1,2(B) , Emanuele Frontoni1,3 , Sergey Kandiy1 , Oleksii Smirnov4 , Yuliia Ulianovska5 , and Olena Kobylianska2 1 Department of Political Sciences, Communication and International Relations, University of
Macerata, Via Crescimbeni, 30/32, 62100 Macerata, Italy [email protected], [email protected] 2 Department of Information and Communication Systems Security, Faculty of Comupter Science, V. N. Karazin Kharkiv National University, 4 Svobody Sq., Kharkiv 61022, Ukraine 3 Department of Information Engineering, Marche Polytechnic University, Via Brecce Bianche 12, 60131 Ancona, Italy 4 Cybersecurity & Software Academic Department, Central Ukrainian National Technical University, 8, University Avenue, Kropyvnytskyi 25006, Ukraine 5 Department of Computer Science and Software Engineering, University of Customs and Finance, Vernadskogo Street, 2/4, Dnipro 49000, Ukraine
Abstract. Heuristic algorithms are used to solve complex computational problems quickly in various computer applications. Such algorithms use heuristic functions that rank the search alternatives instead of a full enumeration of possible variants. The algorithm selects, at each iteration, an alternative with the best value for the heuristics. In this paper, we investigate the complex computational problem of finding highly nonlinear substitutions (S-boxes) in the space of 8-bit permutations. The generation of S-boxes is an important field of research, since nonlinear substitutions are widely used in various cryptographic applications. For instance, S-boxes in symmetric ciphers are responsible for cryptographic strength to linear, differential, algebraic, and other types of cryptanalysis. We propose new heuristics in the form of a cost function, calculated with the Walsh-Hadamard transform. We use the Hill Climbing Algorithm to find highly nonlinear substitutions. Our experiments demonstrate that the new heuristics give good results – it takes about 80,000 iterations of the algorithm to generate 8-bit S-boxes with a nonlinearity of 104. For the optimized parameters, the number of iterations of the algorithm is comparable to the best known results, which confirms the significance and value of the research. Keywords: Heuristic Search · Nonlinear Substitutions · Hill Climbing Algorithm · Cost Function · S-boxes Generation
1 Introduction Heuristic algorithms are used to rapidly solve complex computational problems in various computer science applications [1, 2]. The algorithms are based on the use of heuristic functions (or simply heuristics), which rank possible search alternatives. Heuristics are © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 288–298, 2023. https://doi.org/10.1007/978-3-031-36115-9_27
Heuristic Search for Nonlinear Substitutions
289
commonly represented as a cost function of the search [3–6]. In this case, the possible alternatives are ranked in ascending order of cost, i.e. the best alternative has the lowest value of the cost function. The search algorithm selects the best alternatives and iteratively repeats this process. Thus, instead of going through all the alternatives completely, the heuristic algorithm selects only some alternatives – the best ones according to the cost function value. This significantly reduces search complexity. However, it also does not guarantee the best, optimal solution. As a rule, heuristic algorithms give an approximate solution, but this solution would be found quickly enough [1, 2, 7]. In this paper, we discuss the complex computational problem of generating highly nonlinear substitutions (S-boxes) for cryptographic applications [8–10]. This is an important field of research, since S-boxes are used in many cryptographic algorithms to enhance strength against analytical attacks [11–13]. For example, the nonlinearity of S-boxes provides resistance to linear cryptanalysis [14]. Substitutions are required to be random, in order to be resistant to algebraic cryptanalysis. Consequently, substitutions should not contain hidden mathematical constructs, which describe the cipher with simple algebraic equations [15–17]. In this sense, the generation of random highly nonlinear S-boxes is a relevant computational problem [13, 18, 19]. We consider the generation of bijective permutations in the space of all 256! possible permutations of 8-bit numbers. This is a huge space of possible states, and it is computationally impossible to find a solution by trying all 256! ≈ 10507 alternatives. We use a heuristic search to find highly non-linear substitutions and propose new heuristics, based on calculating the Walsh-Hadamard spectrum. The goal of our research is to optimize the generation of random 8-bit S-boxes with non-linearity 104. Existing techniques require a large number of iterations of heuristic search algorithms. We hope that the new heuristics will improve the generation speed.
2 Related Works The problem of rapid generation of random S-boxes was considered in many related papers. For example, [8, 20–23] and many others investigated various heuristics for cryptographic applications, in particular iterative techniques for generating Boolean and vector functions, as well as cost functions for ranking different search alternatives. In further works these techniques were generalized and extended to various computational algorithms: simulated annealing [10, 24], hill climbing [8, 25, 26], genetic algorithms [27–30], artificial immunity techniques [31], etc. The search for various heuristic functions was also investigated in [30, 32, 33]. The WHS cost function optimisation was explored in [34, 35], and various algorithms with WHS and other heuristics were studied in [30, 36]. Reference [32] proposed a new cost function, which proved to be a more effective heuristic for finding S-boxes. Moreover, the authors used genetic algorithms and local search methods. Another heuristics, proposed in [33], are one of the most effective to date. As shown in [30], the use of this cost function gives the lowest computational complexity of the heuristic search. For instance, a local search algorithm requires about 150,000 iterations; a combination of a genetic algorithm and a search through a decision tree requires about 116,000 iterations. Meanwhile, the Hill Climbing (HC) algorithm has the lowest computational complexity. Combined with
290
O. Kuznetsov et al.
the cost function from [33], the HC algorithm requires about 70,000 iterations and this appears to be the best known result. Examination of the related works reveals that, in most cases, the authors solve the problem of rapid generation of highly nonlinear bijective S-boxes. Consequently, the authors search for a substitution with nonlinearity of 104 in the space of 256! ≈ 10507 possible permutations of 8-bit numbers. The nonlinearity of the substitution is calculated by the Walsh-Hadamard transform. In order to increase the nonlinearity of the S-box, the absolute values of the Walsh-Hadamard coefficients should be reduced [37]. Therefore, it is important to estimate the spectrum of coefficients, and this is implemented in all known heuristics.
3 Bijective 8-bit S-boxes In combinatorics, a permutation is an ordered set of numbers 0, 1, ..., N − 1. Such a set of 0, 1, ..., N − 1 corresponds to each number i with an i-th element from the se [38, 39]. Thus, a permutation is a bijection on the set {0, 1, ..., N − 1}. The number N is called the order (degree) of the permutation. We consider permutations of 8-bit numbers, i.e. for N = 28 = 256. Denote each element in the input of the permutation as Xi , i = 0, 1, ..., 255, and the elements in the output as Yj , j = 0, 1, ..., 255. Each Xi and Yj requires 8 bits to encode, as indicated in Fig. 1. The input binary vector (x0 , x1 , ..., x7 ) is fed to the select lines of the Multiplexer. Consequently, this vector specifies the number of one of the 256 outputs of the Xi , i = 0, 1, ..., 255 Multiplexer. Each Xi , i = 0, 1, ..., 255 is connected to one of the 256 permutations Yj , j = 0, 1, ..., 255 outputs. The demultiplexer performs the reverse conversion, encoding the output number Yj , j = 0, 1, ..., 255 of the permutation as a binary vector (y0 , y1 , ..., y7 ). Thus, a permutation of the set {0, 1, ..., N − 1}, N = 28 = 256 can be represented as a substitution of x0 x1 ... x7 , S= y0 y1 ... y7 which implements a vector mapping S : (x0 , x1 , ..., x7 ) → (y0 , y1 , ..., y7 ). The mathematical apparatus of Boolean functions is used to describe the S-box structure analytically [37]. Each binary substitution output is written as the value of a Boolean function: y0 = f0 (x0 , x1 , ..., x7 ), y1 = f1 (x0 , x1 , ..., x7 ), ... y7 = f7 (x0 , x1 , ..., x7 ). The apparatus of Boolean functions is very useful for evaluating the cryptographic properties of S-boxes. For example, the nonlinearity of the substitution N (S) is calculated using the Walsh-Hadamard transform.
Heuristic Search for Nonlinear Substitutions
x0
x1
291
x7 ... Multiplexer
X0
X1
X2
...
X255
Permutation Y0
Y1
Y2
...
Y255
Demultiplexer ... y0
y1
y7
Fig. 1. Data conversion scheme in an 8-bit bijective S-box
4 The Walsh-Hadamard Transform and Nonlinearity of Substitution Denote the scalar product of the binary vectors w = (w0 , w1 , ..., w7 ), and x = (x0 , x1 , ..., x7 ) as w, x = w0 x0 + w1 x1 + ... + w7 x7 Therefore, the Walsh-Hadamard transform of the Boolean function f (x) is given by the next expression: WHT (f (x), w) = (−1)f (x)⊕w,x (1) x∈{0,1}8
Expression (1) allows the calculation of Walsh-Hadamard coefficients WHT (f (x), w) for all 256 possible values of w = (w0 , w1 , ..., w7 ). In order to calculate the nonlinearity N (S), the Walsh-Hadamard transform should be performed for all linear combinations of coordinate Boolean functions f0 , f1 , ..., f7 . Denote the set of coordinate functions as a vector F = (f0 , f1 , ..., f7 ). Then the set of all linear combinations of Boolean functions f0 , f1 , ..., f7 is given by the scalar product of all binary vectors u = (u0 , u1 , ..., u7 ) and the vector F: u, F = u0 f0 + u1 f1 + ... + u7 f7 .
292
O. Kuznetsov et al.
For each of the 255 values u, F (except u = (0, 0, ..., 0)), perform the Walsh-Hadamard transform and find all 256 values of WHT (u, F, w). The maximum absolute value of the Walsh-Hadamard coefficients over all u = (0, 0, ..., 0) and w, denote as WHTmax = max |WHT (u, F, w)|. u=0,w
Hence, the nonlinearity N (S) is calculated according to the formula: N (S) = 27 −
WHTmax 2
(2)
Table 1 gives an example of the correspondence between some WHTmax and N (S) values. We are interested in highly nonlinear S-boxes. In particular, the related works most frequently solved the problem of finding S-boxes with N (S) ≥ 104, i.e. with WHTmax < 48. In this sense, the used heuristics should consider the values of the WalshHadamard coefficients (1), in order to minimise the cost of finding highly nonlinear substitutions. Table 1. Example of matching some WHTmax and N (S) values N (S)
WHTmax 80
88
76
90
72
92
68
94
64
96
60
98
56
100
52
102
48
104
44
106
5 New Heuristics for Finding Nonlinear Substitutions One of the first and most studied heuristic functions for finding highly nonlinear substitutions is the CFWHS cost function [24]: ||WHT (u, F, w)| − X |R , (3) CFWHS = u=0 w
where X and R – selectable heuristics parameters.
Heuristic Search for Nonlinear Substitutions
293
Research of function eq. (3), with optimisation of X i R parameters, was carried out in [27, 34, 35] and many other studies. However, to date, the CFWHS function is considered as an inefficient one [32, 33]. Another example of a cost function is the heuristics, proposed in [32]: CFPCF =
N
2−i W (S)k−i ,
(4)
i=1
where W (S) is a value vector |WHT (u, F, w)|, in which the i-th position contains the number of coefficients with values |4i|; k is a maximum position number with a non-zero value. The cost function (4) performed very well in [32]. At the same time, the authors used genetic algorithms and local search methods. The most effective heuristics for finding S-boxes, with N (S) ≥ 104, is the cost function from [33]: ||WHT (u, F, w)| − z|, (5) CFWCF = u=0 w z∈C
where C = {0, 4, ..., 32}. The use of function (5) in the local search algorithm requires about 150 thousand iterations, according to [33]. A genetic algorithm with a decision tree search requires about 116,000 iterations [33]. The HC algorithm showed the lowest computational complexity, requiring about 70,000 iterations [33]. In this paper, we offer a new cost function for the heuristic search for highly nonlinear substitutions. The proposed heuristics are as follows: CFnew =
2
|WHT (u,F,w)|−Y 4
,
(6)
u=0 w |WHT (u,F,w)>X
where X and Y – selectable parameters. The logic behind using parameter X in function (6) is to consider only those coefficients, which could potentially reduce the nonlinearity of the substitution. For instance, in order to achieve N (S) ≥ 104, it is necessary to reach WHTmax < 48, according to Table 1. Meanwhile, the values with |WHT (u, F, w)| 20 the search complexity increases, and for X > 32 we were unsuccessful in finding any substitutions with N (S) = 104. The observations reveal that: • it is necessary to consider almost all values of the coefficients of |WHT (u, F, w)|, in order to find highly nonlinear substitutions, i.e. the parameter X cannot be large; • the search complexity is almost independent of the Y parameter; • obtained results, for the number of iterations of the algorithm, look like a random process realisation. According to Table 2, the smallest number of iterations (on average) is observed for X = 12. In some cases (in bold) we managed to find substitutions with N (S) = 104 in about 70,000 iterations of the algorithm. This compares with the best known results from [30, 33]. Thus, this work advances the field of knowledge in the direction of improving cost functions for heuristic search of random S-boxes. This can be useful in various fields of computer science and cybersecurity [40–43], including the improvement of symmetric key encryption algorithms and hash functions [44]. Notably, in this paper, we used only the HC algorithm. Possibly, the cost function (6) will give an alternative result for other heuristic search algorithms. This is the subject of our further research. Acknowledgment. 1. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 101007820. 2. This publication reflects only the author’s view and the REA is not responsible for any use that may be made of the information it contains.
References 1. Banzhaf, W., Hu, T.: Evolutionary computation. In: Banzhaf, W., Hu, T. (eds.) Evolutionary Biology. Oxford University Press (2019). https://doi.org/10.1093/obo/9780199941728-0122 2. Gilli, M., Maringer, D., Schumann, E.: Chapter 13 – Heuristics: a tutorial. In: Gilli, M., Maringer, D., Schumann, E. (eds.) Numerical Methods and Optimization in Finance (Second Edition), pp. 319–353. Academic Press (2019). https://doi.org/10.1016/B978-0-12-8150658.00025-X 3. Gandomi, A.H., Yang, X.-S., Talatahari, S., Alavi, A.H.: Metaheuristic algorithms in modeling and optimization. In: Metaheuristic Applications in Structures and Infrastructures, pp. 1–24. Elsevier (2013). https://doi.org/10.1016/B978-0-12-398364-0.00001-2
296
O. Kuznetsov et al.
4. Çataloluk, H., Çelebı, F.V.: A heuristic algorithm for Chan-Vese model. In: 2018 26th Signal Processing and Communications Applications Conference (SIU). pp. 1–4 (2018). https://doi. org/10.1109/SIU.2018.8404820 5. van der Stockt, S.A.G., Engelbrecht, A.P., Cleghorn, C.W.: Heuristic space diversity measures for population-based hyper-heuristics. In: 2020 IEEE Congress on Evolutionary Computation (CEC), pp. 1–9 (2020). https://doi.org/10.1109/CEC48606.2020.9185719 6. Tunç, A., Ta¸sdemir, S., ¸ Sa˘g, T.: Comparison of heuristic and metaheuristic algorithms. In: 2022 7th International Conference on Computer Science and Engineering (UBMK). pp. 76–81 (2022). https://doi.org/10.1109/UBMK55850.2022.9919459 7. Salhi, S.: Hybridisation search. In: Heuristic Search, pp. 129–156. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-49355-8_5 8. Burnett, L.D.: Heuristic Optimization of Boolean Functions and Substitution Boxes for Cryptography (2005). https://eprints.qut.edu.au/16023/ 9. Álvarez-Cubero, J.: Vector Boolean Functions: applications in symmetric cryptography (2015). https://doi.org/10.13140/RG.2.2.12540.23685 10. McLaughlin, J.: Applications of search techniques to cryptanalysis and the construction of cipher components. https://etheses.whiterose.ac.uk/3674/ (2012) 11. Rodinko, M., Oliynykov, R., Gorbenko, Y.: Optimization of the high nonlinear s-boxes generation method. Tatra Mt. Math. Publ. 70, 93–105 (2017). https://doi.org/10.1515/tmmp-20170020 12. Biham, E., Perle, S.: Conditional linear cryptanalysis – cryptanalysis of DES with less than 242 complexity. In: IACR Transactions on Symmetric Cryptology, pp. 215–264 (2018). https:// doi.org/10.13154/tosc.v2018.i3.215-264 13. Freyre Echevarría, A.: Evolución híbrida de s-cajas no lineales resistentes a ataques de potencia (2020). https://doi.org/10.13140/RG.2.2.17037.77284/1 14. Mihailescu, M.I., Nita, S.L.: Linear and differential cryptanalysis. In: Mihailescu, M.I., Nita, S.L. (eds.) Pro Cryptography and Cryptanalysis with C++20: Creating and Programming Advanced Algorithms, pp. 387–409. Apress, Berkeley, CA (2021). https://doi.org/10.1007/ 978-1-4842-6586-4_19 15. Ars, G., Faugère, J.-C.: Algebraic Immunities of functions over finite fields. In: INRIA (2005) 16. Bard, G.V.: Algebraic Cryptanalysis. Springer US, Boston, MA (2009). https://doi.org/10. 1007/978-0-387-88757-9 17. Courtois, N.T., Bard, G.V.: Algebraic cryptanalysis of the data encryption standard. In: Galbraith, S.D. (ed.) Cryptography and Coding 2007. LNCS, vol. 4887, pp. 152–169. Springer, Heidelberg (2007). https://doi.org/10.1007/978-3-540-77272-9_10 18. Lisitskiy, K., Lisitska, I., Kuznetsov, A.: Cryptographically properties of random s-boxes. In: Proceedings of the 16th International Conference on ICT in Education, Research and Industrial Applications. Integration, Harmonization and Knowledge Transfer. Volume II: Workshops, Kharkiv, Ukraine, 06–10 Oct 2020, pp. 228–241 (2020) 19. Gorbenko, I., Kuznetsov, A., Gorbenko, Y., Pushkar’ov, A., Kotukh, Y., Kuznetsova, K.: Random s-boxes generation methods for symmetric cryptography. In: 2019 IEEE 2nd Ukraine Conference on Electrical and Computer Engineering (UKRCON), pp. 947–950 (2019). https:// doi.org/10.1109/UKRCON.2019.8879962 20. Clark, A.J.: Optimisation heuristics for cryptology. https://eprints.qut.edu.au/15777/ (1998) 21. Millan, W., Clark, A., Dawson, E.: Heuristic design of cryptographically strong balanced Boolean functions. In: Nyberg, K. (ed.) EUROCRYPT 1998. LNCS, vol. 1403, pp. 489–499. Springer, Heidelberg (1998). https://doi.org/10.1007/BFb0054148 22. Millan, W., Burnett, L., Carter, G., Clark, A., Dawson, E.: Evolutionary heuristics for finding cryptographically strong s-boxes. In: Varadharajan, V., Mu, Y. (eds.) ICICS 1999. LNCS, vol. 1726, pp. 263–274. Springer, Heidelberg (1999). https://doi.org/10.1007/978-3-540-479420_22
Heuristic Search for Nonlinear Substitutions
297
23. Millan, W., Clark, A., Dawson, E.: Boolean function design using hill climbing methods. In: Pieprzyk, J., Safavi-Naini, R., Seberry, J. (eds.) ACISP 1999. LNCS, vol. 1587, pp. 1–11. Springer, Heidelberg (1999). https://doi.org/10.1007/3-540-48970-3_1 24. Clark, J.A., Jacob, J.L., Stepney, S.: The design of S-boxes by simulated annealing. New Gener. Comput. 23, 219–231 (2005). https://doi.org/10.1007/BF03037656 25. Freyre-Echevarría, A., Martínez-Díaz, I., Pérez, C.M.L., Sosa-Gómez, G., Rojas, O.: Evolving nonlinear s-boxes with improved theoretical resilience to power attacks. IEEE Access 8, 202728–202737 (2020). https://doi.org/10.1109/ACCESS.2020.3035163 26. Kavut, S., Yücel, M.D.: Improved cost function in the design of boolean functions satisfying multiple criteria. In: Johansson, T., Maitra, S. (eds.) INDOCRYPT 2003. LNCS, vol. 2904, pp. 121–134. Springer, Heidelberg (2003). https://doi.org/10.1007/978-3-540-24582-7_9 27. Tesar, P.: A new method for generating high non-linearity s-boxes. Radioengineering 19, 23–26 (2010) 28. Ivanov, G., Nikolov, N., Nikova, S.: Reversed genetic algorithms for generation of bijective sboxes with good cryptographic properties. Cryptogr. Commun. 8(2), 247–276 (2016). https:// doi.org/10.1007/s12095-015-0170-5 29. Kapu´sci´nski, T., Nowicki, R.K., Napoli, C.: Application of genetic algorithms in the construction of invertible substitution boxes. In: Rutkowski, L., Korytkowski, M., Scherer, R., Tadeusiewicz, R., Zadeh, L.A., Zurada, J.M. (eds.) ICAISC 2016. LNCS (LNAI), vol. 9692, pp. 380–391. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-39378-0_33 30. Freyre-Echevarría, A., et al.: An external parameter independent novel cost function for evolving bijective substitution-boxes. Symmetry 12, 1896 (2020). https://doi.org/10.3390/ sym12111896 31. Ivanov, G., Nikolov, N., Nikova, S.: Cryptographically strong s-boxes generated by modified immune algorithm. In: Pasalic, E., Knudsen, L.R. (eds.) BalkanCryptSec 2015. LNCS, vol. 9540, pp. 31–42. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-29172-7_3 32. Picek, S., Cupic, M., Rotim, L.: A new cost function for evolution of s-boxes. Evol. Comput. 24, 695–718 (2016). https://doi.org/10.1162/EVCO_a_00191 33. Freyre Echevarría, A., Martínez Díaz, I.: A new cost function to improve nonlinearity of bijective S-boxes (2020) 34. Kuznetsov, A., et al.: WHS cost function for generating S-boxes. In: 2021 IEEE 8th International Conference on Problems of Infocommunications, Science and Technology (PIC S T), pp. 434–438 (2021). https://doi.org/10.1109/PICST54195.2021.9772133 35. Kuznetsov, A., et al.: Optimizing the local search algorithm for generating s-boxes. In: 2021 IEEE 8th International Conference on Problems of Infocommunications, Science and Technology (PIC S T), pp. 458–464 (2021). https://doi.org/10.1109/PICST54195.2021.977 2163 36. Kuznetsov, A., Wieclaw, L., Poluyanenko, N., Hamera, L., Kandiy, S., Lohachova, Y.: Optimization of a simulated annealing algorithm for s-boxes generating. Sensors 22, 6073 (2022). https://doi.org/10.3390/s22166073 37. Carlet, C.: Vectorial Boolean functions for cryptography. Boolean Models and Methods in Mathematics, Computer Science, and Engineering (2006) 38. Sachkov, V.N., Vatutin, V.A.: Probabilistic Methods in Combinatorial Analysis. Cambridge University Press (1997). https://doi.org/10.1017/CBO9780511666193 39. Sachkov, V.N., Kolchin, V.: Combinatorial Methods in Discrete Mathematics. Cambridge University Press (1996). https://doi.org/10.1017/CBO9780511666186 40. Beletsky, A.: Generalized galois-fibonacci matrix generators pseudo-random sequences. IJCNIS 13, 57–69 (2021). https://doi.org/10.5815/ijcnis.2021.06.05 41. Krasnobayev, V., Kuznetsov, A., Kuznetsova, K.: Synthesis of the structure of a computer system functioning in residual classes. Int. J. Comput. Netw. Inform. Secur. 15(1), 1–13 (2023). https://doi.org/10.5815/ijcnis.2023.01.01
298
O. Kuznetsov et al.
42. Iavich, M., Kuchukhidze, T., Gnatyuk, S., Fesenko, A.: Novel certification method for quantum random number generators. IJCNIS 13, 28–38 (2021). https://doi.org/10.5815/ijcnis. 2021.03.03 43. Shekhanin, K., Kuznetsov, A., Krasnobayev, V., Smirnov, O.: Detecting hidden information in fat. Int. J. Comput. Netw. Inf. Security. 12, 33–43 (2020). https://doi.org/10.5815/ijcnis. 2020.03.04 44. Kuznetsov, A., et al.: Performance analysis of cryptographic hash functions suitable for use in Blockchain. IJCNIS 13, 1–15 (2021). https://doi.org/10.5815/ijcnis.2021.02.01
Dangerous Landslide Suspectable Region Forecasting in Bangladesh – A Machine Learning Fusion Approach Khandaker Mohammad Mohi Uddin1(B) , Rownak Borhan1 , Elias Ur Rahman1 , Fateha Sharmin2 , and Saikat Islam Khan1 1 Department of Computer Science and Engineering, Dhaka International University,
Dhaka 1205, Bangladesh [email protected] 2 Department of Chemistry, University of Chittagong, Chittagong, Bangladesh
Abstract. Destruction caused by landslides in Bangladesh’s south-east highlights the importance of landslide-hazard mapping and a better understanding of the geomorphic development of landslide hazardous landscapes. Suitable relief, steep terrain gradients, Permiancyclothems and Pennsylvanian that weather into finegrained soils which have extensive clay, and appropriate precipitation are present in Chittagong, Bandarban, and other hillside regions of Bangladesh. In order to map landslide susceptibility for some of the areas in Rangamati District, Bangladesh, this study compares and evaluates a variety of machine learning models, including Random Forest (RF), Support Vector Machine (SVM), Naive Byes (NB), Random Forest (RF), Logistic Regression (LR), Decision Tree (DT), and the hybrid approaches. This study’s combination of logistic regression and random forest demonstrated 91.1% accuracy in landslide prediction. For risk management and disaster planning, this paper provides a useful analysis for selecting the best model for determining landslide susceptibility. Keywords: Landslide · Machine Learning fusion · geohazards · logistic regression · random forest
1 Introduction Landslides cause loss of life and property worldwide since they are one of the most dangerous geohazards. As a result, government and non-government organizations must move quickly to assess landslide vulnerability and mitigate its negative effects [1]. Bangladesh is particularly vulnerable to a wide variety of natural hazards and calamities. Particularly vulnerable to landslides in Bangladesh is the Chittagong Hill Tracts (CHT) [2]. Heavy rains during the monsoon season are the most likely cause of landslides in the hillside area. People play a unique part in the instability of hillslopes, which was previously unknown. As a result of this, there has been a rise in population, which has led to deforestation, unsustainable farming methods, urban expansion in Bangladesh, and a lack of effective government [2]. Landslide Risk in Bangladesh is shown in Fig. 1. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 299–309, 2023. https://doi.org/10.1007/978-3-031-36115-9_28
300
K. M. M. Uddin et al.
A landslide susceptibility map is a crucial tool for geohazard management through better decision in prone areas and planning for land use. These geographical traits are often referred to as “cause factors.” The landslide hazard assessment considers the determination of these cause elements to be the keystone [3]. Future landslides are thought to be possible in regions that have already experienced landslides and where the variables contributing to such landslides have recently produced favorable conditions for the triggering of landslides [4]. Several natural and man-made things, such as volcanic activity, groundwater extraction, rapid snow melting, long periods of rain, deforestation, hill cutting, changes in land usability, etc., can cause landslides, which is why the term “triggering factors” is used to describe them [5, 6]. Landslide risk can be measured in two ways: qualitatively and quantitatively [7].
Fig. 1. Landslide Risk prediction in Bangladesh
Expert analyses of the causes of the factors provide the foundation for qualitative assessments of landslide susceptibility. Quantitative methods, however, take use of the mathematical connections between the sites of landslides and their underlying causes. Mixed and semi-quantitative methodologies are frequently used in qualitative assessments to process the viewpoints of experts. Systems that employ machine learning are thought to be superior to methods based on expert judgment and analytical procedures
Dangerous Landslide Suspectable Region Forecasting
301
for the spatial prediction of landslides [1]. These methodologies evaluate landslide susceptibility by using algorithms for machine learning to look at the spatial correlation between previous landslide occurrences and a set of conditioning variables, from which the probable likelihood of landslide occurrence is determined. In many parts of the world, various machine learning techniques have been developed and put to use for creating landslide susceptibility maps, and artificial neural networks based on biological neural networks have been used to predict the geographic distributions of landslides [8]. Fuzzy inference techniques have been used to analyze the geographical distribution of landslides. Additionally, new strategies have been created, including Logistic Regression (LR) and Support Vector Machines (SVM). This work makes an effort to identify landslide hotspots in Bangladesh’s tropical region using machine learning approaches.
2 Literature Review The majority of natural hazards in mountainous regions are landslides. Landslides represent 9% of all natural disasters and are ranked as the third-worst natural disaster in the world. Researchers from all across the world have been attempting to forecast landslides. On the other hand, multivariate statistical tools determine how landslides are associated with a variety of potential causes. Multivariate approaches [9] include general additive models, logical regression, adaptive regression spline, and simple decision trees. Normality and collinearity assumptions make it difficult for bivariate and multivariate models to perform their intended functions. Machine learning-based structures are less restricted by these presumptions than traditional models are [10], which enables them to take into account the fact that landslides are not linear [11]. Bivariate and multivariate statistical models, according to some, frequently perform worse than machine learningbased models like gradient boosting, random forest, and support vector machines [12, 13]. LR has been extensively used to investigate landslides and has proven to be an effective method for predicting where landslides will occur. LR is a better method for mapping the susceptibility of landslides than probability value and multi-criteria decision – making process. Among the most widely used techniques for categorizing complex data is FLDA (Fisher’s linear discriminant analysis), which is straightforward enough to be used to a formal analysis of each data point in the projected region. Only a few researches have used FLDA to address landslide issues. Haojie Wang et al. [14] showed that the BN is seen as a promising way to figure out how dangerous something is. But it is still rarely used to figure out how dangerous a landslide is. Some of the studies on assessing landslides have also used the NB method successfully. It was claimed that this technique, which was used to forecast landslides in space, is an efficient machine learning method for determining landslide susceptibility. Random Forest (RF), K-Nearest Neighbor (KNN), and Extreme Gradient Boosting were three different methods of e-machine learning (ML) that were utilized by Yasin Wahid Rabbyet al [2]. The XGBoost model is the most recent of these three, and it is a brand-new method for modeling landslides that delivers highly precise results. However, both the RF and KNN models are often used and have adequate accuracy. The
302
K. M. M. Uddin et al.
performance of the cutting-edge machine learning algorithm Extreme Gradient Boosting (XGBoost) has not been compared to the published research nearly as frequently as it should be, despite the fact that RF and KNN have been used extremely frequently.
3 Methodology The techniques used in this study include data collection, data analysis, forecasting, and the creation and validation of landslide susceptibility maps. In data preparation, some data preprocessing technique, prediction machine learning approach. Algorithm 1 depicts the proposed method of this experiment to forecast the landslide in the Rangamati area. Step by step algorithm shows that data collection, data preprocessing, feature selection, and machine learning algorithm are the main functions and these functions will be described briefly later. The block diagram of this study is given Fig. 2.
Fig. 2. Block diagram of Landslide susceptibility’s proposed method.
Dangerous Landslide Suspectable Region Forecasting
303
3.1 Data Collection The Rangamati district is in the southeast of the Chittagong Hill Tracts. It is 6116.19 km2 in size (CHT). It is between the latitudes of 22°27 and 23°44 north and the longitudes of 91°56 and 92°33 east. There are 243,999 persons living in the research region, of whom 48% are Bengalis and 52% are tribes. There are 97 individuals per km2 and a male to female ratio of 1.1. (BBS 2011). The research region is 70.39 m above sea level on average. The study area has a tropical monsoon environment with moderate temperatures, significant rainfall, frequently high humidity, and obvious seasonal fluctuations because the country is in a tropical monsoon region. Rabby et al. [2] used the Rangamati landslide dataset and this dataset is actually used in this experiment also. Datasets metadata description is given in Table 1. 3.2 Data Preprocessing The likelihood of a landslide Data from Rangamati is used to forecast the location of the next landslide. Preprocessing has been done using the core Python modules NumPy, Pandas, and Matplotlib. The computation approach of mean, average, and max was utilized to locate and handle the missing data. To scale the characteristics in this research, a standardizing procedure was applied. With this technique, a dataset’s independent variables are kept inside a predetermined range. The standardized equation for scaling features is shown in (Eq. 1), where mean() returns the average value of feature x and std() returns the feature’s standard deviation. x = (x − mean(x))/std (x)
(1)
3.3 Machine Learning Models Predicting the class of a set of data points is the process of classification. There are numerous categorization algorithms available right now, but it is impossible to say which one is best. It relies on the application and type of data set that is readily available. Several machine classification algorithms were utilized in this investigation to find the intended outcome.
304
K. M. M. Uddin et al. Table 1. Metadata description of the Dataset
Metadata
Description
Aspect
An aspect map shows both the direction and grade of terrain at the same time [15]
Curvature
Plan curvature refers to the outlines on a topographic map or the curve of a slope in a horizontal surface [16]
Earthquake
In the event of an earthquake, the ground’s surface may shake and vibrate because of the propagation of seismic waves
Elevation
For the assessment of landslide vulnerability, elevations are frequently used. Elevation differences may be attributed to different environmental conditions as rainfall and plant species [9]
Flow
Landslides known as flows occur when fluid-like material slides down a slope [10]
Lithology
According to the results of several research, Lithology significantly affects the regional variance of landslide prevalence, kind, and depth [11]
NDVI
The RS readings are analyzed and the normalized difference vegetation index (NDVI), a straightforward graphical indication, is frequently used to determine whether or not the target being examined includes green, healthy vegetation [12]
NDWI
Surface water features can be distinguished using the normalized difference water index (NDWI) [13]
Plan Curvature
The curvature of a slope in a horizontal plane, or the curvature of contours on a topographic map, is referred to as plan curvature [14]
Profile Curvature The maximum slope’s direction is parallel to the profile’s curvature. Negative value depicts an upwardly convex surface at that particular cell. A positive profile shows that the cell’s surface is concave upward [17] Slope
Hillsides can have straight contours known as planar zones, convex outward plan curvatures known as noses, and concave outward plan curvatures known as hollows. Landslide debris congregates in hollows near the slope’s base in a small area [18]
Landslide
This metadata indicates the landslide risk [19]
3.3.1 Random Forest (RF) Random Forest is a widely used machine learning algorithm for supervised methods. It may be used in ML to solve classification and regression-related issues. The model’s performance is combined and improved using an ensemble of classifiers, which is based on the idea of ensemble learning. A group of decision trees are trained using the “bagging” approach to create the “forest.” The fundamental idea of the bagging technique is that by mixing many instructional modalities, pupils may perform better [19]. c fi(1 − fi) (2) i=1
Dangerous Landslide Suspectable Region Forecasting
305
3.3.2 Decision Tree (DT) The decision Tree method belongs to the class of algorithms known as “supervised learning algorithms”. The ability of the decision tree approach to address both regression and classification issues set it apart from other supervised learning techniques. By learning straightforward decision rules from prior data, a Decision Tree is used to build a training model that may be used to predict the class or value of the target variable (training data). In Decision Trees, we begin at the root of the tree to forecast the class label of a record. The root attribute and the record’s attribute values are contrasted. We go to the following node along the branch that corresponds to that value based on the comparison [20]. c −pi log2 pi (3) E(S) = i=1
3.3.3 Ensemble Technique An ensemble model is a method of using multiple machine learning models in a single model. It is used when we are not confident with our single model for prediction. It is a collection of weak models to produce more reasoning than a single model. It is a quite popular model for exploiting the strength of different models. Ensemble models can be of two types such as Homogeneous ensemble model and Heterogeneous ensemble model. The homogeneous ensemble model uses the same type of classifiers whereas the heterogeneous ensemble model uses different types of classifiers. Bagging and boosting are examples of a homogeneous ensemble model and stacking is of a heterogeneous ensemble model. Stacking is also known as blending. It is a stacked generalization that considers heterogeneous weak learners and learns them in parallel. The stacking model can be divided into two parts. One is called level 1 and another is level 2. The models that are used in level 1 are also known as the lower-level models and the model used in level 2 is also called the meta-model. The number of the model in lower-level depends on the model creator but the main fact that needs to be in mind is that the bigger the model the slower it will become. The output of the lower-level model is then passed down to the meta-model as an input. The meta-model can be any machine learning model of the creators’ choice. It can even be the same as one of the models that are used in the lower level. The input will be used to train the meta-model and the output of the meta-model will be considered as the final output. 3.3.4 Accuracy Measurement The confusion matrix has four outcomes that evaluate each classifier’s performance on positive and negative classes separately. These outcomes are true positive (TP), falsenegative (FN), true negative (TN), and false positive (FP). Equation 4 helps to calculate accuracy and Eq. 5 and 6 are used to calculate Precision, and Recall, respectively. F-1 score and Sensitivity are calculated by Eqs. 7 and 8, respectively [21, 22]. Accuracy =
(TP + TN ) (TP + FP + FN + TN )
(4)
TP (TP + TN )
(5)
Precision =
306
K. M. M. Uddin et al.
Recall = F1_score =
TP (TP + FN )
(6)
2 × Recall × Precision Recall + Precision
(7)
FP + FN TP + TN + FP + FN
(8)
Error rate =
4 Experiment and Results The experiment is carried out on a local computer using Jupyter Notebook, NumPy (1.0.1), Matplotlib (1.0.1), Scikit-Learn (1.0.1), and Pandas (1.11.0). Based on the experiment, two different sets of results have been analyzed. This experiment used different machine learning algorithms, including a decision tree (DT), a random forest (RF), and a Gaussian Naive Bayes (NB). After preprocessing the data and selecting the features to apply, SVM surpassed other machine learning classifiers with an accuracy of 75.99 percent. Then, to improve the accuracy, this study used a fusion method of machine learning algorithms. This experiment shows that this study uses NB, DT and RF, and DT and LR, DT and GB together to find the best solution.
Fig. 3. Comparison among machine learning classifiers based on accuracy for landslide forecasting in the tropical region in Bangladesh
For the purpose of forecasting landslides in Bangladesh’s tropical area, comparison of machine learning classifiers based on accuracy is shown in Fig. 3 and Table 2 depicts the performs of the different ML algorithms. By using the voting classifier, this classifier ensemble approach was made possible. With a 91.1 percent accuracy rate, the fusion of RF and LR did better than other fusions of machine learning classifiers.
Dangerous Landslide Suspectable Region Forecasting
307
Table 2. Accuracy, f-1 score, recall, precision and sensitivity of different ML algorithms. ML Algorithms
Accuracy
F-1 score
Recall
Precision
Sensitivity
Support Vector Machine
75.99
0.77
0.77
0.78
76%
Gaussian NB
72.44
0.75
0.76
0.76
62%
Logistic Regression
69.81
0.73
0.73
0.71
64%
Decision Tree
63.78
0.69
0.69
0.7
39%
Random Forrest
74.44
0.69
0.69
0.7
46%
Adaboost
67.47
0.73
0.81
0.81
58%
GradientBoostingClassifier
69.21
0.69
0.69
0.71
65%
RF and LR Fusion
91.1
0.86
0.84
0.85
74%
DT and RF
73.22
0.75
0.76
0.78
72%
LG and NB
77.65
0.76
0.77
0.78
60%
RF and NB
78.33
0.76
0.77
0.79
61%
RF,LG,NB,DT
77.94
0.78
0.78
0.8
73%
GB and AB
78.2
0.77
0.78
0.8
59%
Table 3 shows that the proposed fusion method performed better than various neural network approaches and other machine learning model approaches. The profile curvature and slope have the highest correlation of prediction in landslides, according to the proposed fusion technique approach. In every indicator, the suggested model outperformed the six benchmarks. The proposed model achieved the highest percentage of accuracy, scoring 91.1% overall. Table 3. Comparison table between various approach of finding landslide prediction References
Landslide Prediction Technique
Area
Accuracy
Mohammad Azarafza et al. [21]
CNN–DNN model
Iran
90.9%
Haojie Wang et al. [14]
CNN model
Hong Kong
92.5%
BinhThaiPham et al. [1]
NB model
India
90.1%
Proposed Method
Fusion approach of RF and LR
Bangladesh
91.1%
5 Discussion In this work, landslide susceptibility mapping in certain locations of Bangladesh was modeled using different machine learning techniques. Data preparation is essential for employing these techniques to get the better performance. The main objective of these models is to combine the output of different algorithms to obtain maximum accuracy. The
308
K. M. M. Uddin et al.
results demonstrate a considerable increase in landslide susceptibility prediction accuracy when compared to the benchmark models. Modeling the system was challenging because there weren’t many reference landslides in the data, the triggering factors’ data depended heavily on the spatial resolutions of satellite sensor imagery, and the quality of the DEM data directly impacted the quality of the input database, and the predictive model needed powerful processors to handle the inputs during landslide susceptibility assessments. SVM has 75.99 percent accuracy after preprocessing the data and selecting the features to use, outperforming other commonly used machine learning classifiers. The study then used a fusion method of machine learning algorithms to improve accuracy. This experiment demonstrates how this study combines SVM and NB, DT and RF, RF and LR, DT and GB to find the best solution. This study suggests that the areas which are in great risk of landslides are the South-East areas of Bangladesh. This study could pave the way for future machine learning research in geological data. This study employed a geological dataset from Bangladesh’s southeast that included information on landslide features. As a consequence, this study illustrates how machine learning algorithms work when a geological dataset is employed. Then again, this study reveals how altering some characteristics might change when different natural disasters occur.
6 Conclusions One of the most difficult jobs in geohazard assessments is mapping the vulnerability of slides. In this study, a novel fusion strategy of a prediction model is used to analyze the susceptibility to landslides in Rangamati, Bangladesh. Various triggering conditions and historical landslide data were used to fit the model. The suggested machine learning fusion model outperformed a wide range of benchmark approaches and provided accuracy of 91.1%. The statistical analysis and presentation of the predicted data show that Bangladesh’s hillside faces a serious risk of landslide. The study’s findings can be utilized to prepare in advance for any potential landslides in the south-east of the country. So, the study has the potential to save innumerable lives as well as the general public’s wealth.
References 1. Pham, B.T., Pradhan, B., Bui, D.T., Prakash, I., Dholakia, M.B.: A comparative study of different machine learning methods for landslide susceptibility assessment: a case study of Uttarakhand area (India). Environ. Model. Softw. 84, 240–250 (2016) 2. Rabby, Y.W., Hossain, M.B., Abedin, J.: Landslide susceptibility mapping in three Upazilas of Rangamati hill district Bangladesh: application and comparison of GIS-based machine learning methods. Geocarto Int. 37(12), 3371–3396 (2022) 3. Rabby, Y.W., Ishtiaque, A., Rahman, M.S.: Evaluating the effects of digital elevation models in landslide susceptibility mapping in Rangamati district, Bangladesh. Remote Sens. 12(17), 2718 (2020)
Dangerous Landslide Suspectable Region Forecasting
309
4. Guzzetti, F., Mondini, A.C., Cardinali, M., Fiorucci, F., Santangelo, M., Chang, K.T.: Landslide inventory maps: New tools for an old problem. Earth Sci. Rev. 112(1–2), 42–66 (2012) 5. Arora, M.K., Das Gupta, A.S., Gupta, R.P.: An artificial neural network approach for landslide hazard zonation in the Bhagirathi (Ganga) Valley, Himalayas. Int. J. Remote Sens. 25(3), 559–572 (2004) 6. Althuwaynee, O.F., Pradhan, B., Park, H.J., Lee, J.H.: A novel ensemble bivariate statistical evidential belief function with knowledge-based analytical hierarchy process and multivariate statistical logistic regression for landslide susceptibility mapping. CATENA 114, 21–36 (2014) 7. Aleotti, P., Chowdhury, R.: Landslide hazard assessment: summary review and new perspectives. Bull. Eng. Geol. Env. 58(1), 21–44 (1999) 8. Zare, M., Pourghasemi, H.R., Vafakhah, M., Pradhan, B.: Landslide susceptibility mapping at Vaz Watershed (Iran) using an artificial neural network model: a comparison between multilayer perceptron (MLP) and radial basic function (RBF) algorithms. Arab. J. Geosci. 6, 2873–2888 (2013) 9. Dou, J., et al.: Optimization of causative factors for landslide susceptibility evaluation using remote sensing and GIS data in parts of Niigata, Japan. PLOS ONE 10(7), e0133262 (2015) 10. Achour, Y., Pourghasemi, H.R.: How do machine learning techniques help in increasing accuracy of landslide susceptibility maps? Geosci. Front. 11(3), 871–883 (2020) 11. Henriques, C., Zêzere, J.L., Marques, F.: The role of the lithological setting on the landslide pattern and distribution. Eng. Geol. 189, 17–31 (2015) 12. Pettorelli, N.: The normalized Difference Vegetation Index. Oxford University Press (2013) 13. Ji, L., Zhang, L., Wylie, B.: Analysis of dynamic thresholds for the normalized difference water index. Photogramm. Eng. Remote. Sens. 75(11), 1307–1317 (2009) 14. Wang, H., Zhang, L., Yin, K., Luo, H., Li, J.: Landslide identification using machine learning. Geosci. Front. 12(1), 351–364 (2021) 15. Cellek, S.: The Effect of Aspect on Landslide and Its Relationship with Other Parameters. In Landslides. IntechOpen (2021) 16. Ohlmacher, G.C.: Plan curvature and landslide probability in regions dominated by earth flows and earth slides. Eng. Geol. 91(2–4), 117–134 (2007) 17. Dey, S.K., et al.: Prediction of dengue incidents using hospitalized patients, metrological and socio-economic data in Bangladesh: a machine learning approach. PLoS ONE 17(7), e0270933 (2022) 18. Schuldt, C., Laptev, I., Caputo, B.: Recognizing human actions: a local SVM approach. In: Proceedings of the 17th International Conference on Pattern Recognition, 2004. ICPR 2004, vol. 3, pp. 32–36. IEEE (2004) 19. Akyol, K., Karacı, A.: Comparing the performances of ensemble-classifiers to detect eye state. I.J. Inform. Technol. Comput. Sci. 14, 33–38 (2022) 20. Maharjan, M.: Comparative analysis of data mining methods to analyze personal loans using decision tree and naïve bayes classifier. In. J. Educ. Manage. Eng. 12(4), 33–42 (2022). https:// doi.org/10.5815/ijeme.2022.04.04 21. Latif, S., Dola, F.F., Afsar, M.D.M., Esha, I.J., Nandi, D.: Investigation of machine learning algorithms for network intrusion detection. Int. J. Inform. Eng. Electr. Bus. 14(2), 1–22 (2022) 22. Rahman, M.M., Rana, M.R., Alam, M.N.A., Khan, M.S.I., Uddin, K.M.M.: A web-based heart disease prediction system using machine learning algorithms. Netw. Biol. 12(2), 64–80 (2022)
New Cost Function for S-boxes Generation by Simulated Annealing Algorithm Oleksandr Kuznetsov1,2(B) , Emanuele Frontoni1,3 , Sergey Kandiy2 , Tetiana Smirnova4 , Serhii Prokopov5 , and Alisa Bilanovych2 1 Department of Political Sciences, Communication and International Relations, University of
Macerata, Via Crescimbeni, 30/32, 62100 Macerata, Italy [email protected], [email protected] 2 Department of Information and Communication Systems Security, Faculty of Computer Science, V. N. Karazin Kharkiv National University, 4 Svobody Sq., Kharkiv 61022, Ukraine 3 Department of Information Engineering, Marche Polytechnic University, Via Brecce Bianche 12, 60131 Ancona, Italy 4 Cybersecurity & Software Academic Department, Central Ukrainian National Technical University, 8, University Ave, Kropyvnytskyi 25006, Ukraine 5 Department of Economic and Information Security, Dnipropetrovsk State University of Internal Affairs. Ave. Gagarina 26, Dnipro 49005, Ukraine [email protected]
Abstract. The simulated annealing algorithm relates to heuristic techniques for approximating optimization problems. It is well suited to finding a solution in a discrete state space. This algorithm simulates the physical processes that occur during metal annealing. While the temperature of the metal is high, the atoms of the substance can pass between the cells of the crystal lattice. As the substance cools, the probability of transitions decreases, and the process freezes. This is simulated in the computational algorithm for finding the global optimum. At each iteration, the simulated annealing algorithm forms several alternatives and chooses one of them. While the temperature is high, the adoption of worsening alternatives is allowed. As the temperature drops, the probability of worsening steps decreases. Cost functions are used to evaluate and compare possible alternatives. We are looking at the computational problem of generating S-boxes in the space of all 8-bit permutations. This is a difficult task, because the search space is very large – there are 256! permutations. We offer a new cost function that reduces the difficulty of finding S-boxes by an algorithm simulated annealing. Generated S-boxes can be used in cryptographic applications, such as symmetric key encryption algorithms and hash functions. Our experiments show that the new cost function allows you to quickly generate highly nonlinear S-boxes for cryptographic applications. For the simulated annealing algorithm, we obtained the best result, and this is the main contribution of this work. Keywords: Global Optimization · Nonlinear substitution · Simulated Annealing Algorithm · Cost Function · S-boxes Generation
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 310–320, 2023. https://doi.org/10.1007/978-3-031-36115-9_29
New Cost Function for S-boxes Generation
311
1 Instruction Simulated annealing algorithm refers to the heuristic class of global optimum search [1, 2]. This algorithm is inspired by the physical processes that occur during the cooling of metals. After the metal is heated, slow cooling gives the atoms enough time to organize into stable structures of the crystal lattice. While the temperature of matter is high individual atoms can pass from one cell of the crystal lattice to another. However, as the metal cools, such events occur less frequently and gradually the process is frozen. For the computational algorithm for global optimum, the metal cooling process is simulated as simple formulas [3, 4]. Possible alternative in the search space is evaluated and ranked using the cost function. We set the initial temperature and calculate the probability of transition between the cells of the crystal lattice. In the search algorithm, this probability will define the adoption of the worst alternative. We are gradually reducing the temperature and the likelihood of adopting deteriorating alternatives is also diminishing. As a result, only improving alternatives will be accepted at low temperature and the process is frozen. Therefore, to implement heuristic search it is necessary to set several parameters of imitation annealing: • Initial temperature; • Cooling coefficient; • Cost function for comparison and ranking of alternatives. In this work we offer a new cost function to solve the problem of generating nonlinear substitutions (S-boxes) in the space of all 8-bit permutations. This is a difficult task, because the search space is very large – there are 256! permutations. We want to find high nonlinear substitutions that are used in cryptographic applications. For example, they are useful in symmetric key encryption algorithms, hashing functions, etc. Thus, the goal of this work is to develop a new cost function that is expected to reduce the computational complexity of the simulation annealing algorithm when searching for S-boxes.
2 Related Works The problem of generating highly nonlinear S-boxes for cryptographic applications has been solved in many related works. The works [5–7] used various heuristic techniques to search for cryptographic boolean functions and nonlinear substitutions. The [8] work used the simulated annealing algorithm to solve this problem. The use of the Walsh-Adamar transform value function was suggested as a heuristic. [9–11] uses genetic algorithms to generate S-boxes. [12, 13] uses local optimization methods. In [14, 15] the hill climbing algorithm was investigated. The [14] work examined techniques of artificial immunity. Last works summarize and develop this direction. In particular, the work [12] proposes a new cost function that has proved very effective for genetic algorithms and local optimization techniques. The works [13, 16] offer another cost function, which also showed high efficiency with the local search algorithm and
312
O. Kuznetsov et al.
with the hill climbing algorithm. [17] investigated several heuristic algorithms using different heuristics. In our last work [18] the simulated annealing algorithm was optimized. A simple cost function was used, based on the calculation of the maximum absolute values of the Walsh-Adamar coefficients. It should be noted that the difficulty of heuristic search of highly nonlinear S-boxes is still very high. For example, for the simulated annealing algorithm, the best known result from [18] is about 450,000 iterations to generate S-boxes with a nonlinearity of 104. In this paper we continue to investigate the algorithm of simulated annealing. We offer a new cost function and optimize its parameters for fast generation of 8-bit S-boxes with non-linearity 104.
3 Methods The simulated annealing algorithm is motivated by physical considerations. The most advantageous energy state of the thermodynamic system is determined by the Boltzmann distribution [19, 20]: p(Ei ) =
1 − kEiT e B , Z
(1)
where: • Ei – the energy of the i-th state; • kB – is Boltzmann constant; • Z – normalization constant defined in such a way that the sum of probabilities for all i is 1. The probability ratio of two states is called the Boltzmann factor, it usually depends only on the difference of energies of states [19, 20]: Ej −Ei p(Ei ) = e kB T . p(Ej )
(2)
The Boltzmann distribution shows that states with lower energy will always be more likely. Therefore, the Boltzmann distribution is usually used to solve a variety of tasks [19, 20]. Simulated annealing algorithm simulates physical processes of thermodynamic system [1–3]. We consider the task of finding the optimal solution in the discrete space of possible states. Let’s label each state Si . For example, for the non-linear substitution problem, Si will denote a specific 8-bit S-box from the space of all 256! permutations. Consider the cost function FCF , whose values rank states Si . In this way, FCF (Si ) is the energy analogue of the i-th state of Ei in formula (1). The probability of finding the system in the i- th state by analogy with formula (1) write in the form: p(Si ) =
1 − FCF (Si ) T e . Z
(3)
New Cost Function for S-boxes Generation
313
The probability ratio (Boltzmann factor analogue) should be written as. FCF (Sj )−FCF (Si ) p(Si ) T =e . p(Sj )
(4)
Formula (4) calculates the probability of a system moving from one state to another. Suppose that the current state of the system is Sj and we have the value of the cost function FCF (Sj ). If for an alternative state Si the value of the cost function satisfies the condition FCF (Sj ) − FCF (Si ) > 0, then Si is accepted as the current state: Sj ← Si . If the alternative state Si is degrading, i.e. if FCF (Sj ) − FCF (Si ) ≤ 0, then Si is accepted as the current state with probability (4). The probability (4) decreases when the temperature T drops (see Fig. 1). Figure 1 shows the cases: • • • •
δ δ δ δ
= −(FCF (Sj ) − FCF (Si )) = 1; = −(FCF (Sj ) − FCF (Si )) = 10; = −(FCF (Sj ) − FCF (Si )) = 100; = −(FCF (Sj ) − FCF (Si )) = 1000.
Thus, to initiate the simulated annealing algorithm, it is necessary to select the cost function FCF , whose values rank the states Si , as well as to optimize the algorithm by the temperature value T . High T temperatures allow the algorithm to accept deteriorating Sj ← Si states, for which FCF (Sj ) − FCF (Si ) ≤ 0. This is useful for leaving the local optimum. At the same time, the adoption of all worsening states leads to an accidental transition from one state to another state, without optimizing the cost function. For this reason, the T temperature is continuously reduced in proportion to the cooling coefficient α: Ti = α · Ti−1 , where Ti is the temperature used to calculate the probability eq. (4) on the i-th iteration of the algorithm. In the work [18] we optimized the algorithm of simulated annealing on various parameters. In particular, the most optimal parameters were: • Initial temperature T0 = 20000; • Cooling coefficient α = 0.95. For comparison and ranking of alternatives in the work [18] we used the cost function R max |WHT (v · S(x), u)| − X , (5) Fmax WHS (S) = v∈{0,1}8 ,v=0
u∈{0,1}8
314
O. Kuznetsov et al.
Fig. 1. Probability of taking deterioration Sj ← Si for different values of δ = −(FCF (Sj ) − FCF (Si ))
where: • X and R - configurable parameters; • WHT (f (x), u) - Walsh-Adamar transform coefficients computed with [21]: WHT (f (x), u) =
(−1)f (x)⊕u·x .
(6)
x∈{0,1}8
We consider the generation of 8-bit S-boxes, which are given as a vector function S(x) = {f0 (x), f1 (x), ..., f7 (x)}, x = (x0 , x1 , ..., x7 ). Cost function (5) inspired by heuristic from work [8]: ||WHT (v · S(x), u)| − X |R , FWHS (S) =
(7)
v∈{0,1}8 ,v=0 u∈{0,1}8
where X and R - configurable parameters. In this paper we offer a new cost function and show that its use allows to significantly reduce the number of iterations of the simulated annealing algorithm for generation of 8-bit S-boxes with nonlinearity 104.
New Cost Function for S-boxes Generation
315
4 New Cost Function and Experimental Results To solve the problem of finding non-linear 8-bit S-boxes, we offer to use the cost function: |WHT (v·S(x),u)|−Y 2 Fnew (S) = 2 , (8) v∈{0,1}8 ,v=0 u∈{0,1}8 |WHT (v·S(x),u)|>X
where X and R - configurable parameters. We have conducted numerous experiments with the cost function (8). To find substitutions with nonlinearity 104 we used the algorithm of simulated annealing with parameters of [18]. The aim of our experiments was to optimize the parameters X and R in (8), i.e. to find the combinations of parameters that require the least number of iterations of the search algorithm. The results of the experiments are shown in Tables 1, 2 and 3. In Fig. 2 the results of the experiments are visualized as a diagram. The Walsh-Adamar transform coefficients calculated by formula (6) are always multiples of 4. For this reason we only consider values of X and R in (8), that are multiples of 4. Moreover, values of |WHT (v · S(x), u)| can only be positive, therefore we only consider X ≥ 0. Table 1. The results of experiments X
Y −48
−44
−40
−36
−32
−28
−24
0
81510.5
82007.1
80159.0
84573.8
82512.3
83204.2
84098.6
4
76745.7
80152.2
77775.5
87041.4
78836.1
82196.8
72215.4
8
77896.1
77147.5
77016.4
78604.4
81195.6
84373.4
85636.3
12
81306.0
80311.2
76357.3
77183.3
85566.4
79498.4
86121.4
16
69922.7
80535.2
72299.1
72995.8
80595.1
75198.7
78454.0
20
78454.8
82545.1
76416.0
81484.9
77798.0
80743.3
81526.4
24
72819.7
74015.6
80954.0
67748.8
71932.4
73653.7
77709.8
28
78504.1
75821.3
88405.7
81132.8
83522.8
79287.4
77723.8
32
83205.8
80829.9
75084.4
82512.9
72930.8
80382.0
80026.1
36
82917.9
77044.3
81486.0
82037.7
78891.2
76541.9
78188.4
40
85203.4
74981.2
78713.3
80840.9
78588.0
83632.2
86047.3
44
145775.0
153850.1
140859.2
140564.6
143037.3
149241.1
142471.1
48
–
–
–
–
–
–
In this case, we consider only those terms in formula (8) for which. |WHT (v · S(x), u)| − Y ≥ 0,
–
316
O. Kuznetsov et al. Table 2. The results of experiments (continuation)
X
Y −20
−16
−12
−8
−4
0
4
0
80791.1
78309.7
80615.1
87944.2
77346.1
80537.1
0.0
4
74735.9
81232.5
72797.6
76619.7
73580.5
78049.7
70135.7
8
68934.1
80125.6
82580.5
82322.4
73958.8
77840.1
83358.6
12
81978.5
80578.8
84741.8
86711.7
76006.8
84126.6
75120.2
16
74791.3
83375.3
76082.4
75094.1
87580.5
79944.8
80719.5
20
82538.9
82066.4
82771.2
81455.4
86835.6
80526.6
75460.4
24
76279.5
71695.3
81258.0
83892.5
73082.3
73204.3
73171.7
28
78874.1
80055.2
73827.4
74075.7
85799.7
83360.9
76422.2
32
77249.3
76160.7
81596.4
76832.6
79585.0
88636.2
76864.1
36
83097.9
87752.7
84910.5
79248.4
76908.2
78458.9
75015.3
40
84667.1
90001.9
77207.2
81686.5
84297.8
84692.8
79382.6
44
145443.0
143764.6
146710.3
147690.3
155602.7
143380.4
147893.6
48
–
–
–
–
–
–
–
Table 3. The results of experiments (continuation) X
Y 8
12
16
20
24
28
32
36
40
44
48
0 4 8
79440.0
12
82124.2
82615.9
16
87901.7
75372.5
68098.1
20
78759.4
83094.7
84156.3
119882.9
24
68431.8
73465.9
79901.7
111417.2
–
28
78445.8
77738.1
80959.3
112657.6
–
–
32
81088.8
84978.2
78308.8
121169.3
–
–
–
36
80594.5
84572.1
84167.9
117877.9
–
–
–
–
40
81973.8
79641.8
82003.1
119002.4
–
–
–
–
–
44
145833.9
141513.1
146733.6
207419.9
–
–
–
–
–
–
–
–
–
–
–
–
48
–
–
–
–
–
New Cost Function for S-boxes Generation
317
Fig. 2. Average number of iterations of the simulated annealing algorithm with new cost function
i.e. only cases when Y ≤ X ≥ 0. Tables 1, 2 and 3 provide estimates of the number of iterations of the simulated annealing algorithm for generating 8-bit S-boxes with a nonlinearity of 104. Each value in the table is the average of 100 algorithm runs. Cells marked with the «-» symbol correspond to cases when in all 100 runs of the algorithm no S-boxes with non-linearity 104 were found.
5 Comparison of Results and Discussion The obtained results show that the new cost function gives high efficiency of search in a wide range of parameters X and R. On average, the simulated annealing algorithm requires about 80,000 iterations to find 8-bit S-boxes with a nonlinearity of 104. For X > 40 and Y > 16 the difficulty is increasing. For example, with X > 44 and Y > 20 we could not find any S-boxes with non-linearity 104. Table 4 shows the results of the comparison with known results. We compare the efficiency of the simulated annealing algorithm by the average number of iterations, as well as the probability of finding S-boxes. As can be seen from Table 4, the use of the new cost function allowed us to significantly reduce the complexity of S-boxes generation. Compared to the best known result for the simulated annealing algorithm of [18] we managed to reduce the average number of iterations by 5.6 times. The probability of finding 8-bit S-boxes with non-linearity 104 was 100%.
318
O. Kuznetsov et al.
Table 4. Comparison of 8-bit S-boxes search results using simulated annealing algorithm Nonlinearity of S-boxes
Generation probability
Average number of iterations
Works [14, 22]
102
0.5%
–
Work [23]
92
–
–
Work [24]
104
–
3 000 000
Work [18]
104
56.4%
450 000
Our work
104
100%
80 000
The best obtained result (highlighted in Table 1 in bold) was about 67,750 iterations, which roughly corresponds to the best known result for the hill climbing algorithm from [13, 16]. So we showed that the simulated annealing algorithm can produce very good results in generating nonlinear substitutions for cryptographic applications.
6 Conclusion Nonlinear substitution (S-boxes) play an important role in cryptography. They provide nonlinear data transformation and protect against some cryptanalytic attacks. For example, differential, linear algebraic, and many other types of cryptanalysis. Thus, high-nonlinear substitution generation is a topical field of research in modern computer science [25–27]. In this paper heuristic methods of search in space of discrete states were considered. We were looking at the simulated annealing algorithm, inspired by the physics of thermodynamic systems. Our research consisted in developing a new cost function for heuristic search of 8-bit S-boxes with nonlinearity of 104. We have conducted numerous experiments and have shown that the computational complexity of the simulated annealing algorithm can be significantly reduced. By the number of iterations of the search, our cost function shows better values (for the simulated annealing algorithm) and some better than other heuristic algorithms. A promising area for further research is the use of a new value function in other heuristic search algorithms, for example, in local optimization and hill climbing algorithms [28, 29]. Acknowledgment. 1. This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No. 101007820. 2. This publication reflects only the author’s view, and the REA is not responsible for any use that may be made of the information it contains.
New Cost Function for S-boxes Generation
319
References 1. Delahaye, D., Chaimatanan, S., Mongeau, M.: Simulated annealing: from basics to applications. In: Gendreau, M., Potvin, J.-Y. (eds.) Handbook of Metaheuristics. ISORMS, vol. 272, pp. 1–35. Springer, Cham (2019). https://doi.org/10.1007/978-3-319-91086-4_1 2. Eremia, M., Liu, C.-C., Edris, A.-A.: Heuristic optimization techniques. In: Advanced Solutions in Power Systems: HVDC, FACTS, and Artificial Intelligence, pp. 931–984. IEEE (2016). https://doi.org/10.1002/9781119175391.ch21 3. Kirkpatrick, S.: Optimization by simulated annealing: quantitative studies. J Stat Phys. 34, 975–986 (1984). https://doi.org/10.1007/BF01009452 4. Aarts, E.H.L., van Laarhoven, P.J.M.: Statistical cooling: a general approach to combinatorial optimization problems. Philips J. Res. 40, 193–226 (1985) 5. Millan, W., Clark, A., Dawson, E.: Heuristic design of cryptographically strong balanced Boolean functions. In: Nyberg, K. (ed.) EUROCRYPT 1998. LNCS, vol. 1403, pp. 489–499. Springer, Heidelberg (1998). https://doi.org/10.1007/BFb0054148 6. Millan, W., Burnett, L., Carter, G., Clark, A., Dawson, E.: Evolutionary heuristics for finding cryptographically strong s-boxes. In: Varadharajan, V., Mu, Y. (eds.) ICICS 1999. LNCS, vol. 1726, pp. 263–274. Springer, Heidelberg (1999). https://doi.org/10.1007/978-3-540-479420_22 7. Millan, W., Clark, A., Dawson, E.: Boolean function design using hill climbing methods. In: Pieprzyk, J., Safavi-Naini, R., Seberry, J. (eds.) ACISP 1999. LNCS, vol. 1587, pp. 1–11. Springer, Heidelberg (1999). https://doi.org/10.1007/3-540-48970-3_1 8. Clark, J.A., Jacob, J.L., Stepney, S.: The design of S-boxes by simulated annealing. New Gener Comput. 23, 219–231 (2005). https://doi.org/10.1007/BF03037656 9. Tesar, P.: A new method for generating high non-linearity s-boxes. Radioengineering 19, 23–26 (2010) 10. Ivanov, G., Nikolov, N., Nikova, S.: Reversed genetic algorithms for generation of bijective sboxes with good cryptographic properties. Cryptogr. Commun. 8(2), 247–276 (2016). https:// doi.org/10.1007/s12095-015-0170-5 11. Kapu´sci´nski, T., Nowicki, R.K., Napoli, C.: Application of genetic algorithms in the construction of invertible substitution boxes. In: Rutkowski, L., Korytkowski, M., Scherer, R., Tadeusiewicz, R., Zadeh, L.A., Zurada, J.M. (eds.) ICAISC 2016. LNCS (LNAI), vol. 9692, pp. 380–391. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-39378-0_33 12. Picek, S., Cupic, M., Rotim, L.: A new cost function for evolution of s-boxes. Evol. Comput. 24, 695–718 (2016). https://doi.org/10.1162/EVCO_a_00191 13. Freyre-Echevarría, A., et al.: An External parameter independent novel cost function for evolving bijective substitution-boxes. Symmetry 12, 1896 (2020). https://doi.org/10.3390/ sym12111896 14. Ivanov, G., Nikolov, N., Nikova, S.: Cryptographically strong s-boxes generated by modified immune algorithm. In: Pasalic, E., Knudsen, L.R. (eds.) BalkanCryptSec 2015. LNCS, vol. 9540, pp. 31–42. Springer, Cham (2016). https://doi.org/10.1007/978-3-319-29172-7_3 15. Freyre-Echevarría, A., Martínez-Díaz, I., Pérez, C.M.L., Sosa-Gómez, G., Rojas, O.: Evolving nonlinear s-boxes with improved theoretical resilience to power attacks. IEEE Access 8, 202728–202737 (2020). https://doi.org/10.1109/ACCESS.2020.3035163 16. Freyre Echevarría, A., Martínez Díaz, I.: A new cost function to improve nonlinearity of bijective S-boxes (2020) 17. McLaughlin, J.: Applications of search techniques to cryptanalysis and the construction of cipher components https://etheses.whiterose.ac.uk/3674/ (2012) 18. Kuznetsov, A., Wieclaw, L., Poluyanenko, N., Hamera, L., Kandiy, S., Lohachova, Y.: Optimization of a simulated annealing algorithm for s-boxes generating. Sensors 22, 6073 (2022). https://doi.org/10.3390/s22166073
320
O. Kuznetsov et al.
19. Klenke, A.: Wahrscheinlichkeitstheorie. Springer Berlin Heidelberg, Berlin, Heidelberg (2020). https://doi.org/10.1007/978-3-662-62089-2 20. Landau, L.D., Lifshitz, E.M.: Statistical Physics, vol. 5. Elsevier (2013) 21. Carlet, C.: Vectorial Boolean functions for cryptography. Boolean Models and Methods in Mathematics, Computer Science, and Engineering (2006) 22. Clark, J.A., Jacob, J.L., Stepney, S.: The design of s-boxes by simulated annealing. In: Proceedings of the 2004 Congress on Evolutionary Computation (IEEE Cat. No.04TH8753), vol. 2, pp. 1533–1537 (2004). https://doi.org/10.1109/CEC.2004.1331078 23. Wang, J., Zhu, Y., Zhou, C., Qi, Z.: Construction method and performance analysis of chaotic s-box based on a memorable simulated annealing algorithm. Symmetry 12, 2115 (2020). https://doi.org/10.3390/sym12122115 24. McLaughlin, J., Clark, J.A.: Using evolutionary computation to create vectorial Boolean functions with low differential uniformity and high nonlinearity. arXiv (2013). https://doi. org/10.48550/arXiv.1301.6972 25. Beletsky, A.: Generalized galois-fibonacci matrix generators pseudo-random sequences. IJCNIS 13, 57–69 (2021). https://doi.org/10.5815/ijcnis.2021.06.05 26. Kuznetsov, A., et al.: Performance analysis of cryptographic hash functions suitable for use in Blockchain. IJCNIS 13, 1–15 (2021). https://doi.org/10.5815/ijcnis.2021.02.01 27. Iavich, M., Kuchukhidze, T., Gnatyuk, S., Fesenko, A.: Novel certification method for quantum random number generators. IJCNIS 13, 28–38 (2021). https://doi.org/10.5815/ijcnis. 2021.03.03 28. Kuznetsov, A., et al.: Optimizing hill climbing algorithm for S-boxes generation. Electronics 12, 2338 (2023). https://doi.org/10.3390/electronics12102338 29. Kuznetsov, A., et al.: Optimizing the local search algorithm for generating s-boxes. In: 2021 IEEE 8th International Conference on Problems of Infocommunications, Science and Technology (PIC S T), pp. 458–464 (2021). https://doi.org/10.1109/PICST54195.2021.977 2163
Enriched Image Embeddings as a Combined Outputs from Different Layers of CNN for Various Image Similarity Problems More Precise Solution Volodymyr Kubytskyi and Taras Panchenko(B) Taras Shevchenko National University of Kyiv, Kyiv, Ukraine [email protected]
Abstract. The number of images is growing daily. Thus, the task of image similarity detection (also known as the de-duplication problem) becomes more and more strong and important for a wide range of applications. The complexity of the two images comparison is too high because of the structure of the data and the general complication of similarity understanding. There are a lot of approaches mentioned in this work have been developed to solve this class of tasks, including SIFT, SURF, ORB, key points, ResNet50-based and many others. But the quality of these methods measured by F 1 score is still non-satisfactory. In this work, we propose a solution for specified tasks, which exceeds the known state-of-the-art models in quality for image pairs similarity detection. The new image embedding, namely the descriptive image vector built as a combination of outputs at different layers of the proposed pre-trained and fine-tuned Convolutional Neural Network, based on ResNet50, is presented here. The model based on a Neural Network was trained over the introduced image embedding vectors over the pairs of similar images from prepared datasets to detect input images similarity. The measurements show that the proposed approach outperforms the state-of-the-art results by F 1 score for image similarity or images near-duplicate detection tasks, described in this work. The real-world applications for three specified sub-tasks from the real estate domain are considered in this work. The proposed model has significantly improved the existing solution for the considered tasks. It is also promising to help solve more general and connected tasks with higher result quality. Keywords: Image embedding · Image descriptor · Image near duplicates · Image near similarity · Convolutional Neural Network · Intermediate layer
1 Introduction Image processing technologies have been developing rapidly in recent years. Among the vivid examples are DALL-E [1] and DALL-E 2 [2] by Open AI – two versions of the AI system that can create realistic images and art, expand existing images beyond the original canvas, and make realistic edits to images. AI-based methods and models for image identification [3] exceeded the accuracy of human classification ability on average. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 321–333, 2023. https://doi.org/10.1007/978-3-031-36115-9_30
322
V. Kubytskyi and T. Panchenko
By having this tremendous progress in AI-based approaches (primarily artificial neural networks of modern structures) for solving different kinds of tasks with images (classification, clustering, object detection, and other kinds of processing), there are lots of other tasks, targeted at more specific aspects. For example, the detection of sketches, schematical layouts, near-duplicates (similar images) [4–7], images of the same scene or object from different perspectives or angles of view (a kind of similarity again), and many more. In this work, we concentrate on some of these specific tasks, highlight their importance, and propose a solution. Image representation is a crucial component in computer vision tasks, such as image retrieval and classification. The representation of an image should capture its essential features while being compact enough to be processed efficiently. One of the most commonly used image representations is hand-crafted features, such as SIFT [8, 9], which are not always perfect for tasks that require complex patterns in image recognition. In recent years, Convolutional Neural Networks (CNNs) have been shown to be effective in extracting powerful features from images, making them a popular choice for image representation. However, using the output of the final layer of a pre-trained CNN as image embeddings can result in large feature vectors with reduced interpretability. To overcome this issue, this paper focuses on using intermediate layers of pre-trained CNNs for image embeddings, which have been shown to provide compact representations that retain the essential features of the image while being computationally efficient. (Image embeddings are compact representations of an image that capture its essential features.) We explore using intermediate layers of pre-trained CNNs for image embeddings. This approach can provide powerful features compared to traditional image representations, as the intermediate layers have learned to recognize complex patterns in images during training on large datasets. The paper investigates the effectiveness of using intermediate layers as image embeddings in various tasks, such as image retrieval and classification. We also support the proposed heuristics with the calculated F 1 measure score. A wide range of methods for image similarity analysis were proposed in recent years [10–38]. An overview of these existing solutions is presented in the next section. They are ranged from key point-based, geometric invariant, CNN-based, embeddingsbased, local and global features oriented. But these methods do not always show acceptable results for real-world specific tasks. In this paper, we present the structure of the solution proposed for the image deduplication problem (also known as near-similar images search) [4–7], introduce new image embedding, describe the developed model in detail, and analyze the results obtained, and its applicative benefits for the LUN.ua, a well-known Ukrainian company in real estate domain. The main contribution of this work is a newly developed and proposed approach for the de-duplication problem (or image similarity detection task). The main idea of the model is a descriptive image vector (“embedding”) introduced as a combination of outputs at different layers of the proposed pre-trained and fine-tuned Neural Network, based on ResNet50 [3].
Enriched Image Embeddings as a Combined Outputs
323
The obtained result is promising because it exceeds the state of the art for the image de-duplication task [4, 5, 27–33] by using a new combination of outputs at different layers to construct an image embedding.
2 Tasks to Solve and State-of-the-Art Methods Overview In the real estate realm, there are many tasks with objects and their images. One of the important issues is the de-duplication problem, which means merging separate advertisements of the same object (building, flat, and other types). Different kinds of similarity should be applied here, including image similarity. As practice shows, this could help to fix this issue in good measure. We will consider the following three types of tasks. 2.1 Near Duplicate Images Identification of near duplicate images (in general) of various graphical contexts (see Fig. 1 for example). This is an important task to better understand that the real-estate object (flat or some other type) is the same in two different advertisements.
Fig. 1. Near duplicate images
2.2 Multi-Angle Views The second task is the identification of the same object viewed from different angles (different perspectives) like in Fig. 2. This also helps to identify the same real estate object from different advertisements and photos. 2.3 Schematical Layouts It is important to classify the images of a real estate object – is it a façade, a schematical plan, or a living room photo? Identification of floorplans like in Fig. 3 is an important task, but out of the scope of this paper. Here we concentrate on floorplan images similarity, which is our third task.
324
V. Kubytskyi and T. Panchenko
(a)
(b)
Fig. 2. Multi-angle views
2.4 Existing Approaches and the State-of-the-Art Overview To solve the issues mentioned above, a wide range of approaches has been developed in recent years. Let us overview the known methods and ideas behind them: • • • • • • • • •
sub-image retrieval [10, 11]; local-based binary representation [12]; keyframe identification with interest point matching and pattern learning [13]; keypoint-based with scale-rotation invariant pattern entropy analysis [14]; geometric invariant features [15]; color histogram, local complexity based on entropy [16]; min-hash and TF-IDF weighting [17], and other signatures [18]; affinity propagation [19]; CNN-based methods and ideas [20–22]: global and local features matching, and intermediate layers aggregation; • color histograms and locality-sensitive hashing [9], SIFT [8, 9], approximate set intersections between documents computing [9, 17]. Several datasets and benchmarks for progress tracking also appeared [23–26]. Some methods proposed recently are based on image embeddings [27–33]. It implies that a numeric vector, a kind of description, should be calculated (or built) for every image. This embedding (vector) should be distinctive to “catch” the specifics of each
Enriched Image Embeddings as a Combined Outputs
(a)
(b)
325
Fig. 3. Schematical layouts of flats
image. We can find similar models (namely, “embeddings”) for texts, audio, and other kinds of media to be used in data science widely in our days for different analyses. The best-known results [38] by the commonly used measure F 1 (which is the harmonic mean of precision and recall for image pairs true near-duplicates detected by the evaluated method) could be obtained by the following approaches [38]: • by image embedding formed by taking previous before the last layer of pre-trained ResNet50 (the image feature vector), • SIFT / SURF / ORB descriptors, • perceptual DCT hash, • image embedding formed by the combination of intermediate layers of ResNet50 (the proposed method). Now we are going to describe the proposed approach in detail.
3 Methodology. Image Embedding Constructing and Using to Solve the Tasks To construct the new proposed image embeddings, and thus to solve the tasks stated in the previous section, we have done the following: 1) built an image feature extractor (embedding builder),
326
V. Kubytskyi and T. Panchenko
2) adopted and used the distributed embeddings storage, 3) built a decision-making unit. In the next section, we will examine these real-world tasks application results. 3.1 Image Feature Extractor – Embedding Builder To create proposed image embeddings, a pre-trained ResNet50 [3] CNN model for image classification is used. The output layer is removed, and activations from intermediate layers are extracted and combined using concatenation to form a single feature vector, which serves as the image embedding. The image embedding is normalized to ensure that the magnitude of the features does not dominate the similarity measure. The resulting image embedding is robust to image transformations, brightness/contrast changes, and noise/watermarks and is used for comparison tasks. Detailed steps of image embedding creation with the combination of intermediate layers of a pre-trained ResNet50 are as follows. 1. Load the pre-trained ResNet50 model. The first step is to load the pre-trained ResNet50 model that has been trained on a large image dataset such as ImageNet [3]. This allows us to leverage the learned features from the training data to extract high/mid/low-level features from the input images. 2. Remove the final fully connected layer. The pre-trained ResNet50 model is designed for image classification. Therefore, it has a final fully connected layer that outputs the class probabilities. This layer is removed as we do not need it for our image embedding task. 3. Extract intermediate layer activations. We extract the activations of the intermediate layers of the ResNet50 model, as these layers contain different types of features at different levels of abstraction. These activations can be seen as feature maps representing the input image’s feature activations at each layer. To extract the intermediate layer activations, we pass the input image through the ResNet50 model and record the activations at each intermediate layer. Choosing which intermediate layers to extract. The choice of which intermediate layers to extract is based on the desired level of abstraction and the complexity of the features needed for a specific task. In general, the activations from early layers of the CNN contain lower-level features such as edges, textures, and basic shapes, while the activations from later layers contain higher-level features such as objects and parts. The number of intermediate layers to extract depends on the desired size of the image embedding and the computational resources available. A larger image embedding vector may contain more information but may also require more computational resources. On the other hand, a smaller image embedding vector may contain less information but may be computationally less expensive. Generally, a trade-off is made between the size of the image embedding and the computational resources available. Experimentation with different numbers of intermediate layers can help determine the optimal size of the image embedding. For image comparison tasks, extracting activations from both early and late layers is beneficial, as this ensures that the image embedding captures both low-level and
Enriched Image Embeddings as a Combined Outputs
327
high-level features. We do also use mid-level features to enrich image embedding with more details. 4. Combining the intermediate layer activations. The intermediate layer activations are concatenated into a single vector, which forms the image embedding. The concatenation of activations from multiple intermediate layers is important as it captures the rich and diverse set of features from different levels of abstraction. The activations from each layer are first flattened into a one-dimensional vector and then concatenated along the feature dimension to form the final image embedding. There are several ways to combine intermediate layer features to form an image embedding: • Concatenation. Concatenation is the process of combining the intermediate layer features by appending them end-to-end to form a single long feature vector. The resulting feature vector is then used as the image embedding. Concatenation is a simple and straightforward method and has been used successfully in many image comparison tasks. The main advantage of concatenation is that it captures the rich and diverse set of features from different levels of abstraction in CNN. • Summation. Summation combines the intermediate layer features by adding them elementwise to form a single feature vector. Summation is computationally efficient as it requires fewer operations compared to concatenation. However, summation may lead to information loss as it combines features from different levels of abstraction into a single feature vector. • Maximum activation. Maximum activation combines the intermediate layer features by selecting the maximum activation value from each layer to form a single feature vector. Maximum activation is computationally efficient and can capture the most discriminative features from the intermediate layers. However, maximum activation may not capture the rich and diverse set of features from different levels of abstraction in the CNN. • Average activation. Average activation combines the intermediate layer features by averaging the activations from each layer to form a single feature vector. Average activation is computationally efficient and can capture the overall features from the intermediate layers. However, average activation may not capture the rich and diverse set of features from different levels of abstraction in the CNN. So, the choice of combining methods depends on the task and the desired level of abstraction in the image embedding. Concatenation is a widely used method for image comparison tasks as it captures the rich and diverse set of features from different levels of abstraction in the CNN, so we also use it. The resulting image embedding vector serves as a compact representation of the input image and can be used for comparison tasks. The image embedding vector is resistant to linear image transformations, brightness and contrast changes, and is robust to image noise and watermarks. By combining the intermediate layer activations, we can obtain a rich and diverse set of features that represent the input image. 5. Normalize the image embedding. The image embedding vector is normalized to ensure that the magnitudes of the elements do not dominate the similarity measure
328
V. Kubytskyi and T. Panchenko
used for comparison. This is important as it ensures that the embeddings are not affected by the scale of the features. As the result, the process of creating image embeddings involves inheriting a pretrained CNN for image classification, removing the output layer responsible for image classification, and combining intermediate layer features using a method such as concatenation to form a single feature vector that serves as an image embedding, which can then be used for image comparison tasks. Image embeddings can be used for various image comparison tasks such as image retrieval, image similarity, and image classification. For example, in an image retrieval task, given a query image, the image embeddings can be used to retrieve similar images from a database. In an image similarity task, image embeddings can be used to measure the similarity between two images. In an image classification task, the image embeddings can be used as input features for a classifier to classify images into different categories. 3.2 Embeddings Comparison, Search Optimization, and an Appropriate Storage For efficient retrieval of similar images, it is important to index the image embeddings in the database. Indexing is a technique that allows for fast searching of data in a database, and it can be implemented using various algorithms such as KD-Tree, Ball-Tree, or Annoy or DCT hash comparison [39]. DCT hash comparison can be used as a pre-filter for multi-million image comparison tasks to eliminate images that are significantly different from the query image. The DCT hash is a compact and robust representation of an image, obtained by applying the Discrete Cosine Transform (DCT) to a small, downscaled version of the image. The DCT coefficients are then quantized, and the resulting values are concatenated to form the DCT hash [40]. The DCT hash comparison can be performed by calculating the hamming distance between the query image’s DCT hash and the DCT hashes of the images stored in the database. The hamming distance is a measure of the difference between two binary strings and can be calculated as the number of positions at which the corresponding bits are different. Images with a high hamming distance can be considered significantly different and can be eliminated from the database before the image embedding comparison. This can greatly reduce the size of the database that needs to be searched, leading to improved efficiency. As for storing the image embeddings, a document-oriented database, such as MongoDB, is a suitable solution for storing the data. MongoDB is a NoSQL database that stores data in JSON-like documents and can handle large amounts of unstructured data. The image embeddings can be stored as values in the database, with unique identifiers, such as file names or image hashes, serving as the keys. By using MongoDB, the image embeddings can be easily queried and retrieved for comparison without the need for complex relational database management. Indexing can also be applied to the image embeddings to further improve the efficiency of the comparison process.
Enriched Image Embeddings as a Combined Outputs
329
The comparison results can be further processed and filtered based on the required similarity threshold or can be sent to the next block – the decision-making unit, which helps to apply the same image embeddings to different comparison tasks. 3.3 Decision-Making Unit To apply the previously built and stored image embeddings to a specific task of supervised image comparison, we propose to follow the next steps. 1. Collect a dataset of image pairs. First, we need to collect a dataset of image pairs that we want to compare. For example, if we want to compare near-duplicate images, we need to collect pairs of images that are considered to be near-duplicates and which are not considered as such. If we want to compare photos of rooms taken from different angles, we need to collect pairs of photos of the same room taken from different angles and photos of various rooms taken from random view angles. 2. Generate image embeddings. For each image in the sample dataset, we run the image embedding builder and form a vector that describes the image features. These image embeddings will be used as input for the decision-making unit. 3. Form a new vector for each image pair. Each pair from the sample forms a new vector, which is obtained by combining the image embeddings of the two images. The combination can be constructed using Euclidean distance, L 1 distance, cosine similarity, or a combination of these metrics. 4. Train a neural network. The new image pair vector thus formed should be used to train a new fully-connected N-layer neural network with one output neuron. The output of the neural network will be used as a decision on the similarity between the two images in the pair. 5. Fine-tune the network. The trained neural network can be further fine-tuned based on specific task requirements, such as the choice of a distance metric, the number of hidden layers, and the number of neurons in each layer (a set of hyper-parameters). By following these steps, we can apply the previously built image embeddings to a specific task of supervised image comparison, where the goal is to compare pairs of images and decide on their similarity.
4 Real-World Tasks Application Results and Comparison Let’s examine the results on three practical applications mentioned in the Introduction section, namely: finding near duplicate images, clustering rooms based on photos from different shooting angles, and identifying similar floorplans (or schematical layouts). To solve the proposed problem of near-duplicate identification, collecting a dataset of pairs of sample images was necessary. In the first case, we need to collect pairs that are considered to be near-duplicates and which are not considered as such. In the second, there are pairs where the photo of the same room is taken from different angles, and various rooms are taken from random view angles. For the third, we need pairs of floorplans of the same flat (real estate object).
330
V. Kubytskyi and T. Panchenko
LUN.ua have prepared 3 datasets for these three sub-types of the image comparison tasks: • near duplicate images of various graphical contexts (dataset size: 80 000 image pairs), • multi-angle photos of the rooms (dataset size: 12 500 image pairs), • schematical layouts / floorplans (dataset size: 8 800 image pairs). Then we run the embeddings builder and form a vector for each image in the sample dataset. Each pair from the sample forms its new vector, which is obtained by combining the Euclidean metrics, L 1 , and cosine distance. The new image pair vector thus formed has been used then to train a new fully-connected N-layer neural network with one output neuron. As a result of training, the decision block has been trained to compare pairs of images to get an expected result of similarity. Building in a proposed way the image embedding and applying it to the task of the image near duplicate detection showed incredible results on the private dataset (LUN real estate images) – 8–10 times fewer mistakes in near-duplicate determination compared to SIFT [8, 9], SURF [41], ORB [42] key points algorithms, and 3–6 times fewer mistakes in comparison with ResNet50 [3], the state of the art result in the field. The obtained comparative experimental results are presented in Table 1, where one can ensure that the proposed model outperforms the best-known results for these 3 tasks. Table 1. Results comparison (F 1 measure score) for the 3 tasks Task \ Method
(1)**
(2)
Near duplicate images of various graphical contexts (dataset size: 80 000 image pairs)
0.94*
0.84 0.82 0.76
Multi-angle photos of the rooms (dataset size: 12 500 image pairs) 0.87*
0.79 0.77 0.66
0.79*
0.70 0.67 0.66
Schematical layouts / floorplans (dataset size: 8 800 image pairs)
(3)
(4)
* the best results are in bold for each task (the higher F measure score – the better) 1 ** (1) the proposed method’s F score, (2) ResNet50 features vector F score, (3) 1 1 SIFT/SURF/ORB descriptors F 1 score, (4) Perceptual DCT hash F 1 score
Hence, the proposed approach demonstrates acceptable results and outperforms the best-known models by F 1 score, and now is used in production at LUN.ua. Detailed measurements of the F 1 score and other metrics can be found in [38]. The dataset description as well as more details on calculations could also be viewed there. We will not concentrate on these questions here.
5 Summary and Conclusion The three specific tasks of the image similarity detection general problem (or image de-duplication problem) were the research topic in this paper. The new image embedding was introduced in this work. The method of data preparation as well as the step-by-step algorithm for the proposed model construction was
Enriched Image Embeddings as a Combined Outputs
331
presented here. The usage of a combination of different layers outputs from a pre-trained ResNet50 model to obtain new image embeddings (with calibration and normalization) helped to train a decision-making unit (a neural network), which gave an acceptable solution for images similarity task, and specifically three applicative sub-tasks for the real-estate domain. The main difference between the proposed method from the related research is the new combination of outputs at different layers of the proposed pre-trained and fine-tuned Convolutional Neural Network, based on ResNet50, for the construction of a descriptive image vector called “image embedding”, which helped to exceed the state of the art results for the tasks specified by the commonly used F 1 measure. The proposed and described in the paper new enriched image embedding is the main contribution to the research field. Measurement of the solution quality, namely the F 1 score, assures that the suggested approach is appropriate and applicable to the considered tasks. The optimized model proved its high efficiency and has been used in LUN.ua Ukrainian company for real-world tasks solutions from the real estate domain. The obtained results demonstrated a new way to solve an important image similarity problem with better quality. This way the presented model advances the existing methods toward results with higher precision, which means more accurate image near-duplicate detection.
References 1. Ramesh, A., et al.: Zero-shot text-to-image generation. In: International Conference on Machine Learning, pp. 8821–8831. PMLR, (2021) 2. Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., Chen, M.: Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125 (2022) 3. Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., Fei-Fei, L.: ImageNet: a large-scale hierarchical image database. In: IEEE Computer Vision and Pattern Recognition, CVPR (2009) 4. Thyagharajan, K.K., Kalaiarasi, G.: A review on near-duplicate detection of images using computer vision techniques. Arch Computat. Methods Eng. 28(3), 897–916 (2021) 5. Kaur, G., Devgan, M.S.: Data deduplication methods: a review. Int. J. Inform. Technol. Comput. Sci. 9(10), 29–36 (2017) 6. Mohidul Islam, S.M., Debnath, R.: A comparative evaluation of feature extraction and similarity measurement methods for content-based image retrieval. Int. J. Image, Graph. Sig. Process. 12(6), 19–32 (2020) 7. Er. Numa Bajaj, Er. Jagbir Singh Gill, Rakesh Kumar. An Approach for Similarity Matching and Comparison in Content based Image Retrieval System[J]. IJIEEB, 2015, 7(5): 48–54 8. Lowe, D.G.: Distinctive image features from scale-invariant keypoints. Int. J. Comput. Vision 60(2), 91–110 (2004) 9. Chum, O., Philbin, J., Isard, M., Zisserman, A.: Scalable near identical image and shot detection. In: Proceedings of the 6th ACM International Conference on Image and Video Retrieval, pp. 549–556 (2007) 10. Ke, Y., Sukthankar, R., Huston, L.: Efficient near-duplicate detection and sub-image retrieval. ACM Multimedia 4(1), 1–5 (2004)
332
V. Kubytskyi and T. Panchenko
11. Ke, Y., Sukthankar, R., Huston, L.: An efficient parts-based near-duplicate and sub-image retrieval system. In: Proceedings of the 12th Annual ACM International Conference on Multimedia, pp. 869–876 (2004) 12. Nian, F., Li, T., Wu, X., Gao, Q., Li, F.: Efficient near-duplicate image detection with a local-based binary representation. Multimed. Tools Appl. 75(5), 2435–2452 (2016) 13. Zhao, W.L., Ngo, C.W., Tan, H.K., Wu, X.: Near-duplicate keyframe identification with interest point matching and pattern learning. IEEE Trans. Multimed. 9(5), 1037–1048 (2007) 14. Zhao, W.L., Ngo, C.W.: Scale-rotation invariant pattern entropy for keypoint-based nearduplicate detection. IEEE Trans. Image Process. 18(2), 412–423 (2009) 15. Lei, Y., Zheng, L., Huang, J.: Geometric invariant features in the Radon transform domain for near-duplicate image detection. Pattern Recogn. 47(11), 3630–3640 (2014) 16. Li, Y.: A fast algorithm for near-duplicate image detection. In: 2021 IEEE International Conference on Artificial Intelligence and Industrial Design, AIID: IEEE, pp. 360–363 (2021) 17. Chum, Q., Philbin, J., Zisserman, A.: Near duplicate image detection: min-hash and TF-IDF weighting. BMVC 810, 812–815 (2008) 18. Liu, L., Lu, Y., Suen, C.Y.: Variable-length signature for near-duplicate image matching. IEEE Trans. Image Process. 24(4), 1282–1296 (2015) 19. Xie, L., Tian, Q., Zhou, W., Zhang, B.: Fast and accurate near-duplicate image search with affinity propagation on the ImageWeb. Comput. Vis. Image Underst. 124, 31–41 (2014) 20. Zhou, Z., Lin, K., Cao, Y., Yang, C.N., Liu, Y.: Near-duplicate image detection system using coarse-to-fine matching scheme based on global and local CNN features. Mathematics 8(4), 644 (2020) 21. Kordopatis-Zilos, G., Papadopoulos, S., Patras, I., Kompatsiaris, Y.: Near-duplicate video retrieval by aggregating intermediate CNN layers. In: Amsaleg, L., Þór Guðmundsson, G., Gurrin, C., Þór Jónsson, B., Satoh, S. (eds.) Multimedia Modeling, pp. 251–263. Springer International Publishing, Cham (2017). https://doi.org/10.1007/978-3-319-51811-4_21 22. Zhang, Y., Zhang, S., Li, Y., Zhang, Y.: Single- and cross-modality near duplicate image pairs detection via spatial transformer comparing CNN. Sensors 21(1), 255 (2021) 23. Barz, B., Denzler, J.: Do we train on test data? Purging Cifar of near-duplicates. J. Imaging 6(6), 41 (2020) 24. Matatov, H., Naaman, M., Amir, O.: Dataset and case studies for visual near-duplicates detection in the context of social media. arXiv preprint arXiv:2203.07167 (2022) 25. Tralic, D., Zupancic, I., Grgic, S., Grgic, M.: CoMoFoD: New database for copy-move forgery detection. In: Proceedings of the 55th International Symposium ELMAR, pp. 49–54 (2013) 26. Morra, L., Lamberti, F.: Benchmarking unsupervised near-duplicate image detection. Expert Syst. Appl. 135, 313–326 (2019) 27. Barz, B., Denzler, J.: Hierarchy-based Image Embeddings for Semantic Image Retrieval. (2019). https://arxiv.org/pdf/1809.09924v4.pdf 28. Berman, M., J´egou, H., Vedaldi, A., Kokkinos, I., Douze, M.: MultiGrain: a unified image embedding for classes and instances. arXiv preprint (2019). https://arxiv.org/pdf/1902.055 09v2.pdf 29. Yu, Z., Zheng, J., Lian, D., Zhou, Z., Gao, S.: Single-Image Piece-wise Planar 3D Reconstruction via Associative Embedding. arXiv preprint (2019). https://arxiv.org/pdf/1902.097 77v3.pdf 30. Rau, A., Garcia-Hernando, G., Stoyanov, D., Brostow, G.J., Turmukhambetov, D.: Predicting Visual Overlap of Images Through Interpretable Non-Metric Box Embeddings. arXiv preprint (2020). https://arxiv.org/pdf/2008.05785v1.pdf 31. Feng, G., Hu, Z., Zhang, L., Lu, H.: Encoder Fusion Network with Co-Attention Embedding for Referring Image Segmentation. arXiv preprint (2021). https://arxiv.org/pdf/2105.01839v1. pdf
Enriched Image Embeddings as a Combined Outputs
333
32. Asadi-Aghbolaghi, M., Azad, R., Fathy, M., Escalera, S.: Multi-level Context Gating of Embedded Collective Knowledge for Medical Image Segmentation. arXiv preprint (2020). https://arxiv.org/pdf/2003.05056v1.pdf 33. He, K., Zhang, X., Ren, S., Sun, J.: Deep Residual Learning for Image Recognition. arXiv preprint (2015). https://doi.org/10.48550/arXiv.1512.03385 34. Lytvynenko, T.I., Panchenko, T.V., Redko, V.D.: Sales forecasting using data mining methods. Bull. Taras Shevchenko Natl. Univ. Kyiv, Ser.: Phys.-Math. Sci. 4, 148–155 (2015) 35. Bieda, I., Panchenko, T.: A systematic mapping study on artificial intelligence tools used in video editing. Int. J. Comput. Sci. Netw. Secur. 22(3), 312–318 (2022) 36. Bieda, I., Kisil, A., Panchenko, T.. An approach to scene change detection. In: Proceedings of the 11th IEEE International Conference on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications (IDAACS’2021), vol. 1, pp. 489–493 (2021) 37. Bieda, I., Panchenko, T.: A Comparison of scene change localization methods over the open video scene detection dataset. Int. J. Comput. Sci. Netw. Secur. 22(6), 1–6 (2022) 38. Kubytskyi, V., Panchenko, T.: An effective approach to image embeddings for e-commerce. In: Information Technologies and Implementations (2022) 39. Tang, Z., Yang, F., Huang, L., Zhang, X.: Robust image hashing with dominant DCT coefficients. Optik 125(18), 5102–5107 (2014) 40. Zeng, J.: A novel block-DCT and PCA based image perceptual hashing algorithm. arXiv preprint (2013). arXiv:1306.4079 41. Bay, H., Tuytelaars, T., Van Gool, L.: SURF: speeded up robust features. In: Leonardis, A., Bischof, H., Pinz, A. (eds.) ECCV 2006. LNCS, vol. 3951, pp. 404–417. Springer, Heidelberg (2006). https://doi.org/10.1007/11744023_32 42. Rublee, E., Rabaud, V., Konolige, K., Bradski, G.: ORB: An efficient alternative to SIFT or SURF. In: 2011 International Conference on Computer Vision, pp. 2564–2571. ICCV, Barcelona, Spain (2011)
A Novel Approach to Network Intrusion Detection with LR Stacking Model Mahnaz Jarin1(B) and A. S. M. Mostafizur Rahaman2 1 Department of ICT, Bangladesh University of Professionals, Dhaka, Bangladesh
[email protected] 2 Department of CSE, Jahangirnagar University, Dhaka, Bangladesh
Abstract. IDS is one of the most researched subjects of network security. Singleclassifier IDS tends to fail in many scenarios where ensemble or hybrid approaches have been adopted. In the ensemble approach, many techniques have been adopted that give high efficiency yet complicate the system even when to make the system less complicated widely used classifiers like SVM, KNN, RF etc. is vastly used while others lag behind. Keeping a view to that, we propose a stacking ensemble approach with Logistic regression classifier for network-based anomaly detection. Also, using it to detect both binary and multiclass classification, shows its multiclass usability. The model was evaluated on the NSL-KDD dataset, for both binary and multiclass classification. Achieving an accuracy of 98.80%, a precision of 98.80%, and a recall of 98.80% for binary classification. 98.66%, 98.63%, And 98.66% were respectively the accuracy, precision, and recall for the multiclass classification. Keywords: Logistic regression · intrusion detection · multiclass classification
1 Introduction Intrusion detection systems come in various forms. Some are built for host-based detection, and some are for network intrusion detection. These categories can be further divided into branches like signature-based detection, anomaly detection, etc. different machine learning, deep learning, and unsupervised learning techniques have been widely used in developing these systems. Ensemble models have been one of them. Researchers from [1–6] adopted many ensemble techniques that can produce a high detection rate, but with that it can make the system computationally complex, and heavy for many devices. Even they vastly used machine learning classifiers like SVM, RF, KNN, and gradient boost (XGB).but a few of them has explored another promising yet simple classifier like logistic regression (LR). The logistic regression classifier has lagged behind in use of intrusion detection especially as the Meta classifiers. Some approaches have adopted it in building the ensemble model like in [7] but it is used just as a weak learner. Again. [8] Used it for Host based detection specifically for cloud environment. These shows a gap of utility of LR in network based intrusion detection that can unfold improved detection rates. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 334–343, 2023. https://doi.org/10.1007/978-3-031-36115-9_31
A Novel Approach to Network Intrusion Detection
335
Also fill the gap that is sustaining. Leveraging the advantages of LR like fast training, efficient discrete output in network-ids is the primary objective of this contribution. With a view to that, this paper focused on achieving three goals, first building a network-based ensemble model with LR as the final estimator. Secondly, securing a decent detection rate both for binary and multiclass classification as LR works best for binary detections. Finally to build a system that would be lightweight and won’t jam daily use devices. [1] uses DT-ensemble SVM for binary classification on NSL-KDD dataset achieving an accuracy of 99.36% and false alarm rate(FAR) of 0.38%. [2] Proposes an approach using Particle Swarm optimizer (PSO) with SVM and KNN for intrusion detection having an accuracy of 92.82% on KDD Cup99 dataset. Kunal and Mohit [3] demonstrated a rank-based attribute selection method to reduce attributes and then used an ensemble approach with RF, KNN, j48graft, and REP Tree.99.68% and 99.72% accuracy was achieved for multiclass and binary classification accordingly. [4] proposes CFS-BA, using the ensemble of C4.5, Forest PA, and RF algorithms. This model was evaluated on the AWID, NSL-KDD, and CICIDS2017 datasets. Getting the highest accuracy of 99.89% for their approach in the CICIDS2017 dataset. [5] built an GBM-RF model for multiclass classification with highest 99.77% accuracy for DoS attack. [6] presents a KNN, extreme machine learning (ELM) based method for intrusion detection with 84.29% accuracy. Abbas et al. [8] presented an ensemble model for IoT that used LR, NB, and DT. They achieved 88.96% accuracy for binary classification and 88.92% accuracy for multiclass classification on the CICIDS2017 dataset was used. [8] Uses LR for host based detection applicable for cloud environment which has 97.51% accuracy. [9] Introduced a stacking model with SVM, KNN, RF, and LR. Achieving accuracy of 94% for emulated and 97% for the real-time dataset. Kumar [3] proposed a hybrid model using NN and MOGA. The model works in two phases and uses the major voting technique for the final output. It has an accuracy of 88% and 97% respectively for NSL_KDD and ISCX-2012. Bhati et al. [10] debated that XGBoost can smoothen the bias-variance tradeoff, by using this they proposed an ensemble approach evaluating on KDD CUP99 dataset. That achieved an accuracy of 99.95% for detecting normal data. Gao et al. [11] took the NSL-KDD dataset and proposed a MultiTree algorithm by coping up the size of the training set and building several decision trees. They used several base classifiers like RF, and KNN and introduced an adaptive voting algorithm. The MultiTree had an accuracy of 84.2%, final accuracy for the voting algorithm was 85.2%. [12] Presents a RF feature selection with XGB hybrid. From [13–17] shows different ensemble approaches where [16] had the highest accuracy of 99.81% for binary classification using adaboost reptree. [18] Experiments with RF- DT ensemble and feature selection for IoT on NF-ToNIoT-v2 NF-BoT-IoT-v2 datasets. [19] Uses the algorithms and Fusion of Multiple Classifier (FMC) and Fuzzy Ensemble Feature selection (FEFS). Kumar et al. [20] conducted a broad review on ensemble IDS, where they concluded that modern-day IDS needs the ability to handle a wide range of traffic on speedy networks. Also, they mentioned the challenges such as the lack of an ideal distributed test bed to examine the ids. They stressed the use of ensemble models in the field of IDS as they have the potential to
336
M. Jarin and A. S. M. M. Rahaman
deliver more effective results. [20] Proposed white box attacks against NIDS and their defense mechanisms. [21] On the NSL KDD dataset, a cost-effective feature selection IDS was proposed. Hybrid, wrapper, and filter techniques are applied in [22] for feature selection and SVM to classify for NIDS. 1.1 Stacking Model Among the three types of ensemble learning, we choose stacking approach for our model. Since the objective was to build an ensemble model giving utmost importance to Logistic Regression, the stacking model would help us to approach our work in two phases. Phase one consists of using base classifiers, the goal was to choose classifiers that would complement each other’s weaknesses, and won’t make the training set difficult to train. Since our final estimator was fixed, the base classifiers have to be efficient enough to make the final outcome better on average to avoid the poor choice of any classifier. As Logistic regression can’t work better with variables that aren’t linearly separable, the multiclass classification could raise an issue. Our model uses eight machine-learning classifiers as base models, these classifiers are Linear SVM since it’s computationally easy gives accuracy while there is a clear margin, and polynomial SVM helps to expand the classifier to a higher degree. KNN is used as its ideal for non-linear data, handles multiclass problems efficiently, easily trainable. RF for its class balance tendency and high performance. DT has the ability to handle multi output data, the using cost of the tree is logarithmic making it lightweight. XGB for its flexibility and accuracy performance. Also it can boost any loss function. LightGBM for faster training speed, giving better results for XGB. Finally, LR is used both in as a base classifier also a final estimator.
2 Methodology Our stacked approach uses eight classifiers to build new set of training data that would be used to identify unknown data. The base classifiers for say A1 , A2 , . . . , A8 are trained using the training dataset and each of the classifiers creates their own individual predictions for say P1 , P2 , . . . , Pn , these predictions form a new training dataset for the second classifier od also known as the meta classifier. The meta-classifier is trained on the newly made dataset and optimized it for errors. After training, the output of the meta classifier performs as a generalized model. In our model the base classifiers, for example, A1 , A2 , . . . , A8 are SVM (linear and polynomial), KNN, RF, LR, DT, XGB, and LightGBM. As the final estimator, we used LR. That works in, p = e(c
0 +c1 *x)
/(1 + e(c
0 +c1 *x)
)
(1)
The p is the predicted outcome and c0 is the intercept term or generally known as bias and c1 is the coefficient term for each single value input that is generated by the base learners. Figure 1 shows the flow of our system.
A Novel Approach to Network Intrusion Detection
337
Fig. 1. Stacking ensemble model
2.1 Dataset and Preprocessing To conduct the experiment the KDDtrain+ was chosen. This is from the NSL-KDD dataset group. This dataset consists of a total of 42 attributes, were 4 of which were non-numeric. It has 23 different attacks under the label attribute. For preprocessing, the dataset was checked for the number of rows and columns. Looked for if any null values were present. The appropriate column names were given to it. The 23 attacks were classified under their corresponding attack type. The four types were: DoS, probe, U2R, and R2L. Then, the numeric attributes were normalized. Among the four non-numeric attributes, the difficulty column was dropped. The remaining three, namely flag, protocol_type, and service went through one-hot encoding and a categorical dataframe was built. Two datasets were built, one was for binary classification, and the other was for multiclass. For the first one, the labels were changed to “normal” and “abnormal”. A binary dataframe was created with them and labels were encoded after that. A dataset with labels and encoded columns was created. For multiclass, the same process was repeated. Only here the attack labels are kept the same as the four categories assigned to them at first. Another label was “normal”. A dataset with multi-labels and encoded columns was created. 2.2 Feature Extraction and Model Training In this stage, from the binary dataset, the numeric attributes and the encoded label attributes were taken to create a dataframe. Using Pearson correlation, the features having greater than 0.5 correlations with encoded attack label features were selected. 9 attributes were found here. These attributes were taken and combined with the categorical dataframe of one-hot-encoded labels. Finally, for the final binary dataset real attack labels, one-hot-encoded and encoded features were combined. Again, the same steps were repeated for the multi-class dataset. The final dataset was saved. These two sets were read to be machine fed. For this experiment, these datasets were fed to SVM (Linear and Polynomial), KNN, RF, LR, DT, GradiantBoost, and LightGBM serially in two rounds with a ratio of 0.75:0.25 for training testing split respectively. First for binary classification, and second for multi-class. These models were cross-validated using tenfold cross-validation. RF had 500 estimators. KNN had
338
M. Jarin and A. S. M. M. Rahaman
K = 5 neighbors. Then, using LR as the final estimator, the stacking model was fed the input of all the predictions to deliver the output for binary and multiclass classification. 2.3 Experimental Setup and Sequence The meta-classifier was fed and the final result was calculated for binary and multiclass classification. The whole experiment was executed in google colab on a corei5 machine with 8 GB ram and with no external GPU. It was written using the python language. The experiment was evaluated for higher accuracy, low false negative rate. The accuracy, precision, recall and F1score was calculated first, then the confusion matrices were generated to show the exact number of correctly identified intrusions, missed intrusion etc. and were expected that would run smoothly on the machine. Also, first the accuracy. Evaluation Metrics: Accuracy and recall are the most crucial evaluation indicators for our IDS evaluation. Greater accuracy is achieved by correctly identifying intrusion across all samples; the lower the false intrusion rate, the higher the recall rate. However, confusion metrics were generated for each classifier and both classes. In addition to the accuracy, the F1 score and precision were calculated too. Accuracy = (TP + TN )/(TP + TN + FP + FN )
(2)
Recall(R) = TP/(TP + FN )
(3)
Precision(P) = TP/(TP + FP)
(4)
F1score = (2.R.P)/(R + P)
(5)
FNR = 1 − recall
(6)
Confusion Matrix: In this matrix, the TN = Intrusion Correctly Detected , FP = Incorrectly Detected , FN = Intrusion Missed , TP = Intrusion Detected (Table 1). Table 1. Confusion matrix
real
predict Negative positive
Negative
Positive
True negative ( ) False negative ) (
False positive ( ) True positive ( )
A Novel Approach to Network Intrusion Detection
339
3 Result Analysis Accuracy is the most important parameter for our result analysis. Secondly, we considered the recall rate from there the False Negative Rate (FNR) is calculated and finally from the confusion matrix the number of classified intrusions, missed intrusions is observed. Figure 2 shows the overall accuracy, recall, precision and F1 score for the stacking model for binary and multiclass classification.
Fig. 2. Performance of the stacking model
All the metrics showed the same results for both the classification that was 98.80%, from the recall the FNR for binary classification was 1.2%, and for multiclass the accuracy and precision was 98.66%, recall was 98.63% and F1 was 98.64%.was it was 1.37%. The training dataset had an accuracy of 99.08% for binary classification. For multiclass it was 98.99%. In Fig. 3, the precision, recall, and f1 scores for detecting normal and abnormal data detection are given for binary classification. This showed the model performed well in terms of accuracy and had the highest recall rate for the abnormal classification. In Fig. 4, from the confusion matrix for binary classification, it has shown that, among 16,674 samples it detected 16,546 intrusions correctly, also detected 219 false intrusions and missed 228 intrusions. The model performs well in detecting attacks, but the number of missed intrusions can be lower. For multiclass classification, based five categorical types the performance of the model is shown in Fig. 5. Also Fig. 6 shows the confusion matrix for multiclass classification. As per Fig. 5, the precision for detecting DoS attacks was 99.33%, for probe it was 96.99%, for R2L was 90.97%. Normal samples had precision of 98.64% and for U2R attack it was 20%. The recall rate was highest for the detection of normal data with
340
M. Jarin and A. S. M. M. Rahaman
Fig. 3. Binary detection metrics
Fig. 4. Confusion matrix of stacking Model (Binary)
99.14%.after that consequently there were Dos with 98.65%, probe with 97.35%, R2L with 88.92% and U2R with 6%. From the recall rate, the FNR for normal sample were 0.86%. Dos had FNR of 1.35%. For probe the FNR was 11.68%. It was highest for U2R attacks with 94%. In Fig. 6 the confusion matrix shows among the 15 U2R samples it detected only two correctly. 11,322 DoS attacks were identified correctly. 237 R2L attacks, 2862 probe attacks and 16,537 normal data correctly. The model was compared with some pre-existing models, in which it showed better performance in terms of accuracy for multi-class classification. The comparison is shown in Table 2. In Table 3, the comparison with some pre-existing models for binary classification in shown. From there it can be seen that, [16] had higher accuracy then our model with adaboost reptree approach. Analysis: Our stacking approach showed excellent results in terms of accuracy for binary and multiclass detection. It ran smoothly on the machine. It had high precision, recall and F1 score for both normal and abnormal data samples in binary classification. In multiclass classification, the model performed best in detecting the normal data then
A Novel Approach to Network Intrusion Detection
F1 98.64 99.14 98.89
R
20 6 10
90.97 88.32 89.63
96.99 97.35 97.17
99.33 98.65 98.99
P
341
DOS
PROBE
R2L
U2R NORMAL
Fig. 5. Attack detection metrics
Fig. 6. Confusion matrix of stacking Model (Multiclass)
Table 2. Comparison with other ensemble approaches for Multiclass classification Multiclass classification Author
Ensemble methods
Dataset
Accuracy
Adeel Abbas [8]
NB (M), DT, LR
CICIDS2017
88.96%
Xianwei Gao [11]
Ensemble voting
KDDtest+
85.2%
Majd latah [6]
Elm and KNN
KDDtest+
84.29%
Simone A. Ludwig [13]
Neural network ensemble
NSl-KDD
85.93% and 98.28%
Our proposed model
Stacking ensemble with LR
KDDtrain+
98.66%
DoS attacks, probe and R2L was moderately good but it showed terrible performance in detecting U2R attacks since the recall was very low and FNR rate was so high that interprets that the model would miss most of the U2R categorical attacks. While compared with four pre-existing ensemble approaches, LR- stcaking outperformed neural network ensemble, KNN and ELM, NB (M), DT, LR ensemble. It shows
342
M. Jarin and A. S. M. M. Rahaman Table 3. Comparison with other ensemble approaches for Binary classification
Binary Classification Author
Ensemble methods
Dataset
Accuracy
M A Jabbar [14]
ensemble classifier (RF + AODE)
kyoto
90.51%
Bayu Adhi Tama [15]
Two-step ensemble
NSL_KDD and UNSW-NB15
85.8%
Celestine Iwendi [16]
Adaboost Reptree
KDD CUP99
99.80%
Poulmanogo Illy [17]
Decision Tree Bagging Ensemble
KDDTest+
85.81%
Our proposed model
Stacking ensemble with LR
KDDtrain+
98.80%
the strong hold of LR to be used in other NIDS approaches. In binary classification, it got close to adaboost reptree approach, but outperformed others. Again, there are a lot of pre-existing approaches available that are also needed to be explored for having a better understanding of the quality of the IDS.
4 Conclusion and Future Work This paper proposed an ensemble-based stacking approach for intrusion detection with LR. It showed excellent performance not only in binary classification but also in the multiclass approach. Together Host-based approaches LR is fully capable of using in other Network-IDS. With other vastly used classifiers, this model showed the capability of LR to be of use too. It was fast, ran smoothly, and provides a high detection rate with low FNR. With the low detection rate of U2R attacks, we envision increasing the U2R detection rate. Also, the use of different datasets and introducing other ensemble techniques with LR is our goal eventually. We want the system to be tested in real-life network traffic.
References 1. Gu, J., et al.: A novel approach to intrusion detection using SVM ensemble with feature augmentation. Comput. Secur. 86, 53–62 (2019) 2. Aburomman, A.A., Ibne Reaz, M.B.: A novel SVM-KNN-PSO ensemble method for intrusion detection system. Appl. Soft Comput. 38, 360–372 (2016) 3. Kumar, G.: An improved ensemble approach for effective intrusion detection. The J. Supercomput. 75, 275–291 (2020) 4. Zhou, Y., Cheng, G., Jiang, S., et al.: Building an efficient intrusion detection system based on feature selection and ensemble classifier. Comput. Netw. 174, 107247 (2020) 5. Rajadurai, H., Gandhi, U.: A stacked ensemble learning model for intrusion detection in wireless network. Neural Comput. Appl. 34, 15387–15395 (2022)
A Novel Approach to Network Intrusion Detection
343
6. Latah, M., Toker, L.: Towards an efficient anomaly-based intrusion detection for softwaredefined networks. IET Netw. 7(6), 453–459 (2018) 7. Besharati, E., Naderan, M., Namjoo, E.: LR-HIDS: logistic regression host-based intrusion detection system for cloud environments. J. Ambient Intell. Human. Comput. 10(9), 3669– 3692 (2019) 8. Abbas, A., Khan, M.A., Latif, S., et al.: A new ensemble-based intrusion detection system for internet of things. Arab. J. Sci. Eng. 47, 1805–1819 (2022) 9. Rajagopal, S., Kundapur, P.P., Hareesha, K.S.: A Stacking ensemble for network intrusion detection using heterogeneous datasets. Secur. Commun. Netw. 2020, 1–9 (2020) 10. Bhati, S., Chugh, G., Al-Turjman, F., Bhati, N.S.: An improved ensemble based intrusion detection technique using XGBoost. Trans. Emerg. Telecommun. Technol. 32, e4076 (2021) 11. Gao, X., Shan, C., Hu, C., Niu, Z., Liu, Z.: An adaptive ensemble machine learning model for intrusion detection. IEEE Access 7, 82512–82521 (2019) 12. Faysal, J.A., et al.: XGB-RF: a hybrid machine learning approach for IoT intrusion detection. Telecom 3(1), 52–69 (2022) 13. Ludwig, S.A.: Applying a neural network ensemble to intrusion detection. J. Artif. Intell. Soft Comput. Res. 9(3), 177–188 (2018) 14. Jabbar, M.A., Aluvalu, R., et al.: RFAODE: a novel ensemble intrusion detection system. Procedia Comput. Sci. 115, 226–234 (2017) 15. Tama, B.A., Comuzzi, M., Rhee, K.-H.: TSE-IDS: a two-stage classifier ensemble for intelligent anomaly-based intrusion detection system. IEEE Access 7, 94497–94507 (2019) 16. Iwendi, C., Khan, S., Anajemba, J.H., Mittal, M., Alenezi, M., Alazab, M.: The use of ensemble models for multiple class and binary class classification for improving intrusion detection systems. Sensors 20(9), 2559 (2020) 17. Illy, P., Kaddoum, G., Miranda Moreira, C., Kaur, K., Garg, S.: Securing fog-to-things environment using intrusion detection system based on ensemble learning. In: IEEE Wireless Communications and Networking Conference (WCNC), pp. 1–7. Marrakesh, Morocco (2019) 18. Le, T.-T.-H., Kim, H., et al.: Classification and explanation for intrusion detection system based on ensemble trees and SHAP method. Sensors 22(3), 1154 (2022). https://doi.org/10. 3390/s22031154 19. Priyadarsini, P.I., Anuradha, G.: A novel ensemble modeling for intrusion detection system. Int. J. Electr. Comput. Eng. (IJECE) 10(2), 1963 (2020) 20. Kumar, G., Thakur, K., Ayyagari, M.R.: MLEsIDSs: machine learning-based ensembles for intrusion detection systems—a review. J. Supercomput. 76(11), 8938–8971 (2020) 21. Mukeri, A.F., Gaikwad, D.P.: Adversarial machine learning attacks and defenses in network intrusion detection systems. Int. J. Wireless Microwave Technol. 12(1), 12–21 (2022) 22. Khine, P.T.T., Win, H.P.P., et al.: New intrusion detection framework using cost sensitive classifier and features. Int. J. Wireless Microwave Technol. 12(1), 22–29 (2022). https://doi. org/10.5815/ijwmt.2022.01.03 23. Sakr, M.M., Tawfeeq, M.A., El-Sisi, A.B.: An efficiency optimization for network intrusion detection system. Int. J. Comput. Netw. Inform. Secur. 11(10), 1–11 (2019)
Boundary Refinement via Zoom-In Algorithm for Keyshot Video Summarization of Long Sequences Alexander Zarichkovyi(B) and Inna V. Stetsenko Igor Sikorsky Kyiv Polytechnic Institute, 37 Prospect Peremogy, Kyiv 03056, Ukraine [email protected]
Abstract. This study presents a novel algorithm for keyshot video summarization in the context of real-world videos, such as those found on platforms like YouTube or Facebook. Our proposed boundary refinement algorithm via zoom-in addresses the problem of finding keyshots in long sequences, which is a major challenge for real-world video summarization. Existing state-of-the-art techniques tend to subsample such videos with small frame rate, leading to a coarse prediction of the summarization boundaries, or make several predictions on overlapping frames, which results in significant computation overhead and degraded performance due to loss of the context. In contrast, our approach refines these boundaries by zooming in on selected fragments, resulting in a more accurate and efficient summary. Additionally, we propose a coarse-to-fine score alignment procedure that minimizes the score discrepancy between the different stages of algorithm resulting in improved quality. The developed algorithm was evaluated on two standard datasets, SumMe and TVSum, and achieved state-of-the-art performance with scores of 56.2% (+0.6%) and 63.1% (+0.5%), respectively. Our proposed method demonstrates significant improvements in keyshot video summarization and has the potential to enhance the effectiveness of video summarization in real-world applications. Keywords: Computer Vision · Video Processing · Keyshot Frame Selection · Long Video Summarization · Algorithms
1 Introduction Video summarization is a crucial task in computer vision, as it involves creating a condensed version of a video that effectively captures the most important and relevant information. The process is illustrated in Fig. 1. There are several reasons why video summarization is important in the field of computer vision. Firstly, with the exponential growth of video content available on the internet [1, 2], it has become increasingly difficult for people to keep track of all the videos they watch or want to watch. Video summarization provides a solution to this problem by enabling users to get a quick and concise overview of a video, which can help them determine whether or not they want to watch the full video. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 344–359, 2023. https://doi.org/10.1007/978-3-031-36115-9_32
Boundary Refinement via Zoom-In Algorithm
345
Secondly, video summarization can improve the efficiency and effectiveness of video search engines. By creating a summary of a video, search engines can better match search queries with relevant video content, leading to a more user-friendly experience [3, 4]. Thirdly, video summarization can enhance the accessibility of video content for individuals with disabilities. For instance, individuals who are deaf or hard of hearing can use video summarization to obtain a written summary of the audio content in a video. Finally, video summarization has numerous practical applications, such as video surveillance. In this context, it can assist security personnel in quickly identifying and analyzing suspicious activities.
Fig. 1. The process of video summarization. N number of frames in the video is summarized to M number of frames where M is far smaller than N
Overall, video summarization is an important task in the field of computer vision because it can help people access, understand, and make use of video content more in a convenient and efficient way. In addition to the benefits mentioned above, video summarization has the potential to revolutionize the way we consume and interact with video content [1]. For example, video summarization can be used to create personalized video highlights or summaries based on an individual interests or preferences. It could allow people to watch only the most relevant and interesting parts of a video, rather than having to sit through the entire thing [5–7]. Video summarization can also be used to automatically create subtitles or closed captions for videos, which can make them more accessible to a wider audience. This is especially important for people who are deaf or hard of hearing, as well as for people who are learning a new language and want to improve their listening and comprehension skills. In addition, video summarization can be used to create video summaries for educational purposes. For example, teachers could use video summarization to create shorter, more focused videos that cover specific topics or concepts, which could be more engaging and effective for students. The potential applications of video summarization are vast, and as technology continues to advance,
346
A. Zarichkovyi and I. V. Stetsenko
it is likely that we will see even more innovative and creative uses for this important tool in the field of computer vision. According to a recent study of the literature [1], most video summarization methods use RNNs [8], LSTMs [9], GRUs [10], or Wavelet transformation [11] to model temporal dependencies and predict the importance of frames in a video. However, these methods have shown weaknesses related to modeling long sequences and backpropagating signals. To address these limitations, some works [12–14] have proposed using attention mechanisms to model temporal dependencies without gradient vanishing or explosion. Additionally, future works have improved these methods by modeling temporal dependencies on different levels of granularity [15–18]. Existing solutions for video summarization were developed to solve problems for specific datasets that contain short sequences. However, real-world videos, such as those found on platforms like YouTube or Facebook, tend to have an average length of tens of minutes [19], making the application of these approaches challenging. To address this issue, researchers tend to subsample long sequences with small frame rates [20] or make multiple predictions on overlapping frames [21]. However, these approaches have significant limitations, including coarse prediction of summarization boundaries, loss of context, and significant computation overhead and performance degradation on long sequences. The major research objective of this study is to develop a novel algorithm for keyshot video summarization in the context of real-world videos, specifically addressing the challenges of finding keyshots in long sequences. The proposed boundary refinement algorithm via zoom-in is the best solution, as it refines the boundaries of keyshots by zooming in on selected fragments, resulting in a more accurate and efficient summary with smaller computational overhead. State-of-the-art performance was achieved in keyshot video summarization using this approach. The proposed algorithm was tested on the TVSum [22] and SumMe [23] datasets, resulting in state-of-the-art performance of 56.2% (+0.6%) and 63.1% (+0.5%), respectively. The author hopes that this approach will enhance the effectiveness of video summarization in real-world applications, such as YouTube or Facebook, where long video sequences are common.
2 Related Work Most of the early work on video summarization uses hand-made heuristics for unsupervised training. When choosing representative frames for the video summary, they use the importance score to determine which frames are important or representative [1]. For video summarization, supervised learning techniques have recently been investigated. These methods make use of training data that consists of human-produced ground truth for clips. Since they can implicitly learn high-level semantic knowledge that is used by humans to produce summaries, these supervised learning approaches frequently outperform early work on unsupervised methods [1]. Deep learning techniques [12–18, 24–36], have recently become more popular for video summarization by modeling the variable-range temporal dependence among frames and learning how to assign priority to them based on ground truth annotations. For this, various techniques make use of Fully Convolutional Sequence Networks [18] or RNNs [24–26]. The idea behind employing them is to efficiently capture long-range
Boundary Refinement via Zoom-In Algorithm
347
dependencies between video frames, which are essential for the development of meaningful summaries. As shown in [25] variable-range dependency can be modeled using two LSTMs treating the video summarizing task as a structured prediction problem on sequential data. Other studies try to address concerns with the constrained memory of RNNs and make use of either external storage [27] or the hierarchical stacking of numerous LSTM and memory layers [28]. Some algorithms incorporate customized attention mechanisms into classic [29] or sequence-to-sequence [30] RNN-based architectures to represent the evolution of the users’ interest. Particularly, in [29] was proposed to embed an LSTMbased attention layer to model temporal dependencies and form new representations that allow producing diverse video summary. In [30] problem of video summarization viewed as a seq-to-seq learning task and was proposed to integrate an attention layer into an LSTM-based encoder-decoder network. This layer gets the encoder’s output and the previous hidden state of the decoder and computes a vector with attention values, which subsequently affects the video decoding process. Furthermore, several methods aimed to model the dependencies between frames by utilizing variations of the self-attention mechanism were found in Transformer Networks [31]. The first method [12] combines a soft self-attention mechanism with a two-layer fully connected network to regress the frame importance scores. Liu et al. [13] employ a hierarchical approach, first defining shot-level candidate key-frames and then utilizing a multi-head attention model to evaluate their importance and select the key-frames for the summary. Li et al. [32] enhance the training pipeline of the standard self-attention mechanism by incorporating a step that increases the diversity of the visual content in the summary using the computed attention values and human annotations. Ghauri et al. [33] present a variation of the architecture from [12] that incorporates additional representations of the video content, including CNN-based features from the pool5 layer of GoogleNet [34] trained on ImageNet and motion-related features from the Inflated 3D ConvNet [35] trained on Kinetics. These features are fed into self-attention mechanisms, and the outputs are combined to form a common embedding space for representing the video frames. Finally, the representation is used to learn frame importance. The PGLSUM model [36], which is closely related to existing self-attention-based approaches, captures frame dependence at different levels of granularity using global and local multihead attention mechanisms. Furthermore, unlike other methods that ignore the sequential nature of video, the PGL-SUM model [36] incorporates temporal information about the frame position, a crucial factor in estimating frame importance. Existing solutions for video summarization were designed to handle specific datasets with short sequences, making them challenging to apply to real-world videos found on platforms such as YouTube or Facebook, which tend to be tens of minutes long on average [19]. The most practical ways to address this issue are subsampling with small frame rates [20] or making multiple predictions on overlapping frames [21], but these approaches have significant limitations, including loss of context, coarse prediction of summarization boundaries, and high computation overhead. To overcome these challenges, some researchers have developed architectural level solutions for video summarization, such as the clustering-based approach presented in [37]. In this work, a graph-based hierarchical clustering method is used to compute a
348
A. Zarichkovyi and I. V. Stetsenko
video summary by selecting salient frames to represent the video content. A weight map is generated from the frame similarity graph, and clusters or connected components of the graph can easily be inferred on a minimum spanning tree of frames, with a weight map based on hierarchical observation scales computed over that tree. The closest prior work to our research is the Distilled Kernel Temporal Segmentation (D-KTS) method presented in [38]. D-KTS detects shot boundaries in state-of-the-art video solutions and exploits the temporal locality of long videos. Existing state-of-the-art solutions for video summarization [15, 36–41] typically follow the common paradigm of using two modules for processing long sequences: a preprocessing module and a model inference module. However, their preprocessing modules rely on the Kernel Temporal Segmentation [42, 43] (KTS) method. It performs an automatic Kernel-based Temporal Segmentation based on state-of-the-art video features automatically selecting the number of segments. Then, equipped with an SVM classifier for importance scoring that was trained on videos for the category at hand, we score each segment in terms of importance. Finally, the KTS approach outputs a video summary composed of the segments with the highest predicted importance scores, see Fig. 2 for more clarity. The major drawback of KTS is time-consuming under long videos (accounting for about 68% of the total processing time). The KTS computes the variances of every interval and applies the dynamic programming technique to get the shot boundaries. To consider every possible interval and obtain precise boundaries, the complexity of KTS is O(N2 ), and it becomes the bottleneck for processing long videos. To handle a 10-min video, it even needs more than 8 h while using the baseline KTS method.
Fig. 2. Overall scheme of Kernel-based Temporal Segmentation (KTS) for Video Summarization presented in [42]
So, [38] in their research, the authors presented the Distribution-based Kernel Temporal Segmentation (D-KTS) method, which leverages a new dynamic programming algorithm to improve upon the Kernel Temporal Segmentation (KTS) method used in existing state-of-the-art solutions for video summarization. Additionally, they introduced the Hash-based Adaptive Frame Selection (HAFS) to pre-process frames and construct an adaptive frame selection mechanism, improving the feature extraction performance. Specifically, they observed that the visual similarity of two intra-shot frames is higher than that of two inter-shot frames. To reduce the cost of feature extraction, they proposed for the first frame to record its hash vector through the d-hash [44]. Then, compute the
Boundary Refinement via Zoom-In Algorithm
349
hash vector for each frame and make a comparison with the record vector. If the Hamming distance is larger than our threshold, they replaced the record with the current hash vector and extract the feature vector through the neural network. Otherwise, they repeat the feature vector of the record frame without making feature extraction on the current frame. See Fig. 3 that visualize D-KTS pipeline with HASF mechanism. With D-KTS and HAFS they achieved state-of-the-art results on the TvSum and SumMe datasets when using a high frame-rate setup.
Fig. 3. Overview of Distilled Kernel Temporal Segmentation (D-KTS) pipeline with Hash-based Adaptive Frame Selection (HAFS) mechanism presented in [38]
The authors’ work is particularly noteworthy for its findings on the relationship between frame rate and F-score (see Fig. 4 and Fig. 5). Their research demonstrated that there is a positive correlation between frame rate and F-score, with down-sampling to 1 fps (as used in most related works) resulting in negative impacts on the result. Specifically, compared to no sub-sampling (30 fps), the F-score decreases by an average of 20.92% for the TVSum dataset and 20.72% for the SumMe dataset. This significant finding motivated us to research question: “Why does increasing the frame rate have such a positive impact on the F-score?” We hypothesized that the increased frame rate results in more precise identification of borders, which yields improvements compared to the heuristics used in other works [8–18, 24–27, 36, 37, 39–41]. It is worth noting that our approach differs from the D-KTS [38] method in that we do not require additional neural networks for frame pre-selection. Instead, we use a coarse-to-fine score alignment procedure that minimizes the score discrepancy between different stages. Furthermore, our algorithm recommends using the original 1 fps frame rate, which is more applicable for real-world problems where end-users have GPUs with limited memory compared to researchers’ GPUs that often have 40GB or more. As a result, our approach is simpler to train and requires fewer computational resources during both training and inference compared to DKTS.
2
5
0.8122 0.5766
0.7716 0.5348
0.74
0.7393
D…
0.5364
0.4504
0.4153
F-score
1
0.5236
0.6501
A. Zarichkovyi and I. V. Stetsenko
0.621
350
10
15
30
Sub-sampling (FPS)
1
2
5
10
15
0.5957 0.4425
0.5051 0.5874
0.6068
0.5169
0.5016
0.4626
D…
0.3581
F-score
0.4047
0.502
0.5947
Fig. 4. Correlation of F-score and FPS for D-KTS [38] approach using DR-DSN [39] and DSNet [15] backones on TVSum [23] dataset
30
Sub-sampling (FPS)
Fig. 5. Correlation of F-score and FPS for D-KTS [38] approach using DR-DSN [39] and DSNet [15] backones on SumMe [22] dataset
Boundary Refinement via Zoom-In Algorithm
351
3 Algorithm Description The proposed algorithm for video summarization is comprised of two distinct phases: key-frame pre-selection and frame border refinement. The key-frame pre-selection stage aims to identify the most relevant frames, or S’, by maximizing the likelihood of their selection, as is common in traditional methods for key-shot video summarization. Traditional methods then incorporate additional context in the form of short clips, referred to as padding P, to provide additional information about the video content. However, this padding may also include irrelevant information. The frame border refinement phase of the proposed method addresses this issue by cutting the irrelevant information from the tails of the padding to produce more accurate and shorter clips. To achieve this, the model is run on top of the selected frames S’ with padding P, but at a higher frame rate to enable the processing of more accurate borders. This process is formalized as Algorithm 1. Example is depicted in Fig. 6.
Algorithm 1. Proposed algorithm for fine-grained video summarization Inputs: keyframe selection model M, pre-defined processing sequence length T, the set of video frames , the set of corresponding labels for each frame , refinement padding length P, loss function , then pad F with T empty frames 2. From F subsample equal frequency T frames F' // Pre-selection 3. Get importance scores S' = M(F') // Border Refinement 4. From F subsample equally T frames F'' : 4.1. F'' = {} 4.2. While len(F'') < T: 4.2.1. Select frame f with largest score in S' 4.2.2. Add frame f to F'' and remove it from S' // Frame iteration order f, f – 1, f + 1, f – 2, f + 2, …, f – n, f + n 4.2.3. For frames p in range from to : 4.2.3.1. If p not in F'' then add to F'' and remove it from S' 5. Get importance scores S'' = M(F'') // Training 6.
In the algorithm, we acknowledge that the varying distribution of data between the pre-selection and border refinement stages could lead to inconsistent scores for the same frames. To address this issue, we introduce a coarse-to-fine score alignment procedure that aims to minimize the discrepancy between the scores obtained from the two stages. This, in turn, leads to a more robust network that can effectively handle both distributions. It can be done efficiently by minimizing Euclidian distance between scores from different stages.
352
A. Zarichkovyi and I. V. Stetsenko
4 Experiments and Results 4.1 Datasets To compare proposed algorithm with previous work, experiments were conducted on four datasets: TvSum [28], SumMe [29], OVP [45], and YouTube [45]. The datasets OVP and YouTube were used solely for the purpose of augmenting the training dataset. TvSum and SumMe are the only currently available datasets with suitable labeling for keyshot video summarization, although they remain relatively small for the training of deep models. A summary of the main properties of these datasets can be found in Table 1. The TvSum dataset is annotated by frame-level importance scores, while the SumMe with binary keyshot summaries. OVP and YouTube are annotated with keyframes and need to be converted to the frame-level scores and binary keyshot summaries.
Fig. 6. Processing workflow. In grey boxes not selected frames are represented, red boxes are frames selected for processing in pre-selection stage, yellow boxes are selected high confidence frames from pre-selection stage, blue boxes are frames added for border refinement stage as padding frames, in green boxes are frames selected after refinement stage and final result. Numbers inside the boxes are the corresponding model’s score to this frame after processing at each stage.
Boundary Refinement via Zoom-In Algorithm
353
Table 1. Overview of the datasets properties Dataset
# videos
Annotation
Annotation type
Video length (sec) Min
Avg
Max
TvSum [22]
50
20
Frame-level importance scores
83
235
647
SumMe [23]
25
15–18
keyshots
32
146
324
OVP [45]
50
5
keyframes
46
98
209
YouTube [45]
39
5
keyframes
9
196
572
4.2 Evaluation Protocol For fair comparison with the previous state-of-the-arts, we follow evaluation protocol from [16]. To assess the similarity between the machine and user summaries we use the harmonic mean of precision and recall expressed as the F-score in percentages: F = 2(p · r)/(p + r) · 100,
(1)
p p is precision value, r r is recall value. True and false positives and false negatives for the F-score are calculated per-frame as the overlap between the ground truth and machine summaries, as shown in Fig. 7.
Fig. 7. True positives, False positives and False negatives are calculated per-frame between the ground truth and machine binary keyshot summaries
The machine summary is limited to 15% of the original video length and then evaluated against multiple user summaries. Precisely, on the TvSum [22] benchmark, for each video, the F-score is calculated as an average between the machine summary and each of the user summaries Average F-score over videos in the dataset is then reported. On the SumMe [23] benchmark, for each video, a user summary most similar to the machine summary is selected. 4.3 Implementation Details and Setup The experiments are carried out under Python 3.9.13, PyTorch 1.13.1, CUDA 11.7, OpenCV 4.5.2. The operating system is Ubuntu 20.04. Our device has 8x Nvidia A100 80GB GPUs, 2x Intel(R) Xeon(R) Platinum 8275CL CPU @ 3.00GHz, and 1152GB memory.
354
A. Zarichkovyi and I. V. Stetsenko
In our study, we adhered to the established protocol in the field and utilized a 1 FPS framerate subsampling of videos for the preselection stage. The deep representations of frames were obtained through extraction of the output from the Video Swin Transformer [46], Swin-T specifically. The input frames were resized to a resolution of 224x224, resulting in an input token size of 16x56x56. The model was initialized with pre-trained weights for the backbone and randomly initialized weights for the head. Optimization was carried out using the AdamW [47] algorithm with an initial learning rate of 3e-4, and the model was trained for a total of 200 epochs, including a 2.5 epoch warm-up phase. The effective batch size during training was set to 64. Consistent with prior research, the gradients of the backbone were multiplied by a factor of 0.1. An increasing degree of stochastic depth rate and weight decay was employed for larger models, as recommended by [46]. The stochastic depth rate was set to 0.1, and the weight decay was set to 0.02. All experiments used α = 0.5α = 0.5 if not specified otherwise. The best-trained model was selected using a model selection criterion based on the loss curves, as outlined in [36]. If rapid changes in the loss value were not observed, the model was selected based on minimization of the loss value. All experiments were conducted using the publicly available PGL-SUM implementation available at https://github.com/e-apostolidis/PGL-SUM. 4.4 Results Table 2 presents the comparison of the state-of-the-art approaches for SumMe [23] and TvSum [22] datasets. As shown, proposed algorithm achieved state-of-the-art performance of 56.2% (+0.6%) and 63.1% (+0.5%) on the SumMe and TVSum datasets correspondingly. Table 2. Comparison different supervised summarization approaches on SumMe [23] and TvSum [22] datasets. All methods are tests with fixed 1FPS framerate. Approach
SumMe, F1-score
TvSum, F1-score
VasNet [12]
49.7
61.4
RR-STG [14]
53.4
63.0
DSNet [15]
50.2
62.1
PGL-SUM [36]
55.6
62.7
D-KTS w/ DSNet backbone [38]
50.2
62.3
Ours
56.2
63.2
4.5 Ablation Study In the ablation study section, our objective is twofold. Firstly, we aim to evaluate the impact of each modification introduced in our study on the final performance. Secondly,
Boundary Refinement via Zoom-In Algorithm
355
we aim to compare the performance of our algorithms with that of the D-KTS [38] approach at higher framerates. This comparative analysis will provide insights into the efficacy of our approach and its potential to outperform the state-of-the-art technique. Additionally, by isolating the contribution of each change, we can better understand the relative importance of each modification and its impact on the overall performance of the systems. 4.6 Study on Importance of Introduced Components Within this study, we proposed a novel algorithm that incorporates three major modifications to the PGL-SUM [36] framework. First, we replaced the backbone with the Video Swin Transformer, a more effective feature extractor for video summarization tasks. Second, we introduced boundary refinement to improve the accuracy of the extracted keyframes. Third, we implemented a coarse-to-fine score alignment procedure to better align the score predictions across different stages of the algorithm. The effectiveness of each modification is evaluated and reported in Table 3. Our results indicate that the introduction of the Video Swin Transformer backbone leads to approximately 50% of the overall improvement in the performance, while the remaining 50% can be attributed to the proposed boundary refinement and coarse-to-fine score alignment techniques. Table 3. Impact of different component of proposed algorithm Method
SumMe, F1-score
TvSum, F1-score
PGL-SUM [36] (baseline)
55.6
62.7
+ Video Swin Transformer as backbone
55.9 (+0.3)
62.9 (+0.2)
+ Boundary refinement
56.1 (+0.2)
63.1 (+0.2)
+ Coarse-to-fine alignment
56.2 (+0.1)
63.2 (+0.1)
4.7 Comparison with D-KTS at Higher Framerates In order to compare our algorithm with the D-KTS approach [38], we increased the frame rate for pre-selection phase of proposed algorithm and linearly increased the temporal shape of the input dimension for the Video Swin Transformer. As shown in Fig. 8 and Fig. 9, our proposed algorithm outperformed the D-KTS approach at lower frame rates. However, this difference became negligible at higher frame rates of 15FPS and 30FPS, resulting in a difference of only + 0.1% and 0.yyy% on the SumMe and TVSum datasets, respectively. We attribute this to the fact that at higher frame rates, both approaches have access to more frames, leading to a rapid decrease in the difference in boundary detection accuracy. However, as we stated earlier, the primary goal of this study is to provide a practical solution for handling long video sequences without adding complexity or overhead during inference. In our opinion, the proposed algorithm performs better in this regard than D-KTS [38]. It can be used at lower frame rates and can be deployed on smaller GPU accelerators with 12GB of VRAM for processing sequences that are tens
356
A. Zarichkovyi and I. V. Stetsenko
0.740 0.770
0.772 0.773
10
15
0.812 0.812
0.739 0.747
0.650 0.664
5
D-…
F-score
0.621 0.632
of minutes long, which is common on platforms such as YouTube or Facebook [19]. Therefore, we believe that our algorithm is a more practical and feasible approach for real-world applications.
1
2
30
Sub-sampling (FPS)
0.595 0.637
0.607 0.636
0.587 0.638
0.596 0.638
10
15
30
0.463
0.602
5
D-…
F-score
0.502 0.562
Fig. 8. Comparison of proposed algorithm with D-KTS [38] at higher framerates on TVSum dataset [22]
1
2
Sub-sampling (FPS)
Fig. 9. Comparison of proposed algorithm with D-KTS [38] at higher framerates on SumMe dataset [23]
Boundary Refinement via Zoom-In Algorithm
357
5 Conclusions In this study, a novel algorithm is proposed for video summarization that improves upon the state-of-the-art PGL-SUM approach. The algorithm introduces a boundary refinement stage that further reduces the number of non-relevant frames identified by the pre-selection stage, through the use of coarse prediction. Also coarse-to-fine score alignment technique is implemented to ensure consistency between the scores obtained from each stage. Compared to D-KTS, closes competitor, the proposed algorithm is more practical for real world problems as it achieves superior performance at lower frame rates, making it applicable with smaller GPU accelerators with only 12GB of VRAM. Our approach incorporates a Video Swin Transformer backbone and proposed algorithm into PGL-SUM pipeline. The significant improvements on the SumMe and TVSum datasets, with a 56.2% (+0.6%) and 63.1% (+0.5%) improvement, respectively, are shown. These results demonstrate the effectiveness of our algorithm in addressing the challenges of video summarization for long sequences, making it a valuable tool for practical applications in the field.
References 1. Evlampios, A., Eleni, A., Alexandros, I.M., Vasileios, M., Ioannis, P.: Video summarization using deep neural networks. CoRR, vol. abs/2101.06072, http://arxiv.org/abs/2101.06072 (2021) 2. Fu, H., et al.: Self-attention binary neural tree for video summarization. Pattern Recognit. Lett. (2021) 3. Sahu, A., et al.: First person video summarization using different graph representations. Pattern Recognit. Lett. (2021) 4. Fei, M., et al.: Memorable and rich video summarization. J. Vis. Commun. Image Represent (2017) 5. Nagendraswamy, H.S., Chethana kumara, B.M., Guru, D.S., Naresh, Y.G.: Symbolic Representation of Sign Language at Sentence Level. IJIGSP 7(9), 49–60 (2015). https://doi.org/10. 5815/ijigsp.2015.09.07 6. Shivanand, S., Gornale, A.K.B., Yannawar, P.L.: Analysis and detection of content based video retrieval. Int. J. Image Graph. Sig. Proc. (IJIGSP) 11(3), 43–57 (2019). https://doi.org/ 10.5815/ijigsp.2019.03.06 7. Rahman, A., Hasan, S., Rafizul Haque, S.M.: Creation of video summary with the extracted salient frames using color moment, color histogram and speeded up robust features. Int. J. Info. Technol. Comp. Sci. (IJITCS) 10(7), 22–30 (2018). https://doi.org/10.5815/ijitcs.2018. 07.03 8. Fu, T.-J., et al.: Attentive and adversarial learning for video summarization. In: IEEE Winter Conf. on Applic. of Comp. Vision HI, USA: IEEE, pp. 1579–1587 (2019) 9. Hochreiter, S., et al.: Long short-term memory. Neural Comput. 9(8), 1735–1780 (1997) 10. Cho, K., et al.: Learning Phrase Representations using RNN Encoder–Decoder for Statistical Machine Translation. In: 2014 Conf. on Empirical Methods in Natural Language Processing. Doha, Qatar: ACL, pp. 1724–1734 (2014) 11. Kavitha, J., Arockia Jansi Rani, P.: Design of a video summarization scheme in the wavelet domain using statistical feature extraction. IJIGSP 7(4), 60–67 (2015). https://doi.org/10. 5815/ijigsp.2015.04.07
358
A. Zarichkovyi and I. V. Stetsenko
12. Fajtl, J., et al.: Summarizing Videos with Attention. In: Asian Conf. on Comp. Vision 2018 Workshops, pp. 39–54. Springer Int. Publishing, Cham (2018) 13. Liu, Y.-T., et al.: Learning hierarchical self-attention for video summarization. In: 2019 IEEE Int. Conf. on Image Processing. IEEE, pp. 3377–3381 (2019) 14. Zhu, W., Han, Y., Lu, J., Zhou, J.: Relational reasoning over spatial-temporal graphs for video summarization. IEEE Trans. Image Process. 31, 3017–3031 (2022). https://doi.org/10.1109/ TIP.2022.3163855 15. Zhu, W., Lu, J., Li, J., Zhou, J.: DSNet: a flexible detect-to-summarize network for video summarization. IEEE Transactions on Image Processing 30, 948–962 (2021). https://doi.org/ 10.1109/TIP.2020.3039886 (2020) 16. Zhu, W., Lu, J., Han, Y., Zhou, J.: Learning multiscale hierarchical attention for video summarization. Pattern Recognition 122, 108312 (2022). ISSN 0031-3203, https://doi.org/10.1016/ j.patcog.2021.108312 (2021) 17. Zhu, W., Han, Y., Lu, J., Zhou, J.: Relational reasoning over spatial-temporal graphs for video summarization. IEEE Transactions on Image Processing 31, 3017–3031 (2022). https://doi. org/10.1109/TIP.2022.3163855 18. Rochan, M., et al.: Video summarization using fully convolutional sequence networks. In: Europ. Conf. on Comp. Vision 2018, pp. 358–374. Springer Int. Publishing, Cham (2018) 19. Ceci. L.: Statista. Average YouTube video length as of December 2018 (Aug 2021). www.sta tista.com/statistics/1026923/youtube-video-category-average-length 20. Sah, S., Kulhare, S., Gray, A., Venugopalan, S., Prud’Hommeaux, E., Ptucha, R.: Semantic text summarization of long videos. In: 2017 IEEE Winter Conference on Applications of Computer Vision (WACV), pp. 989–997. Santa Rosa, CA, USA (2017). https://doi.org/10. 1109/WACV.2017.115 21. Wenyi, L., Wenjing, L.: Sliding Window Recurrent Network for Efficient Video SuperResolution. CoRR, vol. abs/2208.11608, http://arxiv.org/abs/2208.11608 (2022) 22. Song, Y., Vallmitjana, J., Stent, A., Jaimes, A.: Tvsum: summarizing web videos using titles. In: Proceedings of the IEEE CVPR, pp. 5179–5187 (2015) 23. Gygli, M., Grabner, H., Riemenschneider, H., Van Gool, L.: Creating Summaries from User Videos. In: Fleet, D., Pajdla, T., Schiele, B., Tuytelaars, T. (eds.) ECCV 2014. LNCS, vol. 8695, pp. 505–520. Springer, Cham (2014). https://doi.org/10.1007/978-3-319-10584-0_33 24. Zhao, B., et al.: Hierarchical recurrent neural network for video summarization. In: 25th ACM Int. Conf. on Multimedia NY, pp. 863–871. ACM, USA (2017) 25. Zhang, K., et al.: Video summarization with long short-term memory. In: Europ. Conf. on Comp. Vision 2016, pp. 766–782. Springer Int. Publishing, Cham (2016) 26. Zhao, B., et al.: HSA-RNN: Hierarchical Structure-Adaptive RNN for Video Summarization. In: 2018 IEEE Conf. on Comp. Vision and Pattern Rec., pp. 7405–7414 (2018) 27. Feng, L., et al.: Extractive video summarizer with memory augmented neural networks. In: 26th ACM Int. Conf. on Multimedia, pp. 976–983. ACM, NY, USA (2018) 28. Wang, J., et al.: Stacked memory network for video summarization. In: 27th ACM Int. Conf. on Multimedia, pp. 836–844. ACM, NY, USA (2019) 29. Lebron Casas, L., et al.: Video Summarization with LSTM and Deep Attention Models. In: 25th Int. Conf. on Multimedia Modeling, pp. 67–79. Springer Int. Publishing, Cham (2019) 30. Ji, Z., et al.: Deep attentive and semantic preserving video summarization. Neurocomputing 405, 200–207 (2020) 31. Vaswani, A., et al.: Attention is all you need. In: 31st Int. Conf. on Neural Information Processing Systems, pp. 6000–6010. Red Hook, NY, USA: Curran Associates Inc. (2017) 32. Li, P., et al.: Exploring global diverse attention via pairwise temporal relation for video summarization. Pattern Recognition 111(107677) (2021) 33. Ghauri, J., et al.: Supervised Video Summarization Via Multiple Feature Sets with Parallel Attention. In: 2021 IEEE Int. Conf. on Multimedia and Expo, pp. 1–6. IEEE, CA, USA (2021)
Boundary Refinement via Zoom-In Algorithm
359
34. Szegedy, C., et al.: Going deeper with convolutions. In: 2015 IEEE Conf. on Comp. Vision and Pattern Rec., pp. 1–9 35. Carreira, J., et al.: Quo vadis, action recognition? a new model and the kinetics dataset. In: 2017 IEEE Conf. on Comp. Vision and Pattern Rec., pp. 4724–4733 36. Apostolidis, E., Balaouras, G., Mezaris, V., Patras, I.: Combining global and local attention with positional encoding for video summarization. In: 2021 IEEE International Symposium on Multimedia (ISM), pp. 226-234. Naple, Italy (2021). https://doi.org/10.1109/ISM52913. 2021.00045 37. Belo, L., Caetano, C., Patrocínio Jr, Z., Guimarães, S.: Summarizing video sequence using a graph-based hierarchical approach. Neurocomputing 173, 1001-1016 (2016). https://doi.org/ 10.1016/j.neucom.2015.08.057 38. Ke, X., Chang, B., Wu, H., Xu, F., Zhong, S.: Towards practical and efficient long video summary. In: ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 1770–1774. Singapore, Singapore (2022). https://doi.org/ 10.1109/ICASSP43922.2022.9746911 39. Zhou, K., Qiao, Y., Xiang, T.: Deep reinforcement learning for unsupervised video summarization with diversity-representativeness reward. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 32 (2018) 40. Zhang, Y., Kampffmeyer, M., Zhao, X., Tan, M.: Dtr-gan: Dilated temporal relational adversarial network for video summarization. In: Proceedings of the ACM Turing Celebration Conference-China, pp. 1–6 (2019) 41. Ji, Z., Zhao, Y., Pang, Y., Li, X., Han, J.: Deep attentive video summarization with distribution consistency learning. IEEE transactions on neural networks and learning systems 32(4), 1765– 1775 (2020) 42. Potapov, D., Douze, M., Harchaoui, Z., Schmid, C.: Category-Specific Video Summarization. In: Fleet, D., Pajdla, T., Schiele, B., Tuytelaars, T. (eds.) ECCV 2014. LNCS, vol. 8694, pp. 540–555. Springer, Cham (2014). https://doi.org/10.1007/978-3-319-10599-4_35 43. Lei, Z., Sun, K., Zhang, Q., Qiu, G.: User video summarization based on joint visual and semantic affinity graph. In: Proceedings of the 2016 ACM Workshop on Vision and Language Integration Meets Multimedia Fusion, pp. 45–52. New York, NY, USA (2016). Association for Computing Machinery 44. Hacker Factor: The hacker factor blog (January 2013). http://www.hackerfactor.com/blog/?/ archives/529-Kind-of-Like-That.html 45. De Avila, S.E.F., Lopes, A.P.B., da Luz Jr, A., de Albuquerque Ara´ujo, A.: Vsumm: a mechanism designed to produce static video summaries and a novel evaluation method. Pattern Recognition Letters 32(1), 56–68 (2011) 46. Liu, Z., Ning, J., Cao, Y., Wei, Y., et al.: Video Swin Transformer. CoRR, vol. abs/2106.13230, http://arxiv.org/abs/2106.13230 (2021) 47. Kingma, D.P., Ba, J.: Adam: A method for stochastic optimization. CoRR, vol. abs/1412.6980, http://arxiv.org/abs/1412.6980 (2014)
Solving Blockchain Scalability Problem Using ZK-SNARK Kateryna Kuznetsova1,2(B) , Anton Yezhov1 , Oleksandr Kuznetsov1,2,3 , and Andrii Tikhonov1 1 Zpoken, OU, Harju Maakond, Kesklinna Linnaosa, Sakala tn 7-2, 10141 Tallinn, Estonia
[email protected]
2 Department of Information and Communication Systems Security, School of Computer
Sciences, V.N., Karazin Kharkiv National University, 4 Svobody Sq., Kharkiv 61022, Ukraine 3 Department of Political Sciences, Communication and International Relations, University of
Macerata, Via Crescimbeni, 30/32, 62100 Macerata, Italy
Abstract. Modern users demands for digital information processing and storage systems, which provide reliable, safe and consistent reproduction of recorded facts and issues. We are well acquainted with distributed ledger technology (DLT) that provides the information storage on independent and unaccountable nodes which interact in complete distrust with each other. It yet provides reliable, secure and historically consistent storage. Due to the increase of blockchain network users it is necessary to provide a resource availability service for billions of users in different parts of the world. Moreover, everyone who interacts with the blockchain network in the mode of complete distrust forces to repeatedly check long chain of blocks containing cryptographic integrity and involvement marks of the party that generates them. Thus, the main recent problem of DLTs is scalability. In this paper, we present the main idea of our solution using new cryptographic technology, known as zero-knowledge proof system. We substantiate the general scheme of fast, reliable and secure DLT verification systems and provide time and measurement result of the classic DLT verification and the new one to demonstrate the perspective of the chosen direction. Keywords: Blockchain · Scalability Problem · Distributed Ledger · ZK-SNARK Technology
1 Introduction Blockchain is a distributed decentralized data storage, implemented by linking information into a continuous chain of blocks [1–3]. Each next block is formed using a cryptographic connection using hash function as it is shown in the figure below [4]. In blockchain, there is the well-known trilemma, which formulates three main requirements: security, decentralization and scalability [5]. The security property avoids information theft and misused. It also provides the right to privacy and anonymity. Security is the basic requirement, which attracts investors or help to win approval of partners. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 360–371, 2023. https://doi.org/10.1007/978-3-031-36115-9_33
Solving Blockchain Scalability Problem Using ZK-SNARK
361
Fig. 1. Chain of linked blocks built using hashing
In addition to security, we require decentralized system, i.e. which avoids control by a single decision-making center, server, node or subject. This is a problem of reliability, because a single point of failure always attracts swindlers and attackers. Decentralization also protects from censorship. It is a manifestation of freedom and independence and democracy of decisions. Scalability is the property of a system to continuously evolve as the amount of work increases. For example, if the number of users in a system increases, then their service time does not increase significantly. This requirement is a difficult to implement because it is not always possible to provide extensive scalability (only by increasing the number of computers). During the development every large DLT project faces the limitations in the number of users, resources and time that can be processed by a system [6]. Everyone needs to check long chain of cryptographic primitives, and the longer the project exists, the more entities are involved in it, the more difficult and costly these verifications are. The scheme shown in the Fig. 1 solves the problems of security and decentralization. Complex cryptographic transformations are used to link blocks together and confirm their authenticity. All blocks are accepted by the consensus of the participants, that is there is no main node (single decision-making center) in the system. But there is a lot of work to validate the records in such DLT. The mostly used method for verifying the correctness of blocks in a blockchain is native registry check, which consists in recalculation of a chain of linked hashes and all signatures of all blocks. This method requires a great amount of time. For instance, if we take a millionth block, it is necessary to calculate a million hashes and a million signatures. It is really computationally hard. In addition, native check includes viewing all records, which violates the privacy. Modern blockchain projects solve this problem in their own way [7–10]. Some of them try new technologies to create a network like directed acyclic graph (DAG) [11–14] or residual number system (RNS) [15]. The NEAR project, for instance, partially solves the scalability problem by segmenting (sharding) the ledger [16, 17]. However, this does
362
K. Kuznetsova et al.
not solve the problem globally. If billions of users start checking their blocks at the same time, the network may fail. Therefore, the scalability problem is the main deterrent in the mass introduction of blockchain networks in all spheres of human activity. The solution of this problem today is the primary task of modern informatics and applied crypto-engineering. The purpose of our research is to study the ZK-SNARKS technology and apply it to solve the scalability problem in modern DLT systems. We approve, that this technology can be used to verify the authenticity of a chain of blocks, as well as speed up the process of verifying the integrity of the blockchain.
2 ZK-SNARK Technology to Solve the Scalability Problem ZK-SNARK is an acronym that stands for «Zero-Knowledge Succinct Non-Interactive Argument of Knowledge» [18]. ZK-SNARK is a cryptographic proof that allows one party to prove it possesses certain information without revealing that information. This technology was originally designed to provide privacy [19]. However, such properties as the short proofs and the logarithmic dependence of their verification, instead of native check, which provides the linear one, make it possible to use ZK-SNARK to solve the global problem of scalability of blockchain systems. Suppose we are dealing with some computational algorithm that requires t steps. For example, this is an algorithm for generating a chain of blocks (or their native verification) in modern DLT. Asymptotic estimates show that ZK-SNARK technology allows generating a proof approximately in O(t·log(t)) steps and verifying them in O(log2 (t)) steps. Thus, ZK-SNARK divides the all calculations into two steps: • A complex process of generating proof of computational integrity; • A fast verification of computational integrity by everyone using cheap and lowresource computers. Computational Integrity (CI) guarantees (with some, very small error) the correctness of the performed calculations. For example, there is a proof that the block in the DLT does indeed contain all cryptographic integrity and involvement marks, and they are formed according to the specified protocol. Consider an example of binary logarithm, where t = 220 = 106 . We need to perform 20·106 steps to make a proof and approximately 202 ≈ 400 steps to check it. The verification is significantly faster than the native registry check, which requires 220 ≈ 1 million steps. So this solves the scalability problem. On Fig. 2 we present an asymptotic estimate for native check, proof generation and verification due to the above reasoning. Native check requires constant recalculation of all blocks. This is computationally intensive task for long chain of blocks. Generating proofs takes time. It is even slightly longer than the native check. The good point is that you can generate proofs once, using some powerful server with a multi-core architecture. Verification significantly reduces the required processing time. It is achieved due to the logarithmic dependency. In this case for all users, even with low-powered gadgets, computations will be available.
Solving Blockchain Scalability Problem Using ZK-SNARK
363
As a result, you have a service that stores proofs for each block and allows billions of users to verify their blocks at the same time very quickly.
Fig. 2. Asymptotic estimates of computational complexity of native verification (dark graph), proof generation (sky-blue graph) and proof verification (green graph) [20]
3 Our Solution We implement a recursive proof of the computational integrity of a chain of linked hashes with digital signature verification. This protects against possible data changes and the imposition of a false chain of related hashes. Initial data. For a test case, consider a simplified version, where each block contains three fields (Fig. 3): • Unique block number: Noncei ; • Hash of the previous block: hi-1 ; • Digital signature: EDS i . To simplify, we accept Noncei = i. The result of the n-th hashing is: hn = H (hn−1 ||n) = H (H (hn−2 ||n − 1)||n) = . . . = = H (H (H (. . . H (H (0)||1) . . . ||)n − 2)||n − 1)||n). Additionally, each hash hi is encrypted with a secret key sk, i.e. we form a signature EDS i = EDS(hi , sk). The public key pk is used to verify the signature, i.e. we decrypt EDS i and check for equality hi = D(EDS i , pk). We use only the signature verification algorithm hi = D(EDS i , pk) to generate proofs and CI verification. While implementing a proving scheme, our team decided to divide the research into two directions: aggregation and recursion.
364
K. Kuznetsova et al.
Fig. 3. A simplified version of the linked list with the formation of EDS
3.1 Aggregation A chain of proofs is created by aggregating the proof of the previous block with the current one. As a result, for each specific block, we can check the proof that the block hash and signatures were calculated correctly and the proof chain for previous blocks was verified correctly. This method requires digital signatures to confirm computational integrity of the block, otherwise a possible attacker can simply substitute some blocks. To implement a recursive proof of CI of a chain, it is necessary to consistently implement the following tasks: 1. Implement a hash chain h0 = H(0), hi = H(hi-1 ||i), i = 1,…n; 2. For each hash h0 ,… hn : a. Create a circuit CH i (xH i , wH i ) of the hash algorithm H, where the public input xH i = hi is the result of hashing, the witness wH i is the hash preimage: wH 0 = 0, wH i = hi-1 ||i, i = 1,…n; b. Form public settings (SH pi , SH vi ) = S(CH i (xH i , wH i )), where SH pi are public prover settings, SH vi are public verifier settings; c. Form a proof CI for hashing π H i = P(SH pi , xH i , wH i ); d. Implement verification algorithm V(SH vi , xH i , π H i ) takes values {0, 1} (accept or reject); e. Proof verification, i.e. to make sure that V(SH vi , xH i , π H i ) = accept. 3. For each signature EDS i = E(hi , sk), i = 0,…n: a. Create a circuit CDi (xDi , wDi ) of proof verification hi = D(EDS i , pk), where the public input xDi = hi is the result of hashing, the witness wDi = (EDS i , pk) are the signature and the public key; b. Form public settings (SDpi , SDvi ) = S(CDi (xDi , wDi )), where SDpi are public prover settings, SDvi are public verifier settings; c. Form a proof CI for signature verification π Di = P(SDpi , xDi , wDi ); d. Implement proof verification algorithm V(SDvi , xDi , π Di ) takes values {0, 1} (accept or reject); e. Proof verification, i.e. to make sure that V(SDvi , xDi , π Di ) = accept. 4. For every triple of proofs i-1 = P(S Pi-1 , X i-1 , W i-1 ), π H i = P(SH pi , xH i , wH i ) and π Di = P(SDpi , xDi , wDi ), i = 1,…n: a. Create a circuit C i (X i , W i ) verification algorithm V, where:
Solving Blockchain Scalability Problem Using ZK-SNARK
365
• X i = (V (S Vi-1 , X i-1 , i-1 )), V (SH vi , xH i , πH i ), V (SDvi , xDi , πDi ), for all i = 1,…n. • W 1 = (π0 , h0 , π1 , h1 , EDS 1 , pk), W 1 = ( i-1 , X i-1 , πi , hi , EDS i , pk), for all i = 2,…n. b. Form public settings (S Pi , S Vi ) = S(C(X i , W i )), where S Pi are public prover settings, S Vi are public verifier settings; c. Form a proof of CI i = P(S Pi , X i , W i ); d. Implement proof verification algorithm V(S Vi , X i , i ) takes values {0, 1} (accept or reject); e. Proof verification, i.e. to make sure that V(S Vi , X i , i ) = accept. Thus, each proof i = P(S Pi , X i , W i ), i = 1,…n is the aggregation of three other proofs: 1 Proof CI of previous chain of linked hashes i-1 = P(S Pi-1 , X i-1 , W i-1 ); 2 Proof CI of current hash π H i = P(SH pi , xH i , wH i ); 3 Proof CI of current signature verification π Di = P(SDpi , xDi , wDi ). Condition fulfillment V(S Vi , X i , i ) = accept for all i = 1,…n means that the proof verification i-1 = P(S Pi-1 , X i-1 , W i-1 ), π H i = P(SH pi , xH i , wH i ) and π Di = P(SDpi , xDi , wDi ) were calculated correctly. If V(S Vi-1 , X i-1 , i-1 ) = accept, V(SH vi , xH i , π H i ) = accept and V(SDvi , xDi , π Di ) = accept, it means that: 1. There is a proof of CI of previous chain, i.e. the verification V(S Vi-2 , X i-2 , i-2 ) = accept is computed correctly; 2. There is a proof of CI of current hash, i.e. the value hi = H(hi-1 ||i) is computed correctly; 3. There is a proof of CI of current signature, i.e. verification hi = D(EDS i , pk) is computed correctly. The scheme of forming a chain of recursive proofs of computational integrity with verification of the correctness of electronic digital signatures is shown in Fig. 4. 3.2 Recursion This time the next proof in a chain is created based on the proof of the previous block. We generate a single proof for hash and signatures of the current block, not aggregating them, just making an enlarged circuit. A proof of the next block is formed by verifying the previous block (i.e., by verifying the correctness of hash and signatures of the previous block) and calculating a proof for the current block. This scheme (Fig. 5) allows checking the previous hash so that we solve the block substitution problem. The scheme does not require a digital signature to provide the computational integrity of a chain. To implement a recursive proof of CI of a chain, it is necessary to consistently implement the following tasks: 1. Implement a block chain based on linked hashes h0 = H(0), hi = H(hi-1 ||i), i = 1,…n;
366
K. Kuznetsova et al.
Fig. 4. Scheme of forming a chain of proofs of CI with digital signature verification (first variant)
2. For each block create a circuit C i (X i , W i ) of the hash algorithm H, signature verification and the verification algorithm V for the previous block π i-1 = P(S pi-1 , x i-1 , wi-1 ). X i = (V(S vi-1 , x i-1 , π i-1 ), x i ): V(S vi-1 , x i-1 , π i-1 ) is the result of verification (accept or reject) of the previous proof π i-1 , x i = hi is the result of hashing. W i = (π i-1 , hi-1 , wH i , wDi ): π i-1 , hi-1 are the input data for verification of proof and wi are the input data (witness, hash-preimage) to calculate hashes: w0 = 0, wi = (hi-1 ||i), i = 1,…n, the witness wDi = (EDS i , pk) are the signature and the public key; 3. Form public settings (S pi , S vi ) = S(CH i (xH i , wH i ), CDi (xDi , wDi )), where S pi are public prover settings, S vi are public verifier settings; 4. Implement proof verification algorithm V(S vi , x i , π i ) takes values {0, 1} (accept or reject); 5. Proof verification, i.e. to make sure that V(S vi , x i , π i ) = accept.
Solving Blockchain Scalability Problem Using ZK-SNARK
367
Fig. 5. Scheme of forming a chain of proofs of CI with digital signature verification (second variant)
Condition fulfillment V(S Vi , X i , i ) = accept for all i = 1,…n means that there is a proof of CI of previous chain, i.e. the verification V(S vi-1 , x i-1 , π i-1 ) = accept is computed correctly, CI of current hash, i.e. the value hi = H(hi-1 ||i) is computed correctly and CI of current signature, i.e. verification hi = D(EDS i , pk) is computed correctly.
4 Time and Measurement Results All computations were made on 2,40GHz Intel(R) Xeon(R) CPU E5–2680 v4. 4.1 Aggregation Method
Table 1. Time and measurement results for a chain of proofs for hashes №
Hash
Time to build a circuit, s
Time to make a proof, s
Proof size, bytes
Verification, s
0
5FECEB66FFC86F38D952786C6D696C79 C2DBC239DD4E91B46729D73A27FB57E9
1.111688
0.9969
132640
0.0224
1
F3C1D27925BF4262AE19BFCBCCE1B7612 4F54A4DBF62DC4B09E6B353F34D277D
2.2134259
1.6811
150268
0.0282
2
9C808663A3D1A47F3FCFE26BA11AFD 90D0F00A41984F8058DE3CE8552C6BEF77
2.2696147
1.7133
150268
0.1521
3
B3CC82D30374FC4FA584032CF35E86 A54E70E08A9E3E9982BFE06A6022A4937A
2.3223338
1.7160
150268
0.1571
4
4D23AF03289C3EB09E4C888999102EE53 E6A49D49FD06326A09ED924D22D34A9
2.8574667
2.6339
150268
0.0621
368
K. Kuznetsova et al. Table 2. Time and measurement results for a chain of proofs for signatures
№
Time to build a circuit, s
Time to make a proof, s
Proof size, bytes
Verification, s
0
63.335045
75.2214
208376
0.0170
1
58.583714
61.1627
208376
0.0173
2
64.45872
57.5522
208376
0.0225
71.951866
72.8148
208376
0.0182
59.8327
208376
0.0168
3 4
102.80682
Table 3. Time and measurement results for a chain of aggregated proofs for signatures & hashes №
Time to build a circuit, s
Time to make a proof, s
Proof size, bytes
Verification, s
0
2.6056433
1.4931
146348
0.0122
1
2.6340964
1.7561
146348
0.0120
2
2.5258436
1.6699
146348
0.0171
3
3.8990123
2.5616
146348
0.0178
4
2.6605875
1.7653
146348
0.0138
4.2 Recursion Method 4.3 Discussion The obtained results fully confirm the asymptotic estimates presented in Sect. 2. Thus, the time for creating a proof increases with the increase in the amount of data. We Table 4. Time and measurement results for a chain of recursive proofs №
Hash
0
81DDC8D248B2DCCDD3FDD5E8 4F0CAD62B08F2D10B57F9A831C13451E5C5C80A5
1
AC8BE15C3CC494661BC32FF34 57E273A98896338195A812702108574D9930EAD
2
Time to build a circuit, s
Verification, s
Proof size, bytes
1.4125
0.0213
149084
72.49396
74.2432
0.0457
211432
3730496E35571584A2839C24348 1A461AA7E304203666CB6A358247CB2818914
53.94758
46.4669
0.0337
211432
3
C9C25FCF5183B139A462ECD0 6919572A88059355D3271F5CFDBBCD46F32C6723
50.751015
48.2629
0.0339
211432
4
81DDC8D248B2DCCDD3FDD5E84 F0CAD62B08F2D10B57F9A831C13451E5C5C80A5
52.462822
44.4447
0.0379
211432
1.119945
Time to prove, s
Solving Blockchain Scalability Problem Using ZK-SNARK
369
see from the aggregation example that creating a proof for a digital signature is much more difficult than for a hash. We also note that we use the SHA256 algorithm for the cryptographic connection of blocks, i.e. hashing, and EDDSA25519 curve and the SHA512 hashing algorithm for signature. The verification time is very low, which is explained by the logarithmic dependence. There is practically no difference between verifying a proof for a signature and a hash. For aggregation we first implement proofs for hashes (Table 1) and signatures (Table 2) for each block and aggregate them with the proof from the previous block (Table 3). As you can see the proof size for signature is a little bigger than for hash, but we use aggregation, which compress the final proof. The final recursive proof is much bigger that in the case of aggregation. This is explained by the fact that in the aggregation scheme we create a separate circuit for each entity: hash, signature and aggregation. And the recursion scheme requires only one circuit for all these three entities. Creating such an extended scheme takes more memory and, as a result, the final proof is also large. As you can see from the tables for both implementations verification is really fast. Proof generation takes time, but its speed depends on the number of processor cores, i.e. the task is well parallelized. It is also worth considering that computations were made on the system, which also performs other processes. Because of this there are some measurement errors.
5 Conclusion On the graph below there are the results of computation times for native verification (or recalculation of all hashes), verification of proofs generated for each block and recursive proof. We see that even the proof for a separate block significantly reduces the cost of verification. Recursion combines all these proofs into one. Verification is very fast, which, in fact, solves the main problem of mass introduction of distributed systems, which is the scalability. Thus, zero-knowledge proofs are a great solution to solve the problems of scalability and confidentiality. This is especially important in large distributed computing projects. We can introduce a proof system and replace native block verification with it. This significantly speeds up verification and facilitates the work of a system. In this article, we have shown how ZK-SNARK technology can be used to prove the integrity of the blockchain. As an example, we took a simplified version of the block, that is, without taking into account transactions and other auxiliary data structures necessary for the safe and reliable operation of the blockchain system. Our task was only to demonstrate the prospect of this direction. In the future, this technology can be applied to a specific blockchain ledger, using real network data as blocks. Moreover, the obtained results can be also useful in various computer science applications, including the development and implementation of computationally efficient systems for information protection, reliability and security (Fig. 6).
370
K. Kuznetsova et al.
Fig. 6. Computational complexity of native verification, proof verification, recursive proof verification Acknowledgment. This project is supported by Zpoken, OU, Harju maakond, Tallinn, Kesklinna linnaosa, Sakala tn 7–2, 10141, Estonia.
References 1. Nakamoto, S.: Bitcoin: A Peer-to-Peer Electronic Cash System 2. Lee, S.-W., Singh, I., Mohammadian, M. (eds.): Blockchain Technology for IoT Applications. BT, Springer, Singapore (2021). https://doi.org/10.1007/978-981-33-4122-7 3. Xu, K., Zhu, J., Song, X., Lu, Z. (eds.): CBCC 2020. CCIS, vol. 1305. Springer, Singapore (2021). https://doi.org/10.1007/978-981-33-6478-3 4. Distributed Ledger Technology: beyond block chain. UK: Government Office for Science (2016) 5. Buterin, V.: Why sharding is great: demystifying the technical properties. Vitalik Buterin’s website 7 (Apr 2021) 6. Chowdhury, M.J.M., et al.: A comparative analysis of distributed ledger technology platforms. IEEE Access 7, 167 930–167 943 (2019) 7. Krishnan, S., Balas, V.E., Golden, J., Robinson, Y.H., Balaji, S., Kumar, R. (eds.) Handbook of Research on Blockchain Technology. Academic Press (2020) 8. Oliynykov, R., Kuznetsov, O., Lemeshko, O., Radivilova, T. (eds.): Information Security Technologies in the Decentralized Distributed Networks. LNDECT, vol. 115. Springer, Cham (2022). https://doi.org/10.1007/978-3-030-95161-0 9. Shapoval, O., Kuznetsov, A., Poluyanenko, N., Yakovenko, V., Prokopovych-Tkachenko, D., Kavun, S.: The decentralized voting model using the hyperledger platform paper. In: 2019 International Conference on Information and Telecommunication Technologies and Radio
Solving Blockchain Scalability Problem Using ZK-SNARK
10.
11.
12. 13.
14.
15. 16. 17. 18. 19. 20.
371
Electronics (UkrMiCo), pp. 1–5 (2019). https://doi.org/10.1109/UkrMiCo47782.2019.916 5368 Poluyanenko, N., Kuznetsov, A., Lisickiy, K., Datsenko, S., Nakisko, O., Rudenko, S.: The problem of double costs in blockchain systems. In: Hu, Z., Petoukhov, S., Dychka, I., He, M. (eds.) Advances in Computer Science for Engineering and Education III, pp. 640–652. Springer International Publishing, Cham (2021). https://doi.org/10.1007/978-3-030-555061_57 Ferraro, P., King, C., Shorten, R.: On the stability of unverified transactions in a DAG-based distributed ledger. IEEE Trans. Autom. Control. 65, 3772–3783 (2020). https://doi.org/10. 1109/TAC.2019.2950873 Devarajan, A., Karabulut, E.: Directed Acyclic Graph Based Blockchain Systems (2020). https://doi.org/10.13140/RG.2.2.18968.67849 Kotilevets, I.D., Ivanova, I.A., Romanov, I.O., Magomedov, S.G., Nikonov, V.V., Pavelev, S.A.: Implementation of directed acyclic graph in blockchain network to improve security and speed of transactions. IFAC-Pap. 51, 693–696 (2018). https://doi.org/10.1016/j.ifacol. 2018.11.213 Benˇci´c, F.M., Žarko, I.P.: Distributed Ledger Technology: Blockchain Compared to Directed Acyclic Graph, http://arxiv.org/abs/1804.10013 (2018). https://doi.org/10.48550/arXiv.1804. 10013 Guo, Z., Gao, Z., et el.: Design and optimization for storage mechanism of the public blockchain based on redundant residual number system. IEEE Access 7, (22 Jul 2019) Yu, G., Wang, X., et al.: Survey: Sharding in Blockchains. IEEE Access 8 (9 Jan 2020) Near protocol. Sharding Design: Nightshade Jo, T.: An Exploration of Zero-Knowledge Proofs and zk-SNARKs (2019) Nitulescu, A.: zk-SNARKs: A Gentle Introduction StarkWare: STARK Math: The Journey Begins, https://medium.com/starkware/stark-maththe-journey-begins-51bd2b063c71, last accessed 15 February 2023
Interactive Information System for Automated Identification of Operator Personnel by Schulte Tables Based on Individual Time Series Myroslav Havryliuk(B) , Roman Kaminskyy, Kyrylo Yemets, and Taras Lisovych Lviv Polytechnic National University, S. Bandera Street, 12, Lviv 79013, Ukraine [email protected]
Abstract. The selection of operator personnel to manage technological processes is an important problem in various industries. A large percentage of accidents at enterprises occur due to the fault of workers serving the production. One of the techniques for determining a person’s suitability for such work is reading Schulte’s tables. Today, there is no freely available specialized software to implement this technique. In this work, the authors developed an information system for the automated identification of operator personnel. It implements the method of multidimensional average on the indicators of descriptive statistics of an individual time series. The created system allows users to compare candidates and form their ratings directly. The software product improves the quality of professional selection by automating the process and using an advanced analysis method. Keywords: Time-series Data; Automated System · Professional Selection · Schulte Tables · Personnel Identification · Software Product
1 Introduction The selection of highly qualified specialists, in particular operators, to manage technological processes is an actual problem in many industries. However, information technology has a much faster rate of development than the selection processes of employees of the corresponding level. Special literature reports that a large percentage of accidents at enterprises occur due to the fault of workers serving the production. One of the main reasons for this is the insufficient level of professional selection and training of personnel. Improving the efficiency of personnel selection and certification by automating it with the help of the latest information technologies is an urgent problem in the field of personnel policy of enterprises and institutions. Special techniques are used to select applicants for the position of the operator. They were developed by specialists in the field of psychology and adapted to the specifics of the operators’ activities. One of these techniques is reading Schulte tables. It is designed to determine the stability of attention and working capacity in dynamics. Its main advantage is that the reading process is an example of finding similar images of objects of a given © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 372–381, 2023. https://doi.org/10.1007/978-3-031-36115-9_34
Interactive Information System for Automated Identification
373
class on the monitor, which is a common task in operator work. The technique is also quite simple and has several varieties. In [1], the authors implemented the Schulte table methodology using web development tools. However, this program only has a classic version of the reading table with small modification possibilities. Methods of analyzing the results in this development are also traditional. In [2], the results of the experiment of reading Schulte tables by a group of people are given. Based on the received data, a rating was formed where the participants were placed according to the level of stability of attention. However, only classical methods were used for the analysis. [3] describes the role of the Schulte method in the procedure of professional selection for the operators’ positions. Paper Schulte tables were used for the experiments, and the analysis was also traditional. An attempt to build a model of operator personnel identification is given in [4]. For this, statistical indicators of time series obtained based on experimental data in image processing systems were used. The authors propose using the multivariate average method to assess the working capacity and build a rating of the participants. This study aims to create an interactive information system for the automated identification of operator personnel according to Schulte tables using individual time series. The main contribution of this paper is the practical implementation of the multidimensional average method for analysing results obtained by the Schulte tables.
2 Materials and Methods 2.1 Schulte Tables Methodology The selection of qualified specialists for the position of the operator is an important part of the company’s work. Currently, there are many different psychological methods and criteria for evaluating job applicants. The key indicators for this kind of work are stability of attention and dynamics of working capacity. Using Schulte’s tables technique is convenient for studying these parameters. The simplest Schulte table is a square table divided into 25 identical square cells (5 rows and 5 columns), in which numbers from 1 to 25 are entered in random order (Fig. 1) [5].
Fig. 1. An example of Schulte table
374
M. Havryliuk et al.
The essence of conducting the test according to this method is as follows: the subject is offered the Schulte table. The controller uses a stopwatch to record the time the subject spends searching for and pointing to a number of numbers in the table. The obtained data are used for further analysis. The way to search and identify the required objects is illustrated in Fig. 2.
Fig. 2. The way to search for objects
However, this is only the most common version of the technique. There are many modifications to it: the table can have a different size; instead of a sequence of numbers, any other ordered sequence can be used, such as alphabetic characters; the background color of the cells may vary, most often red and black tables are used [6]; if automation is possible, the table may change after each element found. The use of different types of tables is another reason for the relevance of process automation. 2.2 Operator Activity Modeling Passing tests according to the Schulte method is one of the types of operator activity. In general, the operator transforms the received information into a decision regarding the control of the system’s operation, process, etc. This activity consists in finding the right image on the screen. After finding the image, the operator makes a decision - he presses a button to confirm it. Let G ∗ be an object of a certain defined class that should be detected (in this case, a number or a symbol), which can be randomly placed on the provided image of the training situation (Schulte table) and which can be characterized by a vector of recognition features gi∗ ∈ G ∗ . We denote the set of objects of the same class that are similar in shape to the searched object as G. The operator’s task is to find, detect and recognize the necessary object (number or symbol) among similar ones and make a decision about its placement on the image. During the process of the operator’s work, at the moments ti , specified by the training situation Z(t) monitoring parameters (in this case – at the moments of finding the necessary element of the table), the operator receives images for processing, which are located on the monitor of the automated system. Images xi ∈ X where X = {xi : xi = x(ti ), ti ∈ T , i = 1, N }
(1)
are training models of the environment (Schulte tables) in which the situation Z(t) arose, and illustrate the state of the situation at moments ti i.e. Z(t) → Z(ti ). Each
Interactive Information System for Automated Identification
375
object belonging to the set G has its vector of features gi = gi , •, gr , gi ∈ G, besides, among them, there may be such as those of the desired object, but there must also be signs of difference from the desired object. At the moment of finding the desired object, the operator compares its features with normative features gi∗ ∈ G ∗ , stored in the human memory as a result of educational experience in the form of “standard images” xi∗ ∈ X ∗ . As a result, the operator makes an appropriate decision from several alternatives. Therefore, comparing images xi and xi∗ , operator identifies the situation given to him, chooses and confirms the correct, in his opinion, decision yj ∈ Y , where Y = {yj : yj = y(tj ), j = 1, N , tj = ti + ti ; ti , tj ∈ [0, T ]}
(2)
and tj is the moment of implementation of the decision made by the operator, which the operator selects with a set of commands that are the control vector uj = u1 , •, uh , uj ∈ U , where U is the set of vectors of control commands, and h = 1, 2, • is the number of commands in the j-th situation. Continuous mental tension creates a sense of increased responsibility. When the operator works for a long time, obviously, a person is exposed to a large psychological load. Visual sensitivity decreases, reactions and speed of actions deteriorate, the number of errors increases, and as a result, nervous tension arises. Of course, this condition worsens the efficiency of the operator’s activity. Therefore, the quality of operator work must be linked to certain states ck . A set of such states C = {ck : ck = c(δtk ), δtk = tk+1 − tk ; tk , tk+1 ∈ [0, T ]},
(3)
can be considered discrete. So, having described the general model of individual intellectual work in the system for recognizing and detecting objects among a certain class on training images with a situation Z(t) using the example of Schulte’s methodology, it is possible to formulate the task of operator activity: to find the desired object image xi ∈ X among the on the objects of a certain class, for which the expression gi∗ − gi ≡ 0 is true and then make a decision and implement it using the appropriate command from the vector uj . The goal of operator activity is maximum reliability p → 1 with minimum time expenditure t → 0, i.e. X (gi − gi∗ ) −→ optU (xt , ct , yt ) → min p→1 t→0
(4)
This expression illustrates the operator activity’s main task in optimizing the situation’s solution Z(t)[7]. 2.3 Multidimensional Average Method Since, at the moment, there is no general model for determining the assessment of operator activity, statistical methods can be used to identify applicants [8]. One of the most common methods that allow getting a generalized score based on several features is the search for a multidimensional average [9]. Let there be N objects for identification by
376
M. Havryliuk et al.
M features. Then, xij , i = 1, N , j = 1, M are the values of the i-th object by the j-th feature. Accordingly, xj , j = 1, M are arithmetic average values. Then the multidimensional average ai , i = 1, N for each object can be calculated as: M ai =
xij j=1 xj
N
(5)
The features that are used to search for a multidimensional average in our system are the measures of the descriptive statistics of the time of finding each object (for the Schulte method - a number/symbol): • • • • • •
arithmetic average; median; standard deviation; average deviation; interquartile range; coefficient of variation;
The resulting multidimensional averages can be used to rank objects. The essence of Schulte method involves the minimization of features, so the object with the lowest value of the multidimensional average will be the highest in the rating. This method is advanced for evaluating the operator’s performance, as it allows better use of time series data compared to the classical one.
3 Results The main menu is implemented on the main page of the application. The user has four options for further action (Fig. 3): • • • •
create new Schulte table; read existing Schulte table; see results; exit.
Fig. 3. The main page of the application
Interactive Information System for Automated Identification
377
The user can define parameters for the desired Schulte table on the page for creating a new table. Among the parameters (Fig. 4): • • • • • •
name of the table; type of elements (numbers or symbols of the Ukrainian/English alphabet); number of sequence iterations; table size; static/dynamic location of objects after each found one; background color of cells (white/red-black).
Fig. 4. Page for creating a new table.
After creating a table, all its parameters (including the generated locations of objects on the table) are saved. This is necessary so that all test participants receive absolutely identical Schulte tables. Thus, the comparison of results is direct. All previously created tables are shown on the Schulte table selection page for reading (Fig. 5).
Fig. 5. Page for selecting a table from the available ones
378
M. Havryliuk et al.
After selecting the Schulte table, the user gets to the page of preparation for reading the table. Table parameters are displayed here so that the participant clearly understands the task. Before reading the Schulte table, the user must enter his name (Fig. 6).
Fig. 6. Page for preparing for the test
The Schulte table with the relevant parameters appears immediately after pressing the “Start” button (Fig. 7).
Fig. 7. An example of Schulte table
The required object is considered found when the participant clicks on it. In this case, the corresponding cell is highlighted in green. Also, at this moment, the time spent on finding this object is recorded. All obtained results are saved for further analysis. On the page for viewing the results, the user needs to select the Schulte table and press the “See rating” button (Fig. 8). The rating is displayed in the form of a table, where all participants are ordered according to the defined multidimensional average. The multidimensional average is calculated according to the features described above, which, in turn, are determined based on stored data on reading the Schulte table (Fig. 9). With the help of the formed rating, it is possible to compare applicants for the operator position, which is the main purpose of creating the system.
Interactive Information System for Automated Identification
379
Fig. 8. Page for selecting a table to view the result
Fig. 9. Page for viewing results
4 Discussion Exact analogs of the system for determining psychological characteristics, according to Schulte tables, were not found in free access. In general, many applications with Schulte tables are used as a training game for developing attention[10, 11]. However, they cannot fully fulfill the role of a system for the identification of operator personnel [12]. Their main advantage is a large number of implemented modifications of the technique [13, 14]. Among their main shortcomings: a lack of detailed statistics of passing the test [15, 16]; an inability to save results [17, 18]. When comparing applicants for the operator position, the purity of the experiment is important [19]. Each person must deal with the same Schulte table [20]. The developed system provides such an opportunity for direct comparison of applicants. The program makes it possible to collect and save individual time series for further analysis. The system implements an identification method based on the calculation of a multidimensional average. The multidimensional average method provides rank-based comparisons using the relative values of identified features.
5 Summary and Conclusion The selection of operator personnel to manage technological processes is an important problem in various industries. One of the techniques for determining a person’s suitability for such work is reading Schulte’s tables. Today, there is no freely available specialized
380
M. Havryliuk et al.
software to implement this technique. In this work, the authors developed an information system for the automated identification of operator personnel. It implements the method of multidimensional average on the indicators of descriptive statistics of an individual time series. This method is advanced for evaluating the operator’s performance, as it allows better use of time series data compared to the classical one. The created system allows users to compare candidates and form their ratings directly. The software product can be used for systems of professional selection of operator personnel, in training systems to improve the level of concentration and work capacity, and in diagnostic systems in psychology. In the future, the data obtained with the program’s help can be used to build a model of operator activity. Funding. The National Research Foundation of Ukraine funded this research under project number 2021.01/0103.
References 1. Fedota, M.V., Shasholko, S.I.: Use of interactive trainers as modern means of implementation of speed reading technology. Bulletin of the Bohdan Khmelnytsky National University of Cherkasy 8(2), 182–184 (2020) 2. Kovalchuk, I.: Characteristics of the psychophysiological status of the body in persons with different levels of physical capacity. Student scientific bulletin 36, 7–9 (2015) 3. Hohoman, T.S., Chumayeva, Y.V.: Effectiveness of psychocorrection of mental fatigue of maritime transport operators by means of multifunctional sound-color regulation. Actual issues of psychology in the modern innovative space 10(4), 89–92 (2022) 4. Kaminsky, R.M., Nych, L.Y.: Identification of the intellectual activity of the operator personnel based on experimental data. Bulletin of the Lviv Polytechnic National University. Information systems and networks 715, 134–149 (2011) 5. Nechyporenko, A.S.: New intelligent-based approach for the early detection of disorders: use on rhinological data. International Journal of Image, Graphics and Signal Processing (IJIGSP) 9(8), 1–8 (2017) 6. Bernacki, J., Kołaczek, G.: Anomaly detection in network traffic using selected methods of time series analysis. IJCNIS 7(9), 10–18 (2015) 7. Nych, L.Y., Kaminsky, R.M.: Evaluation of the effectiveness of individual intellectual activity of operators in image processing systems based on experimental data. Bulletin of the Lviv Polytechnic National University. Information systems and networks 699, 193–204 (2011) 8. Hassan, M.M., Mirza, T.: Using time series forecasting for analysis of GDP growth in india. Int. J. Edu. Manage. Eng. (IJEME) 11(3), 40–49 (2021) 9. Mishra, N., Soni, H.K., Sharma, S., Upadhyay, A.K.: Development and analysis of artificial neural network models for rainfall prediction by using time-series data. Int. J. Intelli. Sys. Appli. (IJISA) 10(1), 16–23 (2018) 10. Jain, E.G., Mallick, B.: A study of time series models ARIMA and ETS. Int. J. Modern Edu. Comp. Sci. (IJMECS) 9(4), 57–63 (2017) 11. Izonin, I., Tkachenko, R., Holoven, R., Yemets, K., Havryliuk, M., Shandilya, S.K.: SGD-based cascade scheme for higher degrees wiener polynomial approximation of large biomedical datasets. Machine Learning and Knowledge Extraction 4(4), 1088–1106 (2022) 12. Ramakrishna, M.T., Venkatesan, V.K., Izonin, I., Havryliuk, M., Bhat, C.R.: Homogeneous adaboost ensemble machine learning algorithms with reduced entropy on balanced data. Entropy 25(2), 245 (2023)
Interactive Information System for Automated Identification
381
13. Basystiuk, O., Melnykova, N.: Multimodal approaches for natural language processing in medical data. In: Proceedings of the 5th International Conference on Informatics & DataDriven Medicine. Lyon, France, November 18–20, CEUR-WS.org, pp. 246–252 (2022) 14. Basystiuk, O., Shakhovska, N., Bilynska, V., Syvokon, O., Shamuratov, O., Kuchkovskiy, V.: The developing of the system for autimatic audio to text conversion. Proc. of the IT&AS’2021 Symposium on Information Technologies & Applied Sciences, CEUR-WS.org, pp. 1–8 (2021) 15. Mochurad, L., Hladun, Y.: Modeling of psychomotor reactions of a person based on modification of the tapping test. Int. J. Comp. 20(2), 190–200 (2021) 16. Mochurad, L., Yatskiv, M.: Simulation of a human operator’s response to stressors under production conditions. ICEUR Workshop Proceedings 2753, 156–169 (2020) 17. Zomchak, L, Nehrey, M.: Economic growth and capital investment: the empirical evidence. Lecture Notes on Data Engineering and Communications Technologies 135, 645–652 (2022) 18. Auzinger, W., Obelovska, K., Stolyarchuk, R.: A revised gomory-hu algorithm taking account of physical unavailability of network channels. Communications in Computer and Information Science 1231, 3–13 (2020) 19. Oleksiv, I.B.: Selection of important company stakeholders: theory and practice. Naukovyi Visnyk Natsionalnoho Hirnychoho Universytetu 1, 128–134 (2013) 20. Kuzmin, O.Y., Oleksiv, I.B., Mykhailyak, G.V.: Integral evaluation of company employees’ competence system. Actual Problems of Economics 155(5), 506–513 (2014)
DIY Smart Auxiliary Power Supply for Emergency Use Nina Zdolbitska(B) , Mykhaylo Delyavskyy, Nataliia Lishchyna, Valerii Lishchyna, Svitlana Lavrenchuk, and Viktoriia Sulim Lutsk National Technical University, Lutsk 43000, Ukraine [email protected]
Abstract. Implementation of Internet of Things technologies in various fields is one of the priority areas of modern development. Energy efficiency is one of the urgent problems in Ukraine today. As part of this study, the use of IoT will help monitor and control energy consumption data. Auxiliary power supplies are devices that provide supplemental power to an existing power source. They are often used in automotive, electrical, and industrial applications. DIY auxiliary power supplies are those that are built by the user instead of purchasing a readymade product. Building your own power supply can be a cost-effective option, as well as an enjoyable DIY project. The purpose of the research is to develop a device based on the existing electrical engineering components, which allows you to completely or partially replace the functionality of industrial analogs of power supplies. Keywords: Auxiliary power supply · uninterruptible power supply · inverter · battery · DC-DC converter · communication · lighting
1 Introduction 1.1 Problem Statement The creation of an alternative auxiliary power source for communication devices and the organization of emergency lighting, which will be available for self-production, will not contain expensive components and will have advantages in compared to industrial models, is topical task in emergency situations. The need for lower price devices arises under unforeseen circumstances, such as war, natural disasters, limited resources, low purchasing power of consumers. You may have mission-critical life support systems that must continue to function even during an extended blackout. The key challenges. The auxiliary power supply must simultaneously provide safe and reliable power, high performance, low power consumption, and low bill-of-materials. An auxiliary power supply is an additional power source used to supplement the main power supply of a system or device. It is used to provide extra power when it is needed or when the main power source fails. Auxiliary power sources can range from © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 382–392, 2023. https://doi.org/10.1007/978-3-031-36115-9_35
DIY Smart Auxiliary Power Supply for Emergency Use
383
small battery backups to large fuel-burning generators. They are often used to power critical security systems and other electronics that must remain operational even if the main power source fails. 1.2 Analysis of Recent Research and Publications Implementation of Internet of Things technologies in various fields is one of the priority areas of modern development. Energy efficiency is one of the urgent problems in Ukraine today. As part of this study, the use of IoT will help monitor and control energy consumption data [1]. The Internet of Things in the energy sector optimizes work at all levels: from the moment of data transfer from the user to the state of the power grid and generating capacities [2, 3]. Modern smart home automation technologies are becoming more and more advanced, able to make life easier and save money. Home automation technologies using low-cost components such as the Arduino microcontroller, control modules and sensors can completely or partially replace the functionality of industrial IOT devices [4–6]. An auxiliary power supply is required for many devices, so they are the object of research in many scientific and practical articles. Technologies that include auxiliary power supply systems are used on an industrial scale for suburban and regional electric transport and trains, energy-efficient street lighting systems [7] that are parts of a smart city, in wind turbines technology [8], solar panels [9], in microturbines and other elements of aircraft and automobile structures (provides electrical and pneumatic power to various on-board subsystems), in alarm systems [10] (fire alarm, air alarm, video surveillance). Numerical modeling and analysis using the Matlab/Simulink Simulation System including an auxiliary power supply (lithium battery and a supercapacitor), a DC-DC converter, a DC-AC inverter were carried out in [11]. UPS has found many applications, for example, they are used to power telecommunications and security systems, computers, and devices that require increased reliability of power supply [12, 13]. For ordinary citizens, the use of additional power sources is also very important. In this difficult time for Ukraine, the problem of providing small and large gadgets with uninterrupted power supply is especially relevant. Smartphones, tablets, and laptops need special attention. There is a need for emergency lighting of apartments, houses, and other premises in order to maintain life comfort during scheduled and emergency power outages. Inefficient, cheap, unreliable solutions are available on the market, while expensive ones are not available for mass use. For example, industrial designs of such companies as RIELLO, BLUETTI, and EcoFlow, which offer functionality and high reliability, but also have inflated prices, are popular among European brands on the market. Backup UPS kits from the Ukrainian manufacturer LogicPower are also popular. Wartime has an essential impact on the excessive hype in the Ukrainian market for small auxiliary power devices such as PowerBank and UPS. They are often impossible to buy because they are not physically available from sellers, and product deliveries are delayed and irregular due to various circumstances. Therefore, the question arises of how to solve the problem of powering communications equipment and organizing any emergency lighting with the help of available improvised means and materials [14].
384
N. Zdolbitska et al.
1.3 Purpose of the article The purpose of the research is to develop a device based on the existing electrical engineering components, which allows you to completely or partially replace the functionality of industrial analogs of power supplies. The object of research is the technology for the development of alternative sources of power supply.
2 Development Requirements Auxiliary power supplies can provide additional reliability to critical systems by allowing them to continue operating even if the main power fails. They can also provide additional power during peak demand or when the primary power is operating at lower peak power. Auxiliary power can be used to provide transient power during startup or shutdown operations. It is necessary to take into account the safety requirements for the operation of devices and systems. Comprehensive measures are needed to protect both the person and the device to prevent accidents. The following aspects are especially important: protection against malfunctions, continuity of power supply, operational features. When creating an auxiliary household power supply, it is necessary to provide for forced air cooling, reduce the weight and lower the price of components, and ensure a minimum noise level.
3 Research Results 3.1 Possibility of Use When the power of the main network is turned off, the Internet, which is necessary for work, education, entertainment, and other communications, usually disappears. It often happens that to access the Internet, it is enough to connect to power the router, since on the provider’s side the corresponding communication equipment can have its emergency power supply, thanks to which the provider will deliver the service even in the absence of electrical energy. This method is most likely available from providers with optic fiber lines connected to consumers. Another example is a mobile communication channel through a mobile phone or modem. They also need to be powered from an auxiliary power supply to get a long Internet connection, since the built-in battery usually does not last long, and mobile operators provide 4G/3G service for a rather long period since this is a critical infrastructure. In particular, there is a question of emergency main or additional lighting. Ordinary incandescent light bulbs are not very suitable for such purposes, energy-efficient LED lamps are usually used. LED lamps are the best solution today for both the main and emergency lighting of premises. The only problem is the voltage level that is needed for the lighting network. Usually, this is a mains voltage of 230 V, which is not available in simple power banks or batteries that are used for auxiliary power gadgets.
DIY Smart Auxiliary Power Supply for Emergency Use
385
It’s necessary to calculate the power and find a compromise between consumption and system autonomy. Additional costs are also required for the organization of an integrated solution in the existing lighting system. Therefore, an urgent task is to create such an auxiliary power source, available for self-manufacturing, based on inexpensive components. Such a device can be a homemade (DIY) auxiliary power supply based on 12-V AGM batteries and a combination of computer components with automotive gadgets. In this device, you can reuse secondary components, such as batteries from uninterruptible power supplies, which are periodically replaced at critical sites of enterprises, security alarms, and transport companies. The main emphasis in such a system is on the reuse of elements, due to which economic and environmental benefits are achieved, as well as versatility. An auxiliary power supply with voltages of 5/12/230 V has been developed for charging mobile gadgets and organizing emergency lighting based on the existing component base of electrical engineering, which will allow replacing the functionality of industrial analogs of power supplies in full or in part. 3.2 Advantages and Disadvantages of Auxiliary Power Supply Advantages of auxiliary power supply: – can be used as an additional power supply along with the main power supply (reducing the load on the power grid); – can be used to provide emergency power during a power outage; – it is easy to install and maintain; – reliable and cost-effective solution for providing power to remote areas or locations without grid access; – scalable based on power requirements; – it is an eco-friendly source of renewable energy, can help reduce the use of fossil fuels, and does not produce any emissions; – it is a cost-effective and reliable way to provide electricity and reduce your bills; – can be used portably for a short period; – can be used in emergencies as a reliable backup power source; – the device is quiet (low noise level); – option of voltage selection (5/12/230 V); – a safe power supply that can shield gadgets from sudden voltage changes; – lower cost than industrial counterparts; – ease of maintenance. Disadvantages of auxiliary power supply: – – – – – –
limited energy storage capacity; relatively short battery off-line life; the need for investment of money and time; difficulty finding compatible parts; sensitivity to weather conditions (must be used indoors); unpredictable performance (due to component reuse).
386
N. Zdolbitska et al.
3.3 The Architecture of the Developed Device (MVP) The developed auxiliary power supply (Fig. 1) contains several USB ports and can charge small gadgets with a power consumption of up to 15 watts. The available power of the lighting system will depend on the built-in 230 V inverter and the capacity of the battery. For example, in the proposed system, a low-power automobile inverter is used, which allows healing a maximum load of 150 watts, and the capacity of the installed batteries is sufficient for emergency lighting at maximum power for 4 h.
Fig. 1. Auxiliary power supply
To display the general device structure, its main nodes, and the links between them, as well as to understand its functionality and operating modes, we present a block diagram of the device (Fig. 2). The main nodes of the designed system shown in Fig. 1 are: – – – – – – –
12 V battery pack with 42 Ah capacity of AGM structure (6 elements); 12–220 V inverter with an approximate sine wave at the output; DC-DC converter 12 V to 5 V; a charger that allows you to charge a 12 V battery pack from a 220 V network; a switching scheme that ensures interaction between blocks; the control panel, which is used to switch between work modes; a control and display unit that allows you to visually control the operating mode of the device.
Lead acid batteries are sold in 6 V and 12 V. You need to connect them in series to increase the voltage, or in parallel to increase the available amp-hours (Fig. 3). Maintenance of lead acid batteries: it’s necessary to top it off periodically with distilled water to keep your lead acid battery well maintained and get at least its minimum life expectancy.
DIY Smart Auxiliary Power Supply for Emergency Use
387
Fig. 2. Structural diagram of the device
Fig. 3. Lead acid battery CSB 12V/7.2Ah
Advantages of lead acid batteries: – – – – – – –
wide range of sizes and specifications; relatively low cost; high output power; widely available, many producers worldwide; being able to deliver high currents, being tolerant of high-rate discharges, and having a longer life cycle than other battery technologies. Disadvantages of lead-acid batteries:
– – – – –
big weight and bulky; low cycle life; high self-discharge rate; maintenance required; corrosive electrolyte.
A power inverter (Fig. 4) converts 12-V DC power to standard household 230-V AC power. It enables quick USB charging for different devices, AC power socket can be used as an emergency power supply. It has silent cooling, produces modified sine wave. DC-DC buck converters have compact size, high reliability, low cost, wide input voltage range, making them ideal for applications [15, 16].
388
N. Zdolbitska et al.
Fig. 4. Inverter
Fig. 5. Lead-acid charger
The charger (Fig. 5) can supply enough current to charge the battery and keep up with the inverter’s load. It suitable for battery lead-acid maintenance battery. 3.4 Remote Control and Indication The sensors and structural modules listed in Table 1 were used to expand the functionality of the developed device. For more comfortable use of the auxiliary power source, it can be equipped with a Bluetooth interface for remote control and management. The easiest way is to use a ready-made Bluetooth module like HC-05/HC-06 and an Arduino Uno board (Fig. 6).
Fig. 6. HC-06 Bluetooth module and its connection diagram to Arduino UNO
To connect a Bluetooth module to the Arduino, necessary to connect the module’s TX pin to the Arduino RX’s pin, the module’s RX pin to the Arduino’s TX pin, and the
DIY Smart Auxiliary Power Supply for Emergency Use
389
Table 1. Sensors and structural modules The sensors and structural modules
Functionality 1-bit module AC 230V voltage detect board with optocoupler isolation function: testing whether AC 230V is existed
INA219 module with I2C interfaces and zero drift bi-directional function: current power supply sensor INA3221 triple-channel module with I2C interfaces and zero drift bi-directional function: triple-channel current power supply sensor ACS712 Hall current sensor module function: single channel current sensor 5A LM2596 power converter step down module DC-DC 1.5V-35V
1way 5V relay module with optocoupler
LCD display – measure and display battery capacity and voltage function: indication KCD-11 push button switch 10x15mm 3A 250V / 30A 12V
Car cigarette 12V lighter socket made of heat-resistant plastic
module’s ground pin to the Arduino’s ground pin. It may be necessary to connect the module’s VCC pin to the Arduino’s VCC pin, depending on the module’s requirements. We can send commands to the Arduino board or receive data from the HC-05 module using our own program. The base software for the Arduino platform was developed in the Arduino IDE, and the Android Studio software was used for mobile development. The Bluetooth module allows remote control of the DIY device: – to connect or disconnect it to the batteries using a relay; – to receive data from sensors;
390
N. Zdolbitska et al.
– notify about the presence/absence of 230 V network voltage; – report the main battery voltage; – inform power consumption of the main connected devices on one/three channels of the controller.
3.5 System Load Testing To test the system, an 11W LightMaster LB-680 LED lamp (LightMaster LB-680 11 W/A60/E27/230 V/50 Hz) was used. The lamp was connected to the inverter with a two- wire copper cable 8 m long and with a cross-sectional area of 2.5 mm2 . The measured inverter self-consumption (no-load current) was 0.26 A at a nominal voltage of 12.8 V. The measured inverter cut-off voltage (load off) was 10.6 V. The measured total self-consumption of display devices was 0.055A at a voltage of 12.8 V. A fully charged battery pack has a voltage of 13.7 V. The measured values of the duration of the system’s operation until complete shutdown are shown in Fig. 7. The data are correlated by the number of cells in the battery.
Fig. 7. System load testing
The experimental results are in good agreement with theoretical calculations. Since the batteries actually used for our system were not new, but used ones, they usually have a smaller capacity than the nominal value, so the obtained experimental results differ downward from the theoretically calculated ones.
4 Summary and Conclusion The developed auxiliary power supply is perfect for home use as an alternative power source. This device may be the only solution for households (especially for residents of high-rise buildings) during planned power outages, voltage fluctuations, and systematic blackouts. The DIY system has an open architecture, so there is a possibility for improvement in consumer characteristics, for instance, an increase in battery capacity will increase the duration of the entire system.
DIY Smart Auxiliary Power Supply for Emergency Use
391
The environmental component should also be noted. DIY device does not have harmful emissions, does not create additional noise, unlike generators with internal combustion engines, and meets improved fire safety standards, so it can be placed indoors. Our device is cheaper than manufactured due to the reuse of secondary components (power supplies and a computer system unit case).
References 1. Tsindeliani, D., Povstyana, Y., Lishchyna, N., Yashchuk, A.: Latency reduction in realtime GPS tracking in Android and the web-based GPS Monitoring System. In: 2022 12th International Conference on Dependable Systems, Services and Technologies (DESSERT) (2022) 2. Wikiman, O., Thomas, S., Nzerem, P., Koyunlu, G.: Design and construction of a prototype wireless power transfer device. Int. J. Eng. Manuf. 2019(9), 16–30 (2019) 3. Adamu, S., Bature, U.I., Nasir, A.Y., Hassan, A.M., Jahun, K.I., Toro, U.S.: IOT controlled home automation technologies. In: 2019 2nd International Conference of the IEEE Nigeria Computer Chapter (NigeriaComputConf), pp. 1–7. Zaria, Nigeria (2019) 4. Ali Shah, S.K., Mahmood, W.: Smart home automation using IOT and its low cost implementation. Int. J. Eng. Manuf. 10(5), 28–36 (2020) 5. Cruz, L., Griño, M., Tungol, T., Bautista, J.: Development of a low-cost air quality data acquisition IoT-based system using arduino leonardo. Int. J. Eng. Manuf. 9(3), 1–18 (2019) 6. Akwu, S., Bature, U.I., Jahun, K.I., Baba, M.A., Nasir, A.Y.: Automatic plant Irrigation control system using Arduino and GSM module. Int. J. Eng. Manuf. 10(3), 12–26 (2020). https://doi. org/10.5815/ijem.2020.03.02 7. Jha, A., Maharjan, M.: Smart lighting system using LoRa WAN technology. Int. J. Eng. Manuf. (IJEM) 12(1), 48–53 (2022) 8. Rebello, E., et al.: Developing, implementing and testing up and down regulation to provide AGC from a 10 MW wind farm during varying wind conditions. J. Phys.: Conf. Ser. 1102, 012032 (2018). https://doi.org/10.1088/1742-6596/1102/1/012032 9. Holovan, M., Zdolbitska, N., Lishchyna, V., Hrinyuk, S.: The analysis of the productivity of the solar panels automatic positioning system. Comput.-Integr. Technol.: Educ. Sci. Prod. 41, 23–29 (2020) 10. Terletskyi, T., Kaidyk, O., Tkachuk, A., Zabolotnyi, O., Cagáˇnová, D.: Ensuring the reliability of functioning of non-addressed fire alarm. EAI Endorsed Trans. Energ. Web 22, e11 (2021) 11. Kuo, J.-K., Huang, P.-H., Wang, C.-F.: Numerical modeling and analysis of PEMFC integrated with auxiliary power source. Int. J. Energ. Res. 37, 1635–1644 (2013) 12. Raju, E.S.N.P, Jain, T.: Distributed energy resources and control. In: Distributed Energy Resources in Microgrids (2019) 13. Ahmed, U., Ali, F., Jennions, I.: A review of aircraft auxiliary power unit faults, diagnostics and acoustic measurements. Prog. Aerosp. Sci. 124, 100721 (2021). https://doi.org/10.1016/ j.paerosci.2021.100721 14. Ostapchuk, O.V., Zdolbitska, N.V.: Auxiliary power supply of means of communication and the organization of emergency lighting. In: The role of innovations in the transformation of the image of modern science: Materials of the VI International Scientific and Practical Conference, pp. 139–143. Kyiv, 23–24 Dec 2022
392
N. Zdolbitska et al.
15. Zapata, J.W., Kouro, S., Carrasco, G., Renaudineau, H., Meynard, T.A.: Analysis of partial power DC–DC converters for two-stage photovoltaic systems. IEEE J. Emerg. Sel. Topics Power Electron. 7(1), 591–603 (2019) 16. Zdolbitska, N.V., Zdolbitskyy, A.P., Sopizhuk, R.V., Suprunyuk, V.V.: DC-AC converter with microcontroller-controlled frequency inverter. Comput.-Integr. Technol.: Educ. Sci. Prod. (23), 25–31
A Computerised System for Monitoring Water Activity in Food Products Using Wireless Technologies Oksana Honsor(B) and Roksolana Oberyshyn Lviv Polytechnic National University, Lviv 79070, Ukraine {oksana.y.honsor,roksoliana.r.oberyshyn}@lpnu.ua
Abstract. The food industry needs constant monitoring of both the technological process and the quality and safety of finished products. Control of the water activity value is essential to ensure the appropriate quality level of food products during their production because it affects the growth of bacteria and microorganisms, affecting the quality indicators, shelf life and proper storage of products. Modern enterprises must have an automated quality monitoring system. A specialised computer system for monitoring water activity should be an integral part of the overall quality monitoring system. This makes it possible to quickly react to deviations in the value of the water activity indicator and take preventive actions. The structural scheme of such a specialised computer system will be presented in this article. It is also important to correctly choose the type of water activity analyser based on the main parameters of the sensors, which are described in this article. The basic principles of developing the necessary software and a block diagram of the data flow algorithm in the system are also presented. Keywords: Water activity · computer system · food · quality monitoring
1 Introduction Water activity aw is a parameter that determines the ratio between the water vapour pressure of the product and the saturation pressure of pure water at the same temperature. It is an essential indicator of product quality level in various industries, in particular, food, pharmaceuticals, industrial production, tobacco, and seed storage. The water activity measurement limits are from (0 to 1) aw . The activity of water affects the following properties of the product: microbiological stability, chemical stability, enzyme stability, colour and taste, nutritional value, protein and vitamin content, composition stability, expiration date, solubility and texture, storage and packaging. That is why constant monitoring of this indicator within the general product quality control system is important and necessary. A specialised computer system using wireless technology is best suited for this. The value of water activity in food products can be controlled with the help of various additives (moisturisers), using satisfactory packaging materials and maintaining © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 393–403, 2023. https://doi.org/10.1007/978-3-031-36115-9_36
394
O. Honsor and R. Oberyshyn
favourable ripening and storage conditions. However, the most critical point of control is the final stage of food product manufacturing. If too much water is available, there is a risk of microbial growth and water migration. Food manufacturers today must prove to food quality and safety organisations that the product’s water activity has been reduced to the point where bacteria cannot grow. The diagram shown in Fig. 1 can help better understand the relationship between the value of product moisture and water activity and their influence on the leading chemical and biological indicators of product quality [1, 2].
Fig. 1. The effect aw on oxidation, browning reactions, enzymatic and microbial activity.
The grey zone in Fig. 1, which corresponds to values from (0.3 to 0.5) aw , demonstrates the lowest chemical and biological activity, which leads to less spoilage and an increase in the shelf life of products. There are a number of devices for measuring water activity. They have high accuracy, efficiency, compatibility and ease of calibration, which are modern and have good characteristics (for instance, the device for measuring water activity “AwTherm” (accuracy ± 0.005aw , ± 0.1 K), laboratory instrument “Hygrolab” (accuracy ± 0.008 aw at 23 °C), station probe HC2-AW-USB). However, these devices are primarily stationary and make it possible to record measurement results manually, without storing them in a database, processing and graphical presentation. Modern food enterprises (in particular, those that make cookies and crackers) strive to reach a new level of quality control and apply non-destructive determination of the content of active water in products without interrupting the production process [3]. However, non-destructive methods have unfortunately not been studied at a sufficient level.
2 Related Work The problem of measuring water activity is deeply studied nowadays, but it started many years ago [4, 5]. In particular, the general properties of this indicator and the main methods of its measurement are considered in [6, 7]. The moisture transfer rate
A Computerised System for Monitoring Water Activity
395
during drying and through the packaging film or edible coating of food products during storage can be estimated using water activity. As a result, drying conditions, packaging, or coating material can be selected based on the identified properties of the food products [8]. The importance of correctly choosing methods and means for determining water activity is highlighted in [9, 10] since this indicator is sensitive to temperature and the content of volatile substances in the product under study [11, 12]. Features of the application of types of measuring devices for determining the activity of water, in the case that the tested substance contains volatile substances, are considered in [13, 14]. It is also important to correctly choose the method of measuring water activity for its introduction into the enterprise’s general computerised quality monitoring system. Similar systems for monitoring quality parameters are considered in [15, 16]. Based on a thorough analysis of related information, new ways of solving the problem are proposed.
3 Methodology and Regulatory Principles for Determining Water Activity in Food Products 3.1 Regulation of the Water Activity Indicator in the Main International Regulatory Documents The growing recognition of the principle of water activity is illustrated by its inclusion in FDA and USDA regulations, GMP and HACCP requirements, and most recently, in NSF International Standard 75. Let’s consider the main ones. 1. The HACCP system describes the need to apply aw for food control and safety measures through risk analysis and critical control points [17]. 2. The system of standards Good Manufacturing Practice (GMP) regulates the use of the water activity indicator as a means of microbiological control; 3. The Food and Drug Administration (FDA) addresses the interaction of aw and pH under certain conditions of heat treatment and packaging. This synergism controls or eliminates pathogens that would otherwise be ineffective when used alone. 4. The United States Department of Agriculture (USDA) and the Food Safety and Inspection Service (FSIS) have included water activity as a microbial control in their Generic HACCP Model 10«Heat-treated, Shelf-stable Meat and Poultry Products». 5. Standard ANSI/NSF 75–2000 «Non-potentially hazardous foods». The standard regulates test methods and evaluation criteria (in particular, water activity) that allow determining that a food product meets the requirements of the FDA Food Code for “potentially hazardous food” and does not require refrigeration for safety [18, 19]. 6. International standard ISO 18787:2017. Foodstuffs – Determination of Water Activity – establishes basic principles and specifies requirements for the methods of determining water activity (aw ) of food products for human consumption and animal feed within a measurement range of 0 to 1 [20]. The methods described in this standard are based on the measurement of the dew point or on the determination of the change in the electrical conductivity of the electrolyte or the dielectric constant of the polymer. However, it should be noted that the methods for measuring water activity described
396
O. Honsor and R. Oberyshyn
in the standard cannot be applied to frozen products, to products in the state of waterin-fat emulsion, and also to crystalline products such as sugar, salt or minerals. Also, when the tested product contains volatile compounds, such as alcohols, the method may require the adaptation of special equipment. 3.2 Main Methods and Instruments for Water Activity Measuring Water activity in food products is often referred to as “free water” or “non-chemically bound water”. It is important not to confuse this concept with the mass fraction of moisture in the product because these are conceptually different product properties that reflect other quality indicators. The mass fraction of moisture characterises the moisture content in the product, and its value depends on the mass of the product (1): ω=
mmoisture . mmoisture − mdrysubstance
(1)
The value of water activity does not depend on the mass of the product and reflects the presence of “free” water in the product: aw = p/p0 ,
(2)
p – water vapour pressure above the product. p0 – water vapour pressure above pure water [2]. An important feature of the water activity indicator is its temperature sensitivity. Measurements can only be made when the product sample, sample holder and measurement sensors are at stable temperatures. Many standards require measurements to be made at a specific temperature [16]. Three main modern methods for measuring water activity should be distinguished, on which the principle of operation of modern laboratory equipment is based – the method using an adjustable diode laser, the cooled mirror method, and the capacitive (or conductivity) method. Simplified structural diagrams of devices using these methods are given below. Figure 2 shows the structural diagram of measuring water activity by the capacitive method. The capacity of the sensor varies depending on the number of water molecules in the air. The sample of the researched product should be placed in a special cup, which in turn is placed in a steel holder (sample holder). After that, the cover with the measuring head must be tightly closed. It is essential to ensure a tight connection between the measuring head and the sample holder because only then will the system be closed, and equilibrium in the system will be achieved. Water activity can be measured in two ways: either by using a predictive model or by waiting for the water vapour pressure and temperature to reach equilibrium in the measuring chamber [8]. This measuring circuit also includes a temperature sensor and a humidity sensor. This method has a reasonably high accuracy but requires a long time for determination aw. Commercially available instruments measure over the entire water activity range with an accuracy of ±0.015aw . Figure 3 shows the structural diagram of the device for determining the activity of water by the method of a cooled mirror and using an optical sensor.
A Computerised System for Monitoring Water Activity
397
Fig. 2. Structural diagram of a laboratory device for determining the activity of water by the capacitive method
Fig. 3. Structural diagram of a laboratory device using the chilled mirror method
As air is passed over a chilled surface whose temperature is gradually reduced, the temperature at which the water vapour begins to condense on the surface is the dewpoint temperature, which is directly related to the vapour pressure of the air. The system consists of a sensor block containing a dewpoint sensor, an infrared thermometer, and a fan. A sample is placed in a sample cup which is sealed against the sensor block. The dewpoint sensor measures the dewpoint temperature of the air, and the infrared thermometer measures the sample temperature. The relative humidity of the headspace is calculated as the ratio of dewpoint temperature saturation vapour pressure to saturation vapour pressure at the sample temperature. When the water activity of the sample and the relative humidity of the air is in equilibrium, measurement of the headspace humidity gives the water activity of the sample. Chilled-mirror instruments make accurate (±0.003 aw ) measurements in less than 5 min without the need for calibration [21]. The method using a diode laser (TDL – Tunable Diode Laser) has a high measurement speed and meets the requirements for measurement accuracy [12, 13]. The structural diagram of the device for measuring water activity using TDL is shown in Fig. 4. The sensor emits a precisely tuned infrared laser beam through the space above the sample. A laser beam less than one nanometer wide is specific for a common isotope of water. Other vapour molecules, including alcohols, gasoline, organic solvents, and propylene glycol, do not affect readings. The attenuation of the beam is measured by the LSD laser receiver, and directly from this value, the concentration of water molecules
398
O. Honsor and R. Oberyshyn
in the air of the chamber is determined. The vapour pressure determined by the TDL is then divided by the saturated vapour pressure at the sample temperature to obtain a water activity value.
Fig. 4. Structural diagram of water activity measurement using TDL.
The diode laser sensor is the only existing sensor that can accurately measure water activity in samples containing significant concentrations of volatile substances. The temperature of the sample is measured using an infrared sensor. The temperature value should be maintained at 25 °C, as the water activity indicator is temperature sensitive. 3.3 Analysis of the Main Criteria for Choosing a Method of Measuring Water Activity Several criteria should be considered to correctly choose the method and type of sensor for measuring water activity for a specific product and under particular conditions. Table 1 shows several main criteria characterising the measurement methods discussed in this article above [6, 13]. Table 1. Main characteristics of sensors Criteria/sensor type
Capacitive sensor
Mirror dew point sensor
Diode laser TDL
Measuring range
0.0 – 1.00 aw
0.050 – 1.000 aw
0.000 – 1.000 aw
Measurement accuracy
± 0.02 aw
± 0.003 aw
± 0.005 aw
Measuring time
≤ 5 min
≤ 5 min
≤ 5 min
Temperature control
Not necessary
15 – 50 ºC ± 0.1ºC
15 – 50 ºC ± 0.1ºC
Resolution
± 0.0001 aw
± 0.0001 aw
± 0.0001 aw
Volatile substances influence
Yes
Yes
No
Calibration
Necessary
Necessary
Not necessary
A Computerised System for Monitoring Water Activity
399
First, you should determine the type of product and its consistency, packaging methods and terms, as well as conditions of storage. Also, an important parameter is the measurement’s accuracy and the method’s speed. The content of volatile substances in the product can also affect the measurement result. As expected, the water activity indicators of the mirror dew point sensor were significantly affected by the presence of volatile substances. The capacitive sensor is also sensitive to some types of samples containing volatile substances. In addition, the capacity sensor, being a secondary method of determining water activity, is not as fast and accurate as the cooled mirror sensor. The performance of the TDL water activity sensor against water activity standards was equal to or better than the best water activity testing options currently available. The highest performance in water activity testing combined with ease of use and lack of sensitivity to any type of sample makes TDL the preferred method for measuring water activity [12].
4 A Specialised Computer System for Monitoring Water Activity 4.1 Hardware design Modern enterprises must have a computerised control system for all quality indicators to ensure the appropriate quality of the products produced. It will make it possible to monitor quality indicators, analyse them and promptly respond to possible deviations. According to the HACCP system, analysis of indicator deviations will make it possible to effectively prevent their occurrence at critical control points. Such a system has three subsystems and channels that connect them together (Fig. 5).
Fig. 5. A structural diagram of a computerised system for quality control at the enterprise.
At the first level, the collection and initial processing of data on quality indicators from the relevant control points is carried out. This subsystem consists of multiparameter sensors, controllers and additional wireless communication devices for transmitting information from the sensor to the controller. Controllers collect data and process it. The second layer is a data transmission subsystem consisting of wireless communication devices with built-in security features that transmit data from the controller to a cloud storage environment.
400
O. Honsor and R. Oberyshyn
The third level is the data management subsystem. It contains an application that accesses the cloud storage environment and displays it to the end user. Connecting link I – measuring channels, which include all measuring devices and communication lines from the metering point to the controller. Connecting link II – communication channels. Physical wired and wireless communication lines are used as communication channels. Water activity is an important comprehensive indicator that affects foods’ taste, smell and texture, microbiological stability (bacterial growth) and shelf life. Permanent control, analysis and processing of data on water activity should be an integral part of the computerised quality control system at the enterprise. Consider the structural diagram of a specialised computer system for monitoring the water activity indicator (Fig. 6).
Fig. 6. Block diagram of proposed Wireless Measuring System
The data received from the analyser is passed through a measuring transducer to form an analogue signal that meets the requirements of the next station for further work with it. After that, the converted data is sent to the controller (PIC16F877A can be used, for example). The controller has a built-in ADC that converts an analogue signal into a digital one for further processing. The received digital data is sent to the ARM processor using the RF module, as shown in Fig. 6. The ARM processor model can be used LPC2148, for example. At this stage, the received data can be displayed on the LCD for initial control of their compliance with the requirements. The results are also sent to the remote monitoring station using the ZigBee module [14, 15]. In the third stage, the converted digital data is sent via serial communication to the server equipped with specialised programs and a Graphic User Interface (GUI), as shown in Fig. 6. The received data is presented in a graphical interpretation for better visual perception by specialists. Special programming and numeric computing platforms (for ex, MATLAB) are used and will be saved for further reference. In addition, the obtained data must be compared with the standard values of the water activity. The specialist and authorised persons are notified if unacceptable deviations are detected to take the necessary preventive and corrective actions. One of the options for sending a message can be an SMS notification.
A Computerised System for Monitoring Water Activity
401
4.2 Software Design A detailed block diagram of the algorithm of the entire system, as well as software development, is shown in Fig. 7.
Fig. 7. General workflow of water activity monitoring system
The software for the water activity control system can be conditionally divided into three parts, the main part of which is the programming of the microcontroller. Next, the software for the ARM processor and the design of the graphical user interface for the interpretation of the received data are developed. It is crucial to ensure the correct interaction of the GUI software with the hardware at the remote monitoring station.
5 Conclusion Regular control of water activity in food products is essential and necessary. According to the requirements of regulatory documents, the control of this indicator should be introduced into the general product quality monitoring system. Typically, this indicator is monitored using laboratory methods followed by manual registration and subjective analysis of the results. This article proposes a specialised computer system that makes it possible to monitor water activity by the most appropriate method, to carry out the primary transformation of
402
O. Honsor and R. Oberyshyn
the received data, to transfer them to an ARM processor for further processing, analysis and storage. The results are transmitted to the remote station to the relevant specialists, where it is possible to display and graphically interpret them using a specialised user interface. Immediate notification of responsible persons in case of critical values of water activity is also foreseen. An important feature of the proposed computerised system is wireless data transmission technologies, which make the system convenient and unified. A detailed block diagram of the algorithm of the entire system and software development is proposed. The implementation and testing of this system at a food enterprise (production of cookies and crackers) are planned as a further direction of scientific work. Acknowledgement. This project is supported by the Institute of Computing Technology, Automation and Metrology of Lviv Polytechnic National University.
References 1. Simons, C.: Determination of water activity. Food science. https://cwsimons.com/determina tion-of-water-activity 2. Mathlouthi, M.: Water content, water activity, water structure and the stability of foodstuffs. Food Control 12(7), 409–417 (2001) 3. Saqaeeyan, S., Rismantab, A.: A novel method in food safety management by using case base reasoning method. Int. J. Intell. Syst. Appl. 7(10), 48–54 (2015) 4. Sandulachi, E.: Water activity concept and its role in food preservation. Meridian ingineresk 39–48 (2012) 5. Van den Berg, C.: Description of water activity of foods for engineering purposes by means of the GAB model of sorption. In: McKenna, B.M. (eds.) Engineering and Food, pp. 311–321. Elsevier Applied Science, London, England (1984) 6. Sahin, S., Sumnu, S.G.: Water activity and sorption properties of foods. In: Physical Properties of Foods, pp. 193–228. Springer New York, New York, NY (2006). https://doi.org/10.1007/ 0-387-30808-3_5 7. Water activity: Solutions for laboratories. https://www.hlr.ua/storage/editor/files/2486ca5be fa76fdd0c325a24820c613a.pdf 8. Mitrevski, V., Geramitcioski, T., Mijakovski, V., Andreevski, I.: Water activity VS. equilibrium moisture content. J. Process. Ener. Agric. 20, 69–72 (2016) 9. How to carry out a water activity (Aw) measurement. https://www.processsensing.com/enus/blog/how-to-do-water-activity-measurements 10. Devices for determining water activity. https://labtime.ua/uk/produkciya-c2/laboratornyepribory-c6/oborudovanie-obschelaboratornoe-c31190/pribory-dlya-opredeleniya-aktivnostivody-c10069 11. Rahman, M.S., Shyam, S.S.: Water activity measurement methods of foods. Food Properties Handbook 9–32 (2009) 12. Ren, Y., Jiewen, G., Sicheng, S., Shyam, S.: Tang Juming Understanding water activity change in oil with temperature. Current Res. Food Sci. 3, 158–165 (2020) 13. Campbell, G.S., Galloway, M., Campbell, Z.: Measurement of water activity in the presence of high volatile concentrations using a tunable diode. laser-single laboratory validation, first action 2021.04. J. AOAC Int. 105(3), 649–656 (2021)
A Computerised System for Monitoring Water Activity
403
14. Carter, B.P., Brown, G.: Superior Water Activity Measurement of Oils and Lubricants Using a Tunable Diode Laser. AquaLab Univercity (2015). https://usermanual.wiki/Document/146 44SuperiorWaterActivityMeasurementofOilsandLubricantsUsingaTunableDiodeLaser 15. Barabde, M.N., Danve, S.R.: A review on water quality monitoring system. Int. J. VLSI Embedded Syst.-IJVES 06, 1475–1479 (2015) 16. Li, Z., Wang, K., Liu, B.: Sensor-network based intelligent water quality monitoring and control. Int. J. Adv. Res. Comput. Eng. Technol. 2(4) (2013) 17. Food quality and safety systems – a training manual on food hygiene and the hazard analysis and critical control point (HACCP) system, 167 p. Food and agriculture organization of the united nations, Rome (1998) 18. NSF International: 2000 Nov. 10. Non-potentially hazardous foods, 12 p. NSF International. Report nr ANSI/NSF 75-2000, Ann Arbor (MI) 19. Powitz, R.W.: Water Activity: A New Food Safety Tool. Food Safety Magazine (2007). https:// www.food-safety.com/articles/3960-water-activity-a-new-food-safety-tool 20. ISO 18787:2017: Food Stuffs—Determination of Water Activity, 9 p. ISO, Geneva, Switzerland (2017) 21. Neil, H.: Mermelstein measuring moisture content & water activity. Food Technol. Magazine 63(11) (2009)
Reengineering of the Ukrainian Energy System: Geospatial Analysis of Solar and Wind Potential Iryna Doronina1 , Maryna Nehrey2(B) , and Viktor Putrenko3 1 Kyiv National Economic University named after Vadim Hetman, Kyiv 03057, Ukraine 2 National University of Life and Environment Science of Ukraine, Kyiv 03041, Ukraine
[email protected] 3 American University Kyiv, Kyiv 04070, Ukraine
Abstract. The potential for using renewable energy sources is a priority in the development of decentralized energy and ensuring the country’s energy security. The aim of the study is to calculate the technical potential of using the territory for renewable energy (solar and wind). The potential industrial area was calculated and the technical potential for the introduction of solar and wind energy in Ukraine was calculated. The technical potential of the sun is determined at the maximum level of 6.84 GW and the optimal level of production of 45.43 GW, and correspondingly of the wind −28.76 GW and 190.99 GW. Geospatial analysis has identified favorable areas for renewable energy infrastructure development and estimates of technical potential can be used to inform policy and investment decisions. The issue of the combination of food security and energy security in the process of the formation of the security model of Ukraine is examined. The study is an important contribution to the promotion of sustainable economic growth and the reduction of carbon dioxide emissions in Ukraine. Keywords: Renewable energy · Security · Energy transformation · Geospatial analysis · ArcGis · Potential · Sustainability
1 Introduction Energy is the core of the economy, a crucial component of the efficient development of industrial production, the functioning of housing and municipal services, and the organization of human activity. The energy transition, based on the principles of sustainable development, involves the gradual replacement of fossil energy resources with renewable ones. The concept of the “Brown” economy is being replaced by new ones, such as the “Green Economy” and “Circular Economy”, which are based on the assessment of the area and the availability of resources in this area. Ukraine’s energy system is centralized and unprepared for the challenges caused by Russia’s military aggression. Damage to the energy infrastructure and transmission lines has led to massive power outages in all regions of the country and blackouts. Therefore, measures are being taken at the regional level to decentralize the energy sector with a focus on using local renewable resources. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 404–415, 2023. https://doi.org/10.1007/978-3-031-36115-9_37
Reengineering of the Ukrainian Energy System
405
The new technologies’ application and opportunities for renewable resources on the ground require a general understanding of the country’s renewable energy potential. Another issue is the balance of using the territory’s potential. It should be pointed out that Ukraine consolidates about 30% of the world’s black soil reserves. The focus only on technical capacity is the main limitation of this study. To get a complete picture, an assessment of local feasibility of renewable energy projects should be added. Further research will be needed to assess the economic and social impacts of farmland replacement, and the feasibility and perception of such projects at local levels. We aim to provide a scientific basis for developing environmental policies and implementing energy decentralization in Ukraine. The paper demonstrates the significant potential of renewable energy sources in Ukraine, in particular solar and wind energy, and their potential contribution to sustainable economic growth and decarbonization of the economy. It also identifies the most favourable areas for the implementation of renewable energy infrastructure, quantifies the technical potential for solar and wind energy in Ukraine, and proposes a sustainable energy planning tool that can be replicated in other regions.
2 Materials and Methods An analysis of government regulations and documents of international organizations for the period from 2014 to 2022 showed that there is a significant difference in the approaches and results of assessing the potential of renewable resources in Ukraine. Therefore, different methodologies, assumptions, and timeframes are used for research. Government documents and the 2015 report of the International Renewable Energy Agency (IRENA) [1] estimate the technically achievable renewable energy potential based on scientific and economic research and legally defined targets. The indicators include targets for the replacement of brown generation, following the Paris Climate Agreement, and considering domestic demand in Ukraine. The REmap 2030 assessment methodology [2] analyzes renewable energy development and substitution. It includes cost analysis and technical efficiency (standard plant capacity, utilization rate and conversion efficiency) of renewable energy technologies and traditional (fossil fuel, nuclear power) for each sector: industry, construction, transportation, electricity, and heat. This methodology identifies possible technically and potentially achievable targets for renewable energy facilities. Ukraine’s total onshore wind potential is estimated at 24 GW. The solar energy potential is estimated at 4 GW. The state program documents set out an economically and logically sound approach to determining the technical potential of renewable energy, which was calculated as an indicator of oil equivalent replacement per kWh. The National Renewable Energy Action Plan for the period up to 2020 [3] defines the annual technically achievable energy potential of renewable energy sources, which reaches 68.6 million tons of oil equivalent per year. The reasonable installed capacity of solar energy in Ukraine is 4 GW. At the same time, the potential economically feasible installed capacity of wind energy in Ukraine reaches 15 GW. The Action Plan considered the direction of energy security and the independence of Ukraine from imported energy resources.
406
I. Doronina et al.
As the green business and green investment environment has developed over the past 10 years, based on the highest feed-in tariff in Europe until 2030 [4], private companies have called on Ukrainian government agencies to revise the renewable energy potential (upward). Subsequently, this indicator was used as an argument to increase the investment attractiveness of private businesses in Ukraine. Also, simplified procedures for environmental impact assessment [5] and procedures for obtaining permits at the regional and local without special approval from the public, were introduced. All of this has led to rapid growth and uneven distribution of installed renewable energy capacities in Ukraine and created an imbalance in the Integrated Power System of Ukraine. As of December 31, 2021, the installed capacity of renewable energy facilities with a green tariff (Feedin-tariff) is 8434.12 MW. The idea that a relatively small share of renewables cannot significantly affect the operation and balancing of the energy system is incorrect [6]. Energy policy in the 2017–2018 analysis showed that the state faced the need to introduce strict regulation of renewable energy to promote sustainable development and energy security in the country. During this period, to increase financial opportunities, private investors focused on the possible export of surplus renewable energy and the formation of Ukraine’s image as an “energy exporter to the EU”. However, it should be emphasized that Ukraine, both then and now, lacks the technical capabilities for full-scale energy exports to the EU. Table 1. Comparative analysis of the Ukrainian RES potential assessment Document
Author
Year
Wind onshore, GW
Solar plants, GW
REMAP 2030
IRENA
2015
24
4
Prospects for the development of renewable energy in Ukraine until 2030
State Agency on Energy Efficiency and Energy Saving
2015
16
4
About the National Renewable Energy Action Plan for the period up to 2020
The Ministry of Energy
2015
15
4
Atlas Renewable Energy 2010
IEE of the National Academy of Sciences of Ukraine
2015
24
4
Cost-competitive renewable power generation
IRENA
2017
320,58
70,61
Atlas Renewable Energy 2020
IEE of the National Academy of Sciences of Ukraine
2020
438
82,76
Reengineering of the Ukrainian Energy System
407
The IRENA report [7] estimates Ukraine’s renewable energy potential at 414.78 GW. However, the highest figures for Ukraine’s renewable energy potential are presented in the Atlas of Renewable Energy Potential of Ukraine [8], developed by The Institute of Renewable Energy of the National Academy of Sciences of Ukraine (Table 1). In January 2022, the Cabinet of Ministers of Ukraine launched a public discussion of an action plan for implementing and developing renewable energy. The National Renewable Energy Action Plan for the period up to 2030 [9] outlines the following opportunities – increase in electricity production through the use of more powerful wind turbines and commissioning of new capacities of onshore wind power plants up to 15.3 TWh in 2030 (total capacity of 4.7 GW); – increase in electricity production from solar energy to 10.5 TWh in 2030 (with a total capacity of about 10 GW, including 7 GW of industrial producers and 3 GW of consumers). Ukraine is viewed internationally as an importer of electricity and technology, caused by the large-scale destruction of energy infrastructure due to Russian military aggression. Therefore, the task of decentralized energy and renewable sources primarily for the country’s domestic needs is urgent today. According to the energy balance for 2020 (calculated by the rules of Directive 2009/28/EC), Ukraine’s total final energy consumption amounted to 50.5 million tons of oil equivalent (of which only about 4 million tons from renewable sources). Thus, if Ukraine maintains its level of energy consumption in 2020, it will be able to fully meet its energy needs with energy from renewable sources going forward. Scenarios for renewable energy development and the corresponding benefits of greening development and reducing CO2 emissions are actively discussed in academic circles [10–14]. Scientists are studying both the impact of renewable energy on the economy [15–17] and the promotion of environmental sustainability [18–23]. At the same time, the development of renewable energy from a technical point of view and the possibility of introducing modern technologies in the energy sector is crucial [24–27]. Our study is based on the methodology for determining the geographic and technical potential of Renewable Energy IRENA 2014, 2017 [7, 28], which requires several processing steps and assumptions. Shifting from the theoretical capacity (e.g., the total amount of solar radiation reaching a country or the total theoretical wind energy in a given area) to the technical or implementation potential requires a series of additional data assumptions (Fig. 1). The study aims to determine the geographical and technical potential of renewable energy in Ukraine. Advances in geographic information system programming can help achieve this goal [29]. GIS is a powerful tool with a variety of applications in analyzing and presenting spatial data (Table 2). The use of ArcGIS software allowed us to model maps that can be used to identify ideal project sites in the future and initiate a dialogue with regional and local organizations and communities.
408
I. Doronina et al.
“Global Solar Dataset”,“World Bank Open Data» and “Global Wind Atlas” maps showing the theoretical potential of different renewable energy sources (sun, wind)
define “exclusion and inclusion zones” based on geographic data (in GIS format)
input conversion factors for solar and wind technologies to show the resulting technical potential
Fig. 1. Stages of technical potential assessment (author’s illustration)
Table 2. The dimensions considered for the analysis, dataset selection for the limitation Map km2
GHI Speed Grids Slope wind buffer % zone
Protected Budlings Water Forest Roads, Black area Railways soil
Solar +
-
-
>35% -
-/+
-
-
-
?
Wind -
+
-
>20% -
-
-
-
-
?
• “-” exclusion zones, • “+” inclusion zones
3 Results We found that 75% of the territory of Ukraine has a global horizontal irradiation GHI of more than 1169 kWh/m2 /year. This global solar radiation is sufficient for the widespread introduction of both photovoltaic (solar power plants) and heat and power (solar collectors) equipment. Map data (min-max range) GHI for Ukraine per year from 1095–1461 kWh/m2 /year is presented in Fig. 2.
Reengineering of the Ukrainian Energy System
409
Fig. 2. GHI Ukraine
Given that more than 90% of winds have a speed of 7–8 m/s, which is sufficient for high-performance wind generation. The strongest winds are observed along the coast of the Azov and Black Seas. The average wind speed throughout the territory is from 2 to 10 m/s (Fig. 3).
Fig. 3. Wind speed 100 m altitude
410
I. Doronina et al.
Using ArcGis software and data [30, 31] for analysis, we identified the territories of the Potential industrial sun area - 325,839 km2 and the Potential industrial wind area − 325,023 km2 (Figs. 4, 5).
Fig. 4. Suitable sun area
Fig. 5. Suitable wind area
According to the data of The World Bank [32] Ukraine has 1/3 of the world’s organic black soil stocks, around 70% (410 000 km2 ) of the territory of Ukraine, of these: 36 000 km2 are different recreational areas; 89 000 km2 are used for various agricultural purposes (grasses, fruits, etc.); 285 000 km2 (outside of cities and protected areas) are used for arable land (wheat and sunflower) (Fig. 6). Ukraine is also the world’s leading exporter (52% share) of sunflower oil, the world’s second-largest exporter of grain (wheat, oats, …) and the world’s third-largest exporter of corn.
Reengineering of the Ukrainian Energy System
411
Fig. 6. Arable\fertile lands
Until 2021, there was a complete moratorium on the use of agricultural land in Ukraine [33], not for its intended purpose. However, there were changes after 2021 [34] regarding the rules for the use of agricultural land. Therefore, it is possible to consider changing the purpose of privately owned agricultural land but only after proving its economic feasibility. State-owned land can be changed only if the CMU and state collegial bodies decide to do so, proving the public benefit of the project. As of 2022, according to Geocadastre statistics, letter No. 281/…69–22 dated 01.12.2022, privately owned agricultural land covers an area of 235,113 km2 , stateowned land - 40,940 km2 , and communal land −8,100 km2 . If we take into account the change in the rules for the use of agricultural land and do not take into account the decrease in Ukraine’s productive capacity, we can consider changing the intended use of privately owned agricultural land. We emphasize that only Ukrainian citizens can own agricultural land. After subtracting the territories that cannot be used for RES industrial facilities and the lands of the agricultural land fund (state, municipal, private), we obtained the Potential industrial area maximal extraction for solar 49,786 km2 and wind 48,970 km2 (Tables 3, 4). Potential industrial area optimal extraction (including privately owned land) for solar 276,799 km2 and wind 275,983 km2 (Tables 3, 4). To determine the technical potential, we rely on the IRENA 2017 methodology [7] for the allocation of industrial land for renewable energy, in the amount of 1% of the geographical potential and calculate the technical potential according to their methodology (Tables 3, 4).
412
I. Doronina et al. Table 3. Potential industrial area and technical potential solar
Type of area f
Optimal
Maximal
Territory, km2
325839
325839
Black soil municipal owners, km2
8100
8100
Black soil private owners, km2
235113
na
Black soil state owners, km2
40940
40940
Potential industrial area, km2
49786
276799
Technical potential, GW
8,2
45,4
Table 4. Potential industrial area and technical potential wind Type of area f
Optimal
Maximal
Territor, km2
325023
325023
Black soil municipal owners, km2
8100
8100
Black soil private owners, km2
235113
na
Black soil state owners, km2
40940
40940
Potential industrial area, km2
48970
275983
Technical potential, GW
33,8
190,4
Having considered the proportion of installed renewable energy capacity over the past 12 years, it was found that, under equal economic conditions, more solar projects were installed than wind projects. This implementation was primarily influenced by the difference in the cost of the technologies themselves and the speed of designing solar parks, which made it possible to start construction under high “green tariff” conditions. We emphasize that in Ukraine, along with the existence of a simplified system of environmental impact assessment, there is a rejection by society and private farmers of the construction of large wind generation, which may be the subject of further research and discussion.
4 Summary and Conclusion The paper provides a scientific justification for the need to develop an environmental policy and implement energy decentralization in Ukraine to reduce carbon emissions and promote sustainable economic growth. The study demonstrates the significant potential of renewable energy sources in Ukraine, particularly solar and wind energy, which can be harnessed through the development of grid and off-grid projects at the local level. The focus on environmental policy development and energy decentralization in Ukraine, together with the even distribution of solar and wind potential, provides an opportunity for the decarbonization of industry and the economy. The significant potential of renewable energy and the scale of its distribution will lead to the development
Reengineering of the Ukrainian Energy System
413
of on-grid and off-grid projects at the local level. The feasibility of using agricultural land for energy is decided at the local level, so the study on the economic and social replacement of agricultural land with energy land will continue. The use of geospatial analysis allowed us to determine the Potential industrial area and calculate the technical potential for the introduction of solar and wind energy in Ukraine. 1) The assessment of previously recommended potentials was tied either to domestic market objectives and oil substitution or to continental assessment methods without taking into account local issues. 2) The most favourable areas that can be used for the implementation of renewable energy infrastructure have been identified on a quantitative scale. 3) The technical potential of the sun at the maximum extraction level of 6,84 GW and the optimal extraction level of 45,43 GW and, respectively, wind - 28,76 GW and 190, 99 GW - is determined. The results of the study have several potential applications and extensions. Policy makers can use the technical potential estimates provided in the study to set goals and targets for renewable energy development in Ukraine. Investors can use the identification of favourable areas for the prioritisation of their investments in renewable energy infrastructure. Moreover, the study methodology can be replicated in other regions to identify favourable renewable areas, making it a valuable tool for sustainable energy planning. Furthermore, future research could provide insight into the feasibility and perception of such projects at the local level, and expand on the economic and social impacts of replacing agricultural land with energy.
References 1. International Renewable Energy Agency. Renewable Energy Prospects for Ukraine – IRENA. https://www.irena.org/publications/2015/Apr/Renewable-Energy-Prospects-for-Ukraine. 09 Jan 2023 2. International Renewable Energy Agency. REMAP - 2030. Prospects for renewable energy development in Ukraine until 2030. https://saee.gov.ua/sites/default/files/UKR%20IRENA% 20REMAP%20_%202015.pdf. 04 Jan 2023 3. Cabinet of Ministers of Ukraine, Order on the National Renewable Energy Action Plan for the period up to 2020, p. 902, Kyiv, October 1, 2014. https://zakon.rada.gov.ua/laws/ show/902-2014-%D1%80?find=1&text=%D0%BF%D0%BE%D1%82%D0%B5%D0% BD%D1%86%D1%96%D0%B0%D0%BB#w1_1. 04 Jan 2023 4. National commission for state regulation in the areas of energy and community services. Resolution On Approval of the Procedure for Establishing, Revising and Terminating the “Green” Tariff for Electricity for Business Entities, Electricity Consumers, including Energy Cooperatives, and Private Households Whose Generating Units Produce Electricity from Alternative Energy Sources No. 1817 of 30.08.2019. https://zakon.rada.gov.ua/laws/show/ v1817874-19#Text. 05 Jan 2023 5. Verkhovna Rada of Ukraine. Law of Ukraine on Environmental Impact Assessment, May 23, 2017, No. 2059-VIII. https://zakon.rada.gov.ua/laws/show/2059-19#Text. 11 Dec 2022 6. Doronina, I., Kryshtof, N., Moskalenko, S.: Green-coal paradox in Ukraine and in the world as a challenge for state regulation in the energy industry. Eur. J. Econ. Manage. 6(6), 39–51 (2020). https://doi.org/10.46340/eujem.2020.6.6.4
414
I. Doronina et al.
7. International Renewable Energy Agency. Cost-Competitive Renewable Power Generation: Potential Across South East Europe. International Renewable Energy Agency (IRENA), Abu Dhabi (2017) 8. Atlas of Renewable Energy Potential of Ukraine. Kyiv, Institute of Renewable Energy of the National Academy of Sciences of Ukraine (2020). https://www.ive.org.ua/wp-content/upl oads/atlas.pdf. 21 Dec 2022 9. The Cabinet of Ministers of Ukraine. The State Agency on Energy Efficiency and Energy Saving has developed a draft National Action Plan for the Development of Renewable Energy for the period up to 2030. https://www.kmu.gov.ua/news/derzhenergoefektivnosti-rozrob leno-proekt-nacionalnogo-planu-dij-z-rozvitku-vidnovlyuvanoyi-energetiki-na-period-do2030-roku. 03 Jan 2023 10. Skrypnyk, A., Klymenko, N., Talavyria, M., Goray, A., Namiasenko, Y.: Bioenergetic potential assessment of the agricultural sector of the Ukrainian economy. Int. J. Energy Sect. Manage. 14(2), 468–481 (2019). https://doi.org/10.1108/IJESM-04-2019-0015 11. Matviychuk, A., Novoseletskyy, O., Vashchaiev, S., Velykoivanenko, H., Zubenko, I.: Fractal analysis of the economic sustainability of industrial enterprise. CEUR Workshop Proceedings 2422, 455–466 (2019) 12. Abebe, Y.M., Rao, P.M., Nak, M.G.: Load flow analysis of a power system network in the presence of uncertainty using complex affine arithmetic. Int. J. Eng. Manufact. 7(5), 48–64 (2017). https://doi.org/10.5815/ijem.2017.05.05 13. Davydenko, N., Buriak, A., Titenko, Z.: Financial support for the development of innovation activities. Intellectual Economics 13(2), 144–151 (2019). https://doi.org/10.13165/IE-19-132-06 14. Oliskevych, M., Beregova, G., Tokarchuk, V.: Fuel consumption in Ukraine: evidence from vector error correction model. Int. J. Energy Econ. Policy 8(5), 58–63 (2018) 15. Skrypnyk, A., Nehrey, M.: The formation of the deposit portfolio in macroeconomic instability. CEUR Workshop Proceedings 1356, 225–235 (2015) 16. Tkachenko, R., Kutucu, H., Izonin, I., Doroshenko, A., Tsymbal, Y.: Non-iterative neural-like predictor for solar energy in Libya. CEUR Workshop Proceedings 2105, 35–45 (2018). http:// ceur-ws.org/ 17. Kobets, V., Yatsenko, V.: Influence of the fourth industrial revolution on divergence and convergence of economic inequality for various countries. Neuro-Fuzzy Modeling Techniques in Economics 8, 124–146 (2019). https://doi.org/10.33111/nfmte.2019.124 18. Dimitrov, I., Davydenko, N., Lotko, A., Dimitrova, A.: Comparative study of main determinants of entrepreneurship intentions of business students. In: 2019 International Conference on Creative Business for Smart and Sustainable Growth (CREBUS) 2019 Mar 18, pp. 1–4. IEEE (2019). https://doi.org/10.1109/CREBUS.2019.8840050 19. Miroshnychenko, I., Kravchenko, T., Drobyna, Y.: Forecasting electricity generation from renewable sources in developing countries (on the example of Ukraine). Neuro-Fuzzy Modeling Techniques in Economics 10, 164–198 (2021). https://doi.org/10.33111/nfmte.202 1.164 20. Klymenko, N., Nehrey, M.: Electricity tariff structures modeling for reengineering Ukrainian energy sector. Lecture Notes on Data Engineering and Communications Technologies 135, 493–502 (2022). https://doi.org/10.1007/978-3-031-04809-8_45 21. Prokopenko, O., Chechel, A., Sotnyk, I., Omelyanenko, V., Kurbatova, T., Nych, T.: Improving state support schemes for the sustainable development of renewable energy in Ukraine. Polityka Energetyczna 24(1), 85–100 (2021) 22. Sabishchenko, O., R˛ebilas, R., Sczygiol, N., Urba´nski, M.: Ukraine energy sector management using hybrid renewable energy systems. Energies 13(7), 1776 (2020). https://doi.org/10.3390/ en13071776
Reengineering of the Ukrainian Energy System
415
23. Ostapenko, O., Olczak, P., Koval, V., Hren, L., Matuszewska, D., Postupna, O.: Application of geoinformation systems for assessment of effective integration of renewable energy technologies in the energy sector of Ukraine. Appl. Sci. 12(2), 592 (2022). https://doi.org/10. 3390/app12020592 24. Praiselin, W.J., Edward, J.B.: A review on impacts of power quality, control and optimization strategies of integration of renewable energy based microgrid operation. Int. J. Intell. Syst. Appl. (IJISA) 10(3), 67–81 (2018). https://doi.org/10.5815/ijisa.2018.03.08 25. Aliyu, S.O., Okwori, M., Onwuka, E.N.: A prototype automatic solar panel controller (ASPC) with night-time hibernation. Int. J. Intell. Syst. Appl. 8, 18–25 (2016). https://doi.org/10.5815/ ijisa.2016.08.03 26. Shakhovska, N., Montenegro, S., Kryvenchuk, Y., Zakharchuk, M.: The neurocontroller for satellite rotation. Int. J. Intell. Syst. Appl. 11(3), 1–10 (2019) 27. Mochurad, L., Kotsiumbas, O., Protsyk, I.: A model for weather forecasting based on parallel calculations. In: Advances in Artificial Systems for Medicine and Education VI 2023 Jan 21, pp. 35–46 (2023). Springer Nature Switzerland, Cham. https://doi.org/10.1007/978-3-03124468-1_4 28. Hermann, S., Miketa, A., Fichaux, N.: Estimating the renewable energy potential in Africa. IRENA-KTH working paper. International Renewable Energy Agency, Abu Dhabi (2014) 29. Joerin, F., Thériault, M., Musy, A.: Using GIS and outranking multicriteria analysis fo land use suitability assessment. Int. J. Geogr. Inf. Sci. 15, 153-174 (2001) 30. Global landcover 2000, Europe (2020). https://www.eea.europa.eu/data-and-maps/data/glo bal-land-cover-2000-europe. 15 Dec 2022 31. Terrestrial and Inland Waters Protected Areas https://www.protectedplanet.net/country/UKR. 13 Dec 2022 32. World Bank. Soil Fertility to Increase Climate Resilience in Ukraine (2014). https://www. worldbank.org/en/news/feature/2014/12/05/ukraine-soil. 28 Dec 2022 33. Law of Ukraine. On the moratorium on changing the designated purpose of certain recreational land plots in cities and other settlements. No. 3159-VI of March 17, 2011. https://zakon.rada. gov.ua/laws/show/3159-17#Text 34. Law of Ukraine. On Amendments to Certain Legislative Acts of Ukraine on the Terms of Turnover of Agricultural Land, No. 552-IX, March 31, 2020. https://zakon.rada.gov.ua/laws/ show/552-20#Text
Complex Approach for License Plate Recognition Effectiveness Enhancement Based on Machine Learning Models Yakovlev Anton(B) and Lisovychenko Oleh Igor Sikorsky Kyiv Polytechnic Institute, Peremohy Prospect 37, Kyiv 03056, Ukraine [email protected]
Abstract. OCR is an effective method when it comes to license plate recognition. However, the variability of the input data quality could reduce detection precision and may require extensive computation costs. Modern machine learning approaches for pattern recognition in the field of license plate recognition is an effective method to improve the traditional OCR only approaches. But effectiveness of machine learning models in terms of precision depends on train techniques from one side and quality, amount, variability of input train data from another. This paper addresses proposed methodology and complex approach for license plate recognition system development involving all components of the process like hardware, machine learning and a software implementation with field approbation. Approach is based on the Yolo v5 detection system which uses machine learning and includes research, synthesis and practical software and hardware implementation for all process constituents. Methodology being proposed allows higher detection precision comparingly to classical OCR-only approaches. Statistical methods are used to prove validity of license plate recognition results obtained. A set of 20 000 input image sets used in the experiment leads to 95,57% detection precision at average 1,168 s per image proving effectiveness of the method being proposed. Keywords: Decision support system · machine learning · artificial intelligence · pattern recognition · Yolo v5
1 Introduction The work is devoted to improvement of approaches to build detection systems (DS) and vehicle license plate recognition (LPR) using machine learning models. Complexity of the approach for mentioned task solving lies in full hardware-software cycle implementation. This includes all sequences from hardware setup in accordance with the standard [6] till DS learning methodology with a custom annotated images dataset usage. Solving similar tasks using machine learning systems and artificial intelligence methods is included into the scope of definitions and field of decision support systems (DSS) study. Yolo v5 is a machine learning models based image pattern recognition system. It provides a learning interface with user defined custom datasets and characterized with high © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 416–425, 2023. https://doi.org/10.1007/978-3-031-36115-9_38
Complex Approach for License Plate Recognition Effectiveness
417
performance [7]. Artificial intelligence methods in pattern recognition are widely used in various fields, like road safety [8], health care, production, signal processing, forecast systems, big data processing, etc. Problems of nonlinear machine learning methods usage are widely researched in scientific studies nowadays [5, 14–26]. These studies are focusing on methods of increasing accuracy and false detections reduction, increasing the time characteristics of the detection process [5]. LPR with machine learning methods usage is implemented in some modern commercial systems [32]. However, points like accuracy indicators, input data variability, computational ability growth, capturing devices resolution growth and new highly efficient machine learning models development provokes scientific interest in the raised issue. The problem of approaches with OCR-only method usage still persist as these methods are limited and outdated [24], which could be eliminated or altered using modern artificial intelligence based methods [35]. The purpose of the work is to develop a complex approach and build vehicle license plate DS that will increase effectiveness of its functioning by precision, speed and commercial implementation accessibility indicators.
2 Problem Statement To build the system it is mandatory to determine its components and justify effectiveness and applicability of its usage. Vehicle license plate DS consists of hardware (camera, computers, communication channels) and software (image processing software, training interfaces, detection interfaces, system for storage, reporting and visualization, etc.). Basic requirements for the hardware part of the system are defined in standard [6], which will be used during build [12]. The rest of the work is targeted to graphical pattern recognition detection system train methodics, train process application and results evaluation with confirmation or rejections of theses being proposed. It is necessary to: • To define and to engage hardware components in accordance with the requirements; • Setup hardware environment to form real world variable input images dataset for recognition model train and evaluation; • Create software for automated dataset annotation and determine optimality of parameters for Yolo v5 model train with given input dataset; • To process and analyze system functioning results in scope of precision, speed, applicability for commercial implementation.
3 Problem Solving Methodology of detection system development was created for stated problem solving. It covers all stages of the detection process from input image obtaining to string represented data view. Complex approach to LPR system development with artificial intelligence based DSS was created within the proposed methodology (Fig. 1).
418
Y. Anton and L. Oleh
Fig. 1. Complex approach to increasing the efficiency of detection systems using machine learning models structural diagram
3.1 Hardware Component Hardware component role in stated recognition problem solving comes down to two main functions. Firstly is to provide technical conditions to obtain images in accordance with the standard, which regulates requirements for road safety systems with a function of photo and video capturing [6, 12]. Secondly, ensuring the receipt of the original graphic image that will meet the quality characteristics for further processing (size, resolution, etc.). During this stage existing systems were analyzed as well as available on the market solutions which conformed minimal requirements for stated problem solving. Hardware components interaction features were also defined on all the “image capture - processing - storage” chain and implemented into the hardware system. For stated task accomplishment Axis P1344-E surveillance camera was chosen as a capturing and primary image processing device[9]. 3.2 Analysis of DSS for Pattern Recognition Problem Solving The system for responding to road events when working with road infrastructure must meet the requirements for real-time systems, since it can respond to events in the environment external to the system within the necessary (given) time limits [13]. That is why, when selecting the DSS for solving pattern recognition problems, one of the requirements is the ability to solve the task within the time limits t, which is defined by the t s standard [6] or faster. The temporal representation of the requirements for the process looks like this t ≤ ts , Also, the input data processing time t i should not change or increase significantly when the number of recognition objects within the image being processed increases ti ≤ ti n. The Yolo v5 pattern recognition system based
Complex Approach for License Plate Recognition Effectiveness
419
on machine learning models was chosen based on the specified criteria [7].To solve the task of recognizing specific patterns, this DSS should be trained accordingly and at a sufficient (according to [6] requirements) level of accuracy to cover the variability of the external environment, which is the source of input data. 3.3 An Approach to the Automation of a DSS Training Task The peculiarity of training the chosen pattern recognition DSS consists in the annotation of images with the definition of the boundaries of the detection region(s) in a normalized form. Available software tools for automating the process of annotation of detection regions were investigated, and it was found that some user actions remain unautomated and impose additional time costs [8]. When forming the input dataset for training, the process can involve thousands of images to achieve the required level of accuracy. The lack of automation in the processes that can be automated by modern software engineering tools entails human cost overruns and the impossibility of achieving the required level of accuracy in certain cases. The process of obtaining an annotated dataset Da for training was defined as Da = ni=1 f (i) m r=1 f (r), where i is the input image, r is the annotation region(s) in the image, and n and m are the number of images and regions in the image, respectively. As a result of the defined process, the software application YoloAnno was developed, which implements the proposed automation approaches [10]. Practical use of the application in creating a set of input annotated training data confirmed its effectiveness. 3.4 Creation of a Set of Training Input Data Using the Proposed Approach Using the built hardware system, 20,000 images were collected under different environmental conditions, from which 2,000 images were selected and annotated. For this, the proposed software application was used, which practically implemented all annotation tasks [10]. As a result, a set of graphic data and annotation data of detection regions was formed, ready for use in the Yolo v5 training system. Empirically, the optimal training parameters were established, which made it possible to perform training on the available equipment in the predicted time and obtain the resulting training data. The model of the pattern recognition system was obtained, which was validated to a given level of accuracy on random data from the built hardware graphic information retrieval system. 3.5 Software Validation in Real Conditions To check the effectiveness of the expressed hypotheses, an experiment was conducted using the created software. For the experiment, a set of 20,000 images, which were not involved in any way in the learning process, was collected using the built hardware complex from real road conditions. This set was processed by the built system and distributed according to the specified detection accuracy criteria. The distribution of results is presented in Table 1.
420
Y. Anton and L. Oleh Table 1. Result distribution for testing dataset input on license plate region detection
Type of dataset
Correct detection [%]
Semi-correct detection [%]
False detection [%]
All images
98,25
1,11
0,64
Plate present only
95,57
3,19
1,24
3.6 Interaction with External Detection Systems Using software engineering tools, the standardized API interface was developed for interaction with existing systems currently in operation. With the specified implementation, the created software received graphic data as an input and returned the detection result. Due to the non-linear dependence of the detection time when using machine learning algorithms, it was experimentally confirmed that the detection efficiency increased 5– 8% compared to the use of classical approaches where only OCR algorithms are used [30, 31]. A processing time cost factor t was considered in comparison to the “OCR-only” approach t OCR for an all image area and the proposed “Plate region detection” approach t RD combined with “Detected region OCR” approach t RO . The results of 1500 images processing within all t factor approaches are shown in Table 2. Table 2. Time cost for testing dataset input recognition with various approaches Approach
tmin [sec.]
tmax [sec.]
tavg [sec.]
t OCR
0,592
19,969
3,989
t RO
0,333
2,125
0,517
t RD
0,555
0,867
0,651
t RD + t RO
0,888
2.992
1,168
An Approach to Increasing the Efficiency of Detection Systems. The result of the theoretical justification of the proposed complex approach to increasing the efficiency of license plate recognition in combination with the practical implementation of the hardware and software complex confirmed the proposed theses. The investigated parameters of the detection process, such as license plate region detection accuracy and error rate, detection time, computational cost of the process, implementation cost were improved in the proposed approach and confirmed experimentally using statistical methods to process the obtained results. The obtained results were compared with the available performance data of the existing systems and the comparable efficiency was confirmed, which is presented in Figs. 2 and 3.
Complex Approach for License Plate Recognition Effectiveness
421
Fig. 2. Distribution comparison chart of the threshold for guaranteed correct detection of license plates region for existing systems (1) and using the proposed approach (2). Where higher value of correct detections is a positive trend. (See Table 1)
Fig. 3. OCR processing only to OCR combined with proposed method region detection processing results comparison. Where image processing time increasing is a negative trend (See Table 2)
4 Conclusion Traffic safety systems in operation use OCR approaches to solve the task of license plate recognition, which are inefficient nowadays. A complex approach and methodology for the construction of the DSS was developed when solving the problem of increasing the efficiency of license plate recognition. In the proposed approach, the complex
422
Y. Anton and L. Oleh
use of methods in the development of systems was applied. Under the approach work was focused on the hardware environment for data collection on one side and software application of detection systems based on machine learning models using software engineering tools on another. The consistency of each of the stages of solving the task made it possible to experimentally determine the components of the system and positively influence the final result of the system’s work in the specified efficiency criteria, such as license plate region detection accuracy, time, computational complexity, etc. The use of the Yolo V5 detection system and the proposed approach to data collection, training, evaluation and practical implementation of the results were proposed. The developed system in the conditions of working with real data showed an increase in efficiency according to the selected criteria in comparison with the requirements for similar systems and systems that are already in operation and using OCR based approaches. The proposed approaches and the built system can be applied both in the development of traffic safety systems with the function of recognizing license plates and in the study of methods of creating such systems with modern software engineering tools, training of detection systems in solving similar tasks and the scientific direction of pattern recognition by methods of machine learning in general.
References 1. ISO. 2021. ISO 39001:2012 - Road traffic safety (RTS) management systems — Requirements with guidance for use (2023). https://www.iso.org/standard/44958.html 2. RANKING EU PROGRESS ON ROAD SAFETY: 16th Road Safety Performance Index Report, June 2022 (2022). https://etsc.eu/wp-content/uploads/16-PIN-annual-report_ FINAL_WEB_1506_2.pdf 3. European regional status report on road safety 2019, ISBN 978 92 890 5498 0 (2019). https:// apps.who.int/iris/bitstream/handle/10665/336584/9789289054980-eng.pdf) 4. The European Automobile Manufacturers’ Association: NEW PASSENGER CAR REGISTRATIONS, EUROPEAN UNION. 18 January 2023 (2023). https://www.acea.auto/files/202 10916_PRPC_2107-08-FINAL.pdf) 5. Shan, Du., Ibrahim, M., Shehata, M., Badawy, W.: Automatic License Plate Recognition (ALPR): a state-of-the-art review. IEEE Trans. Circuits Syst. Video Technol. 23(2), 311–325 (2013). https://doi.org/10.1109/TCSVT.2012.2203741 6. DSTU (State standard of Ukraine) 8809:2018 Metrology. Traffic control devices with photo and video recording functions. Remote vehicle speed meters, remote space-time parameters of vehicle location meters. Metrological and technical requirements; (Ukr.) (2018) 7. Ultralytics/yolov5. GitHub (2020). https://github.com/ultralytics/yolov5 8. Yakovlev, A., Lisovychenko, O.: Pdhd do avtomatizac| anotuvann zobraen dl navqann modele xtuqnogo ntelektu. Adaptivn sistemi avtomatiqnogo upravlnn 1(36), 32–40 (2020). https://doi.org/10.20535/1560-8956.36.2020.209755 9. AXIS P1344-E Network Camera (2023). https://www.axis.com/en-gb/products/axis-p1344-e 10. Yakovlev, A., Yakovlev/Yoloanno, A.: GitHub (2020). https://github.com/AntonYakovlev/ Yoloanno 11. Massoud, M.A., Sabee, M., Gergais, M., Bakhit, R.: Automated new license plate recognition in Egypt. Alex. Eng. J. (2013). https://doi.org/10.1016/j.aej.2013.02.005 12. Anton, Y., Oleh, L.: Automated license plate recognition process enhancement with convolutional neural network based detection system to improve the accuracy and reliability of vehicle recognition. In: Hu, Z., Dychka, I., Petoukhov, S., He, M. (eds.) Advances in Computer
Complex Approach for License Plate Recognition Effectiveness
13. 14.
15.
16.
17.
18. 19.
20.
21.
22. 23.
24. 25.
26.
27.
28.
423
Science for Engineering and Education. ICCSEEA 2022. Lecture Notes on Data Engineering and Communications Technologies, vol. 134. Springer, Cham (2022). https://doi.org/10. 1007/978-3-031-04812-8_21 Decision support system - Wikipedia (2023). https://en.wikipedia.org/wiki/Decision_sup port_system Lin, C.-H., Lin, Y.-S., Liu, W.-C.: An efficient license plate recognition system using convolution neural networks. In: 2018 IEEE International Conference on Applied System Invention (ICASI). IEEE (2018) https://doi.org/10.1109/icasi.2018.8394573 Rhead, M., Gurney, R., Ramalingam, S., Cohen, N.: Accuracy of automatic number plate recognition (ANPR) and real world UK number plate problems. In: 2012 IEEE International Carnahan Conference on Security Technology (ICCST). IEEE (2012). https://doi.org/ 10.1109/ccst.2012.6393574 Koval, V., Turchenko, V., Kochan, V., Sachenko, A., Markowsky, G.: Smart license plate recognition system based on image processing using neural network. In: Proceedings of the Second IEEE International Workshop on Intelligent Data Acquisition and Advanced Computing Systems: Technology and Applications, 2003 (2003) https://doi.org/10.1109/ida acs.2003.1249531 Fahmy, M.M.M.: Automatic number-plate recognition: neural network approach. In: Proceedings of VNIS’94 - 1994 Vehicle Navigation and Information Systems Conference (1994). https://doi.org/10.1109/vnis.1994.396858 Shashirangana, J., et al.: License plate recognition using neural architecture search for edge devices. Int. J. Intell. Syst. 37, 10211–10248 (2021). https://doi.org/10.1002/int.22471 Shashirangana, J., Padmasiri, H., Meedeniya, D., Perera, C.: Automated license plate recognition: a survey on methods and techniques. IEEE Access. (2021). https://doi.org/10.1109/ access.2020.3047929 Shrivastava, S., Singh, S.K., Shrivastava, K., Sharma, V.: CNN based automated vehicle registration number plate recognition system. In: 2020 2nd International Conference on Advances in Computing, Communication Control and Networking (ICACCCN) (2021). https://doi.org/ 10.1109/icacccn51052.2020.9362737 Akther, M., Ahmed, M., Hasan, M.: Detection of vehicle’s number plate at nighttime using iterative threshold segmentation (ITS) algorithm. Int. J. Image, Graph. Sig. Process. 5(12), 62–70 (2013). https://doi.org/10.5815/ijigsp.2013.12.09 Kaur, S.: An automatic number plate recognition system under image processing. Int. J. Intell. Syst. Appl. 8(3), 14–25 (2016). https://doi.org/10.5815/ijisa.2016.03.02 Ramshankar, Y., Deivanathan, R.: Development of machine vision system for automatic inspection of vehicle identification number. Int. J. Eng. Manuf. 8(2), 21–32 (2018). https:// doi.org/10.5815/ijem.2018.02.03 Aung, M.M.: Study for license plate detection. Int. J. Image, Graph. Sig. Process. 11(12), 39–46 (2019). https://doi.org/10.5815/ijigsp.2019.12.05 Sakharkar, Y.A., Singh, M., Kumar, K.A., Aju, D.: A reinforcement learning based offload decision model (RL-OLD) for vehicle number plate detection. Int. J. Eng. Manuf. (IJEM) 11(6), 11–18 (2021). https://doi.org/10.5815/ijem.2021.06.02 Pirgazi, J., Sorkhi, A.G., Kallehbasti, M.M.P.: An efficient robust method for accurate and realtime vehicle plate recognition. J. Real-Time Image Proc. 18(5), 1759–1772 (2021). https:// doi.org/10.1007/s11554-021-01118-7 Calitz, A., Hill, M.: Automated license plate recognition using existing university infrastructure and different camera angles. The African Journal of Information Systems 12(2), 4 (2020). https://digitalcommons.kennesaw.edu/ajis/vol12/iss2/4 Ma, L., Zhang, Y.: Research on vehicle license plate recognition technology based on deep convolutional neural networks. Microprocess. Microsyst. 82, 103932 (2021). https://doi.org/ 10.1016/j.micpro.2021.103932
424
Y. Anton and L. Oleh
29. Silvano, G., et al.: Synthetic image generation for training deep learning-based automated license plate recognition systems on the Brazilian Mercosur standard. Des. Autom. Embed. Syst. 25(2), 113–133 (2020). https://doi.org/10.1007/s10617-020-09241-7 30. Schegolihin, Y., Mitrohin, M., Sazykina, V., Semenkin, M.: Gradual labeling of the training set to improve the efficiency of image detection by a neural network on the example of license plate recognition. In: 2021 29th Conference of Open Innovations Association (FRUCT) (2021). https://doi.org/10.23919/fruct52173.2021.9435602 31. “VIDEOKONTROL-Rubezh” complex. Ollie.com.ua (2020). http://www.ollie.com.ua/vid eocontrol/index.html. (Ukr.) 32. USIT, Produkciya - USI. Ukrsi.com.ua (2020). http://ukrsi.com.ua/products/. (Ukr.) 33. Yampolskyi, L.S., Lisovichenko, O.I., Oliynyk, V.V.: Neurotechnologies and neurocomputer systems: textbook, p. 576. Dorado-Druk, Kyiv (2016). (Ukr.) 34. Joshi, P., Escrivá, D., Godoy, V.: OpenCV by Example. Packt Publishing, Birmingham, UK (2016) 35. Omar, N., Sengur, A., Al-Ali, S.G.S.: Cascaded deep learning-based efficient approach for license plate detection and recognition. Expert Syst. Appl. 149, 113280 (2020). https://doi. org/10.1016/j.eswa.2020.113280 36. Sahoo, A.K.: Automatic recognition of Indian vehicles license plates using machine learning approaches. Materials Today: Proceedings 49, 2982–2988 (2022). https://doi.org/10.1016/j. matpr.2020.09.046 37. Jamtsho, Y., Riyamongkol, P., Waranusast, R.: Real-time license plate detection for nonhelmeted motorcyclist using YOLO. ICT Express 7(1), 104–109 (2021). https://doi.org/10. 1016/j.icte.2020.07.008 38. Qian, Y., et al.: Spot evasion attacks: Adversarial examples for license plate recognition systems with convolutional neural networks. Comput. Secur. 95, 101826 (2020). https://doi. org/10.1016/j.cose.2020.101826 39. Silva, S.M., Jung, C.R.: Real-time license plate detection and recognition using deep convolutional neural networks. J. Vis. Commun. Image Represent. 71, 102773 (2020). https://doi. org/10.1016/j.jvcir.2020.102773 40. Slimani, I., Zaarane, A., Okaishi, W.A., Atouf, I., Hamdoun, A.: An automated license plate detection and recognition system based on wavelet decomposition and CNN. Array 8, 100040 (2020). https://doi.org/10.1016/j.array.2020.100040 41. Xiang, H., Zhao, Y., Yuan, Y., Zhang, G., Hu, X.: Lightweight fully convolutional network for license plate detection. Optik 178, 1185–1194 (2019). https://doi.org/10.1016/j.ijleo.2018. 10.098 42. Björklund, T., Fiandrotti, A., Annarumma, M., Francini, G., Magli, E.: Robust license plate recognition using neural networks trained on synthetic images. Pattern Recogn. 93, 134–146 (2019). https://doi.org/10.1016/j.patcog.2019.04.007 43. Kessentini, Y., Besbes, M.D., Ammar, S., Chabbouh, A.: A two-stage deep neural network for multi-norm license plate detection and recognition. Expert Syst. Appl. 136, 159–170 (2019). https://doi.org/10.1016/j.eswa.2019.06.036 44. Asif, M.R., Qi, C., Wang, T., Fareed, M.S., Raza, S.A.: License plate detection for multinational vehicles: an illumination invariant approach in multi-lane environment. Comput. Electr. Eng. 78, 132–147 (2019). https://doi.org/10.1016/j.compeleceng.2019.07.012 45. Rao, W., Yao-Jan, Wu., Xia, J., Jishun, Ou., Kluger, R.: Origin-destination pattern estimation based on trajectory reconstruction using automatic license plate recognition data. Transp. Res. Part C: Emerg. Technol. 95, 29–46 (2018). https://doi.org/10.1016/j.trc.2018.07.002 46. Li, H., Wang, P., You, M., Shen, C.: Reading car license plates using deep neural networks. Image Vis. Comput. 72, 14–23 (2018). https://doi.org/10.1016/j.imavis.2018.02.002
Complex Approach for License Plate Recognition Effectiveness
425
47. Wang, R., Sang, N., Wang, R., Jiang, L.: Detection and tracking strategy for license plate detection in video. Optik 125(10), 2283–2288 (2014). https://doi.org/10.1016/j.ijleo.2013. 10.126 48. Öztürk, F., Özen, F.: A new license plate recognition system based on probabilistic neural networks. Procedia Technol. 1, 124–128 (2012). https://doi.org/10.1016/j.protcy.2012.02.024 49. Yuren, Du., Shi, W., Liu, C.: Research on an efficient method of license plate location. Phys. Procedia 24, 1990–1995 (2012). https://doi.org/10.1016/j.phpro.2012.02.292 50. Erdinc Kocer, H., Kursat Cevik, K.: Artificial neural networks based vehicle license plate recognition. Procedia Comput. Sci. 3, 1033–1037 (2011). https://doi.org/10.1016/j.procs. 2010.12.169
The Analysis and Visualization of CEE Stock Markets Reaction to Russia’s Invasion of Ukraine by Event Study Approach Andrii Kaminskyi1 and Maryna Nehrey2(B) 1 Taras Shevchenko National University of Kyiv, Kyiv 01033, Ukraine 2 National University of Life and Environment Science of Ukraine, Kyiv 03041, Ukraine
[email protected]
Abstract. The “Black swan” event which raised from Russia’s invasion of Ukraine in February 2022, has caused an extensive external shock for world financial markets. Our paper examines the scale of its impact on the three largest stock markets in Central and Eastern Europe: the Czech, Polish, and Hungarian. The choice of markets was due to the close proximity to Ukraine. To do this, we created a set of indicators that allowed us to display the specificity of reaction to shock. Indicators were split into three groups. The first group included indicators that reflect the «hustle» of investors and the reformatting of investment portfolios in the first ten days after the invasion event. The second group’s metrics consider shock in the context of the price-based dimension “deepness of fall – level of recovery”. The third group allows displaying changes in “risk-return” correspondence in the context of ESG score. The K-ratio tool was applied for considering consistency. The result indicated a strong impact on researched markets and different consequences for stock index constituents, especially for companies with high ESG scores. Keywords: Event study approach · stock market data; shock · ESG investing · Russia-Ukraine war · CEE countries · K-ratio
1 Introduction Russia invaded Ukraine at 24 February 2022, and hostilities began. This event had a tremendous geopolitical impact which was expressed in military, real and financial economic, social and many other aspects. This event, which can be characterized as a strong shock, was negatively affected by three directions. The first direction includes the socially specific economic consequences. The second direction is due to the impact on multiple sectors through global supply chain disruption. This is especially related to sharp fluctuations in oil, gas and number of different commodities prices. The third direction is highlight sanctions, which have a direct impact on the aggressor but at the same time have spillover effects. The reaction of stock markets was significant. Investors, under the influence of a shock, started actively buying and selling various assets, which leads to sharp price © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 426–436, 2023. https://doi.org/10.1007/978-3-031-36115-9_39
The Analysis and Visualization of CEE Stock Markets Reaction
427
fluctuations and high volatility of returns. Our study aims was to apply an event study approach to this “Black Swan” shock for Polish, Czech and Hungarian stock markets. These markets are the most capitalized and liquid in the CEE (Central Eastern European) group. Moreover, these countries are geographically close to Ukraine (Poland and Hungary share a border with Ukraine). Stocks of companies which are constituents of baskets of stock indices PX, WIG20, and BUX were object of investigation. The scale of such reaction and their peculiarities are in focus in our research. For our research goals, we have formed a set of indicators. They have allowed us to display the specificity of the shock reaction and to analyze the specificities of the different companies’ stock reactions. Indicators were combined into three groups. The first group included indicators that reflect the «hustle» of investors and the reformatting of investment portfolios in the first ten days after the invasion event. The second group’s indicators described dimension “deepness of fall – level of recovery”. The third group allows to display impact in the context of ESG (Environment, Social, Governance) score. The actuality of this connected with the process of quite actively promoting the implementation of ESG criteria in companies. The K-ratio tool was applied for investigation of stock returns consistency changes. The result indicated a strong impact shock to such consistency. Moreover, the consistency changing through shock pipeline demonstrated interconnection with ESG scores. Thus, most negative K-ratio changes were revealed to high ESG scores companies. The paper is structured as follows. Section 2 presents the literature review. Section 3 presents the methodology, which applied in our investigation. The Sect. 4 shows the results of applying the methodology to the CEE stock market. The special visualization of indicators values displays shock influences. Finally, Sect. 5 presents the conclusions and controversial issues.
2 Literature Review Complex and deep analysis of investigated shock and its development demonstrates significance it to modern economic and, particularly, to stock markets. During the year after the Russian invasion of Ukraine, a lot of literature on the economic consequences of this has been published. The complexity of this aggression-based impact is examined in [1]. In particular, the author highlights such after-effects: global supply chain disruption, growth of oil and gas prices, changes in the banking segment, global inflation acceleration and cost of living and some others. The basic statistical tool of the author‘s research is a two-stage least square regression analysis. About global stock markets, the author indicates his findings about the falling of several world stock indices within five days (from February 18 to February 25). The choice of such interval is interesting. Although, from our point of view, it assumes the presence of a certain anticipating, which is discussible. Our point of view is more to tend the “Black swan” case. The paper [2] provides an in-depth statistical analysis of the reaction of the G20 stock markets to the Russian-Ukrainian conflict. The paper analysis shows that abnormal returns and cumulative abnormal returns show that the European and Asian regions have been affected the most by the launch of the Russian aggression against Ukraine. Moreover, the author‘s estimation indicates Poland, Hungary, and Turkey plummeted
428
A. Kaminskyi and M. Nehrey
essentially more than others (disregarding the Russian stock market). Abnormal returns on event day were Poland (−10.374%), Hungary (−9.488%), and Turkey (−8.262%). An extensive analysis of the impact of the Russian-Ukrainian war on financial markets is presented in [3]. The authors analyzed the effect on markets of 73 countries and found significant results. It turned out that the war has created considerable instability in the markets, which is reflected in the variability of yields (up to 20%). At the same time, the authors estimate the impact on average stock prices to be relatively low. There are many empirical studies (including event study methods applying) is devoted to the analysis of Russian aggression-based shocks in various segments of the investment market. This applies to different types of investment assets, both traditional and alternative. Also, several studies already present the results of the behavior of financial instruments produced by companies from different sectors of the economy. In addition, a few studies describe the response of stock markets in individual countries. There are many empirical studies (including event study methods applying) devoted to the analysis of Russian aggression-based shocks in various segments of the investment market [4–7]. This applies to different types of investment assets, both traditional and alternative [8–14]. Also, a few studies already present the results of the behavior of financial instruments produced by companies from different sectors of the economy [15–17]. In addition, several studies describe the response of stock markets in individual countries [18–22]. Thus, considering that Russia is the significant exporter of oil and gas. Authors of [23] verified hypothesis about Russia aggression affecting the to the energy and metal markets. Authors found that war had a significant influence on gold, platinum, palladium and nickel, markets. Moreover, renewable energy industry also demonstrates a significant increasing in the anomalous returns for the with the renewable energy industry. Methodologically they use an event research technique and highest deviation was on the t + 1 day. We find a study [24] that examines the impact of Russian aggression on the yields and risks in the commodity market very fascinating. The authors estimate an rising in volatility from 35% to 85%. This, according to the authors, exceeds the level observed during the pandemic COVID-19. Authors grounded that crude oil turns into a net transmitter of side effects, while wheat and soybeans become net recipients of side effects. Such markets commodities as silver, gold, copper, platinum, aluminum and sugar become pure transmitters of volatility. Considering the impact of Russian aggression on alternative assets, it is necessary to mention the article [25], which justifies the impact on the pricing of cryptocurrencies. This is explained by an increase in geopolitical risk (expressed by an index) caused by aggression. The authors substantiate the statistical impact of the role of geopolitical risk on cryptocurrency markets. The impact on the aerospace and defense industries was quite strong, as substantiated in [26]. The authors examined new sentiments from war-related news from October 2021 to June 2022. The results showed a negative impact of war sentiment on market behavior. The authors’ conclusion indicates a negative effect of the war on the air transportation market and a positive effect on the defense market.
The Analysis and Visualization of CEE Stock Markets Reaction
429
The paper [27] also considers the event study method for application to 24 Feb 2022 shock also. Author examines 209 stocks from Borsa Istanbul (BIST) in terms of pre-and post-historical period. The abnormal returns were found in the second, third, fourth, and fifth days they were negative and statistically significant.
3 Methodology 3.1 Research Data We used the event study method to analyze the data and to calculate the indicators we are considering. The event is on 24.02.2022. On this date, Russia started a military aggression against Ukraine called a “special military operation”. To apply the event study method, we considered a time interval of one year: 24.08.2021 - 24.08.2022. The “before event” interval was 24.08.2021–23.02.2022. We considered the event as a short time interval of 24.02.2024 - 09.03.2024. It was defined as the ten working days that our study took as “subject for methods applying” The “after event” interval was 10.03.2022–24.08.2022. Our sample includes shares of companies included in the stock indices of the three CEE countries. They are the index of the Czech stock market PX (9 companies), the index of the Polish stock market WIG20 (20 companies), and the index of the Hungary stock market BUX (16 companies). A total of 45 companies were used for the analysis. The following data from investing.com were used about the companies in question: 1) Daily (closed) prices. 2) Daily high and low prices. 3) Daily trading volume. The daily values of these three indices were also used. The ESG scores were a separate data block used for the analysis. We used ESG scores from the company S&P Global [28]. Score values involved four numbers: ESG score, E score, S score, and G score. At the time of the survey, these values were calculated for 25 companies out of 46. Therefore, the ESG survey was implemented for limited samples. 3.2 Shock Estimation by “Price Spread – Volume Volatility” Indicators We used the hypothesis that in shock environment (the “Black Swan”) investors are suddenly faced with uncertainty and this affects their behavior. Specifically, they reformat their portfolios actively by selling and buying stocks. Based on this logic, we have introduced the following two indicators based on daily trading. The first indicator is “shock deepness volume” (SDV) which is defined as: SDV =
Average daily volume over shock(10 days) −1 Average daily volume before shock(time intervalbefore event)
(1)
The second indicator is “recovery rate volume” (RRV) which is defined as: RRV =
Average daily volume after shock(time interval "after event") Average daily volume before shock (time interval "before event")
(2)
430
A. Kaminskyi and M. Nehrey
These are dimensionless indicators. SDV shows the percentage increase in daily trading volume at the time of a shock. Since a shock is characterized by the abruptness of change, the higher the SDV, the more pronounced the shock. RRV ratio shows to what extent the average daily trading volume after the shock exceeds (or is already equal to) that before the shock. In other words, if it is well above 1, the recovery has not yet taken place. This indicator essentially reflects the long memory of shock affecting. Of course, it depends on the selected interval. Thus, our first shock indicator is a pair (SDV; RRV). The following two indicators reflect investors’ hustle (or in slang “rush asunder”). In case of a shock, there is a rapid increase in uncertainty regarding present momentum and future returns. Therefore, there is no classical equilibrium price. The change in price during the day can be very significant. This occurrence we formalize in our next indicator: HLDD =
Highprice − Lowprice (Day) 0, 5 ∗ (Highprice + Lowprice)
(3)
This indicator integrates two effects. The first, reflected in the numerator, shows the investor‘s hustle. Large fluctuations during the day increase the values. The second effect, reflected in the denominator, can show the decrease in the prices and influence (increase) its values. Thus, according to our approach, the higher the HLDD, the stronger the shock manifests itself. The HLDD is calculated for each day. The logic of SD and RR construction can be applied to its values. As a result, we obtain two similar indicators SDHL and RRHL. 3.3 Shock Estimation by Price Changing The following hypothesis in our study was the hypothesis of a sharp decline in share price followed by a recovery. In our research, the fall-recovery pair characterizes the “risk-return” correspondence. For this purpose, we used indicators methodologically consistent with SDV and RRV. Only instead of trading volume it uses price. The first indicator is “shock deepness price”: SDP =
Average proce through 10 days of shock − 1. Average price during time interval "before event")
(4)
The second indicator is “recovery rate price” (RRP) which is defined as: RRP =
Average price after shock (during time interval "after event") Average price before shock (during time interval "before event")
(5)
SDP has the nature of a classical rate of return with some specifications linked to average prices. It was supposed that such an approach nihilate price volatility before the shock to before the shocking price. The logic of using such a form of RRP is to desire an estimate comparison with the before shock period, not with the “bottom price”.
The Analysis and Visualization of CEE Stock Markets Reaction
431
3.4 ESG-based Analysis One of the most dynamic areas in the financial investment market is ESG investing. Today, approximately 40% of assets under management meet ESG criteria. The level of ESG itself is determined by ESG scores, which have been implemented in the market by major analytical agencies - S&P Global, Refiniti, Sustainalytics and others. The main question is how to combine three criteria - risk minimization, return maximization and ESG maximization. The companies from CEE markets step-by-step move towards receiving ESG scores. Our research question in this regard was to compare how companies with different ESG scores passed the shock.
4 Results 4.1 Markets’ Overall Visualization of Shock: Indexes Handling Markets received a significant shock as a response to Russian invasion in Ukraine. This shock was visualized in the following stock indices of CEE countries: PX, WIG, and BUX (Fig. 1). The shock visualization shows the integral impact of the Russian invasion on the CEE stock market. Figures on the left show the volatility of returns. Figures on the right show the 30-day rolling standard deviation. Data visualization analysis allows us to conclude the significant event’s impact on the stock market. 4.2 Shock Estimation by “PRICE Spread – Volume Volatility” Indicators We estimated shock by “price spread – volume volatility” indicators. For all except one company (Photon), the SDHL figure increased (Fig. 2). The average increase is 111% (i.e., more than double the pre-shock value). At the same time, the recovery over the period under review showed: 27% of stocks began to have an HLDD value less than the pre-shock value, and 73%. We can conclude that the shock and further military developments are reflected in high investor fussiness in the face of uncertainty. A similar examination of the indicators related to daily trading volume shows a slightly different picture. Thus, five companies showed lower volume during the shock (Fig. 3). The average increase is 132%. Moreover, the number of companies that have “returned” to pre-shock level is also different - 44%. 4.3 Shock Estimation by Price Changing A price trend analysis shows that the average value of the fall (in fact, the yield expressed by SDP) was -12.46% (Fig. 4). The decline is not evenly distributed across the markets. In particular, Poland -14.84% Czech -4.52% Hungary -13.95%. The average RRP was 88.08%. This suggests that the markets have not yet recovered from the shock of six months.
432
A. Kaminskyi and M. Nehrey
Fig. 1. Stock market indexes visualization
4.4 Investigation of Shock Transition by Stocks with Different ESG Scores The K-ratio was used to project the consistency analysis. It was calculated for the entire period before and after the event. The event impact was estimated as the difference between the K-ratio before and after the event. In Fig. 5, the columns show that in most cases the K-ratio decreased. We further compared changes in K-ratio and ESG scoring. It turned out that they had opposite directions. The K-ratio changed the most for companies with high ESG scores.
5 Summary and Conclusion The Russian invasion of Ukraine had a great impact on the global economy. One of these strong impacts concerns stock markets. In our research, we investigated the response to this event of the three largest CEE stock markets. For research, we applied three approaches. Each approach was considered as some modification of event study logic. The first approach involved analyzing the behavior of investors in the moment of shock. We studied event at 10 days after it appearance. Two indicators related to trading volume
The Analysis and Visualization of CEE Stock Markets Reaction
433
400% 350% 300%
SDHL
250% 200% 150% 100% 50% 0% 0%
50%
100%
-50%
150%
200%
250%
300%
RRHL Fig. 2. Shock visualization in SDHL-RRHL scale
600% 500%
SDV
400% 300% 200% 100% 0% 0% -100%
50%
100%
150%
200%
RRV Fig. 3. Shock visualization in the SDV-RRV scale
and price spread showed huge deviations. This is due to the great uncertainty that caused the shock. Uncertainties have led to the impossibility of precise future pricing and the reformatting of investment portfolios. Within six months of the shock, processes were normalized, but the stock parameters of most companies did not reach before shock levels. There were also applied indicators directly based on price movements showing the depth of the fall and the level of price recovery. It shows that the average value of the fall (in fact, the return expressed by SDP) was -12.46% (Fig. 4). The decline is not evenly
434
A. Kaminskyi and M. Nehrey
160.00%
RRP
PENP
LPPP ALEP OTPB
ALTS CEZP
PGE MOLB TABK PANPMONET ANYB KOFOL DNP KTY PKN IGNY WABEMAST ACPP VIGRGSPA GDRB MTEL PCOP CIGP OPL AUTW PZU APPB BKOM OPUSG KRU PEO SPL1 ERSTAKKO PKO CPS CDR MBK
JSW
140.00% 120.00%
CZG
y = 1,1633x + 1,0257 100.00% R² = 0,6472
KGH
CCC
80.00% 60.00% 40.00% 20.00%
0.00% -60.00% -50.00% -40.00% -30.00% -20.00% -10.00% 0.00% 10.00% 20.00% 30.00% 40.00%
SDP Fig. 4. Price trend analysis
ESG SCORE
100
0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1
80 60 40 20 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 Companies Fig. 5. Correspondence between ESG and changes in K-ratio value
distributed across the markets. In particular, Poland -14.84% Czech -4.52% Hungary -13.95%. These figures demonstrate comparable value with results presented in [2]. The average RRP was 88.08%. This suggests that the markets have not yet recovered from the shock of six months. K-ratio was used to analyze the effect of shock affecting on risk-return correspondence in the context of ESG score values. The change of this indicator during the transition was overwhelmingly negative. An interesting finding was that the K-ratio change is inversely related to the ESG score. The largest declines in K-ratio were seen in stocks with a large ESG score.
The Analysis and Visualization of CEE Stock Markets Reaction
435
References 1. Ozili, P.K.: Global Economic Consequence of Russian Invasion of Ukraine (2022). SSRN. https://doi.org/10.2139/ssrn.4064770 2. Yousaf, I., Patel, R., Yarovaya, L.: The reaction of G20 stock markets to the russia-ukraine conflict ‘black-swan’ event: evidence from event study approach. J. Behav. Experimental Fin. (2022). https://doi.org/10.2139/ssrn.4069555 3. Lo, G., Marcelin, I., Bassène, T., Sène, B.: The Russo-Ukrainian war and financial markets: the role of dependence on Russian commodities. Finance Research Lett. 50(103194) (2022). https://doi.org/10.1016/j.frl.2022.103194 4. Diacona¸su, D., Mehdian, S., Stoica, O.: The reaction of financial markets to Russia’s invasion of Ukraine: evidence from gold, oil, bitcoin, and major stock markets. Applied Econ. Lett. (2022). https://doi.org/10.1080/13504851.2022.2107608 5. Fang, Y., Shao, Z.: The Russia-Ukraine conflict and volatility risk of commodity markets. Finance Res. Lett. 50 (103264) 6. Gaio, L., Stefanelli, N., Pimenta, T., Bonacim, C., Gatsios, C.: The impact of the RussiaUkraine conflict on market efficiency: evidence for the developed stock market. Finance Research Lett. 50 (103302) (2022) 7. Joshipura, M., Lamba, A.: Asymmetric Impact of Russia-Ukraine War on Global Stock Markets (2022). https://ssrn.com/abstract=4273419 or http://dx.doi.org/https://doi.org/10.2139/ ssrn.4273419 8. Zulqarnain, M., Ghazali, R., Ghouse, M.G., Hassim, Y.M.M., Javid, I.: Predicting financial prices of stock market using recurrent convolutional neural networks. Int. J. Intell. Syst. Appl. (IJISA) 12(6), 21–32 (2020) 9. Guryanova, L., Bolotova, O., Gvozdytskyi, V., Olena, S.: Long-term financial sustainability: an evaluation methodology with threats considerations. Rivista di Studi sulla Sostenibilita 1, 47–69 (2020) 10. Izonin, I., Tkachenko, R., Vitynskyi, P., Zub, K., Tkachenko, P., Dronyuk, I.: Stacking-based GRNN-SGTM ensemble model for prediction tasks. In: 2020 International conference on decision aid sciences and application (DASA). IEEE, 2020, pp. 326-330. https://doi.org/10. 1109/DASA51403.2020.9317124 11. Dey, P.P., Nahar, N., Hossain, B.M.: Forecasting stock market trend using machine learning algorithms with technical indicators. Int. J. Information Technol. Comput. Sci. 12(3), 32–38 (2020) 12. Kaminskyi, A., Butylo, D., Nehrey, M.: Integrated approach for risk assessment of alternative investments. Int. J. Risk Assessment Manage. 24(2–4), 156–177 (2021). https://doi.org/10. 1504/IJRAM.2021.126413 13. Mahedy, A.S., Abdelsalam, A.A., Mohamed, R.H., El-Nahry, I.F.: Utilizing neural networks for stocks prices prediction in stocks markets. Int. J. Inf. Technol. Comput. Sci. 12(3), 1–7 (2020) 14. Sova, Y., Lukianenko, I.: Theoretical and empirical analysis of the relationship between monetary policy and stock market indices. In: 2020 10th International Conference on Advanced Computer Information Technologies, ACIT 2020 - Proceedings, pp. 708–711, 9208926 (2020) 15. Jain, V.R., Gupta, M., Singh, R.M.: Analysis and prediction of individual stock prices of financial sector companies in NIFTY50. Int. J. Information Eng. Electronic Bus. 12(2), 33 (2018) 16. Kaminskyi, A., Nehrey, M.: Changing risk-return correspondence during the COVID-19 turmoil: evidence from polish stock market. Research on Enterprise in Modern Economy theory and practice, 1(32), 18–33 (2021). https://doi.org/10.19253/reme.2021.01.002
436
A. Kaminskyi and M. Nehrey
17. Davydenko, N., Buriak, A., Titenko, Z.: Financial support for the development of innovation activities. Intellectual Economics 13(2), 144–151 (2019) 18. Lukianenko, D., Strelchenko, I.: Neuromodeling of features of crisis contagion on financial markets between countries with different levels of economic development. Neuro-Fuzzy Modeling Techniques in Economics, 10, 136–163 (2021). https://doi.org/10.33111/nfmte.202 1.136 19. Kaminskyi, A., Baiura, D., Nehrey, M.: ESG investing strategy through COVID-19 turmoil: ETF-based comparative analysis of risk-return correspondence. Intellectual Econ. 16(2), 5–23 (2022) 20. Caldara, D., Iacoviello, M.: Measuring geopolitical risk. American Economic Rev. 112(4), 1194–1225 (2022). https://doi.org/10.1257/aer.20191823 21. Matviychuk, A., Lukianenko, O., Miroshnychenko, I.: Neuro-fuzzy model of country’s investment potential assessment. Fuzzy Economic Rev. 24(2), 65–68 (2019). https://doi.org/10. 25102/fer.2019.02.04 22. Kaminskyi, A., Nehrey, M., Fedchun, A.: ESG-score effect in risk assessment of direct and portfolio investment: evidence from CEE markets. The Journal of V. N. Karazin Kharkiv National University Series: “International Relations. Economics. Country Studies. Tourism” (IRECST), (15), 49–62 (2022) 23. Umar, M., Riaz, Y., Yousaf, I.: Impact of Russian-Ukraine war on clean energy, conventional energy, and metal markets: evidence from event study approach. Resour. Policy 79, 102966 (2022). https://doi.org/10.1016/j.resourpol.2022.102966 24. Wang, Y., Bouri, E., Fareed, Z., Dai, Y.: Geopolitical risk and the systemic risk in the commodity markets under the war in Ukraine. Financ. Res. Lett. 49, 103066 (2022) 25. Long, H., Demir, E., B˛edowska-Sójka, B., Zaremba, A., Jawad, S., Shahzad, H.: Is geopolitical risk priced in the cross-section of cryptocurrency returns?. Finance Research Letters 49 (103131). https://doi.org/10.1016/j.frl.2022.103131 26. Le, V.H., von Mettenheim, H.-J., Goutte, S., Liu, F.: News-based sentiment: can it explain market performance before and after the Russia-Ukraine conflict? J. Risk Finance 24(1), 72–88 (2023). https://doi.org/10.1108/JRF-06-2022-0168 27. Do˘gan, M.: The impact of the Russia-Ukraine war on stock returns. Social Sci. Res. J. 11(1), 1–9 (2022) 28. S&P Global Sustainable. 2022 (2023). https://www.spglobal.com/esg/
The Same Size Distribution of Data Based on Unsupervised Clustering Algorithms Akbar Rashidov(B) , Akmal Akhatov, and Fayzullo Nazarov Samarkand State University, Samarkand 140107, Uzbekistan [email protected]
Abstract. It is known that dividing data into groups based on certain rules not only helps to separate meaning from the data but also increases the efficiency of the large-volume data processing process. The same size distribution of data is especially effective in approaches such as distributed computing or parallel processing. The main reason for this is that dividing the data into as equal clusters as possible allows achieving the highest efficiency results in these approaches. But the distribution of data in the same size based on the human factor is a complicated process due to the impossibility of pre-planning the data contained in the data flow and the size of the data. In such a situation, unsupervised clustering algorithms are one of the main solutions to the problem of uniform data distribution. In this research work, hierarchical, K-means, Bisecting K-means, and DBSCAN unsupervised clustering algorithms are analyzed in order to solve the given problem, and the results of research on equal clustering of different groups of data are presented using them. Moreover, a new unsupervised equal-size clustering algorithm for equal clustering of data is proposed. During the research, the efficiency indicator of this algorithm for dividing data into equal groups is compared with the indicators of existing algorithms, and the conclusions of the experimental results are presented. Keywords: Unsupervised clustering algorithms · hierarchical clustering · K-means · Bisecting K-means · DBSCAN · new equal clustering algorithm
1 Introduction Nowadays, data analysis, which is, identifying hidden contents from a data set, and separating them into useful, reliable, and interrelated data groups is one of the topics of modern research [1]. In the modern digital age, the increase in the volume of data flow increases the relevance of this research topic. Because as the flow of information increases, people’s natural data analysis capabilities are limited, and the process of extracting meaningful information from big data becomes more complicated [2–4]. In such a situation, the most effective solution for finding hidden patterns in data is unsupervised learning based on data clustering approaches according to certain criteria. Unsupervised clustering is an approach to grouping data by grouping or separating those with similar patterns, and those with different patterns from the data without human intervention [5]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 437–447, 2023. https://doi.org/10.1007/978-3-031-36115-9_40
438
A. Rashidov et al.
In these clustering approaches, the grouping of data mainly focuses on characteristics such as similarity and uniqueness of the data, and usually ignores or does not consider the number of elements in the groups to be very important. But in real life, there are such problems that not only the similarity and difference of data are important in solving these problems, but also the equality of the number of elements in clusters is an important factor [6, 7]. For example, when storing and processing data using parallel and distributed computing mechanisms. In these mechanisms, the more evenly the data is distributed to the parts of the system, the higher the efficiency [8–10]. Because in this case, all distributed functions within the system perform the same number of tasks, and all parts of the system complete their tasks at the same time. In other words, they do not wait for each other’s tasks to be completed. This is the most important indicator of achieving the highest efficiency in the work process. Therefore, achieving equal size distribution based on unsupervised clustering algorithms was taken as the major research objective. In order to solve the problem presented during the research, the capabilities of hierarchical, K-means, Bisecting K-means, and DBSCAN existing unsupervised clustering algorithms to divide data into clusters of equal size are analyzed. However, due to the limited capabilities of existing algorithms, a new unsupervised clustering algorithm is proposed to achieve uniform distribution of data. At the end of the study, the efficiency indicator of the proposed algorithm for dividing data into groups of equal size is compared with the indicators of existing algorithms, and opinions and comments are given on the results of the experiment.
2 Methodology In this research work, first of all, the literature related to the solution of the problem was analyzed. It was found that most of the literature covered only the content of unsupervised clustering algorithms [13–19]. Only in some literatures, studies related to the size of clusters have been conducted [11, 12]. However, in these literatures, the problem of clustering of the same size has not been considered. During the research, the selection of the characteristics of the data was also considered in order to divide the data into clusters. In this process, the ASCII codes of the characters in the data were used and the following features were extracted. The ASCII code of the first character of the data, the number of vowels and consonants, the sum of their ASCII codes, the number of total characters of the data, and the sum of its ASCII codes. At the next stage of the research, the data were distributed using existing unsupervised clustering algorithms based on the features extracted above, and the sizes of the clusters were determined. The research process showed that existing unsupervised clustering algorithms are not always effective in solving the given problem. Therefore, a new vector module square-based clustering algorithm was proposed during the research. These steps are described in the following sections of this article.
The Same Size Distribution of Data Based on Unsupervised Clustering
439
3 Opportunities of Existing Unsupervised Clustering Algorithms to Distribute Data into Clusters of Equal Size 3.1 Hierarchical Clustering Hierarchical clustering is one of the widely used unsupervised algorithms for clustering data into groups. Hierarchical clustering mainly uses two approaches. These are agglomerative hierarchical clustering and divisive hierarchical clustering [11, 13]. In both approaches, the clustering result can be illustrated using a dendrogram (Fig. 1).
Fig. 1. Representation of the result of hierarchical clustering using a dendrogram
In hierarchical clustering algorithms, grouping is done after the complete cluster hierarchy is built. That is, after all the elements in the collection are formed as branches of a single tree, the branches of the tree corresponding to the required number of clusters are separated. This process can be visualized using the dendrograms shown in Fig. 2.
Fig. 2. Line m divides the first dendrogram into 2 clusters, line n divides the second dendrogram into 3 clusters, line l divides the third dendrogram into 4 clusters
In Fig. 2, line m in the first dendrogram divides objects into 2 clusters, line n in the second dendrogram divides objects into 3 clusters, and line l in the third dendrogram divides objects into 4 clusters. As can be seen, only for the objects in the third dendrogram, the size of the clusters is almost equal. It can be concluded that it is not always possible to divide the data into groups of (almost) the same size by means of hierarchical clustering, which usually depends on the state of the objects to be clustered.
440
A. Rashidov et al.
3.2 K-means Clustering The K-means clustering algorithm is a center-based grouping method, and the objects in the set are clustered according to which of the central points known as centroids are close [14–17]. The number of centroids is the same as the number of clusters requested by the user. Centroids are initially chosen arbitrarily. Therefore, clustering will not be perfect in the initial state. Minimization of the sum of squares of the errors of the objects in the clusters relative to the centroids is used to make the clusters perfect (1). S=
k
dist(ci , xj )2
(1)
i=1 xj ∈Ci
where k- is the number of clusters, Ci - is the cluster, ci is the centroid coordinate of the i cluster, xj is the j element coordinate in the Ci cluster. It is known that S reaches its minimum value when the derivative of S with respect to ci is equal to 0 (2). k ∂S ∂S = dist(ci , xj )2 = 0 ∂ci ∂ci
(2)
i=1 xj ∈Ci
Based on this formula, the optimal value of the centroid of the nth cluster can be found as follows. k ∂S ∂S = (cn − xj )2 = 0 ∂cn ∂cn
(3)
i=1 xj ∈ Cn
If the number of objects in the nth cluster is m, then the following solution comes from Eq. 3 m
2·
m j=1
cn = 2 ·
m j=1
xj
⇒
m · cn =
m j=1
xj
⇒
cn =
xj
j=1
m
(4)
The conclusion from formula 4 is that the centroids reach optimal coordinates when they are equal to the average value of the coordinates of the objects in the cluster. In the K-means algorithm, objects are divided into clusters of equal size compared to hierarchical clustering. This is especially evident when objects are located at the same density in the coordinate system. In Fig. 3a, the objects are divided into 3 clusters, and the size of each cluster is almost the same. But if these objects are divided into 2 or 4 clusters, the sizes of the clusters will be very different from each other. In Fig. 3b, it can be seen that two clusters are large in size, and the other two are small in size. Figure 3c shows a perfect set of objects, i.e., a set of objects that are relatively evenly distributed over space. When this set is divided into an arbitrary number of clusters, clusters of almost the same size are formed.
The Same Size Distribution of Data Based on Unsupervised Clustering
441
Fig. 3. The result of object clustering using K-means
3.3 Bisecting K-means Clustering The Bisecting K-means algorithm is an improved interpretation of the K-means algorithm, which is more efficient than K-means in splitting into clusters of equal size. Because the main idea in Bisecting K-means is to create a new cluster by dividing a cluster with a large size or a large sum of squared errors [18]. This process is shown in Fig. 4 can be seen.
Fig. 4. Clustering of objects using the Bisecting K-means algorithm
In Fig. 4, the first set A is given. Collection A is divided into clusters B and C based on the Bisecting K-means algorithm. When the number of clusters is equal to 3, cluster B is divided into two since there are more objects in cluster B than in cluster C. Then three clusters C, D, and E are formed. When the number of clusters is equal to 4, cluster C is divided into two since cluster C has more objects than clusters D and E. After that and 4 clusters named D, E, F, G are formed. As the number of clusters increases, the same processes are continued. 3.4 DBSCAN Clustering DBSCAN is a density-based clustering algorithm that combines objects that are adjacent to each other at a distance of max_dist (maximum distance) into one cluster [19, 20]. The second important parameter of DBSCAN is min_points (minimum number of objects). It means the minimum number of objects forming a cluster. That is, if the number of objects located at max_dist distance is less than min_points, these objects do not belong to any cluster and they are considered as noise.
442
A. Rashidov et al.
This clustering method is more powerful than other clustering methods. In other words, this algorithm is the most effective approach to identifying hidden meanings. But this algorithm is ineffective in solving the problem of equal clustering. Because in the DBSCAN clustering method, it is impossible to specify the number of clusters in advance by the user. When the number of clusters changes depending on max_dist and min_points, it is impossible to say exactly how many clusters will be formed in advance. In addition, some information contained in the data to be clustered may not belong to any clusters, that is, they may be considered as noise. Taking into account these shortcomings, it can be said that it is impossible to use the DBSCAN algorithm in solving the problem of equal-sized clustering of data streams.
4 Clustering Algorithm Based on Square of Vector Modulus In this research, a new clustering algorithm is proposed in order to solve the problem of distributing the data flow into clusters of equal size. This clustering algorithm is based on the construction of vectors directed from the coordinate origin to the feature coordinates of objects. It is known that any n-dimensional vector a with coordinates (x1 ; x2 ; x3 ; ... ; xn ) has a length of a certain positive number and it is found by formula 5. |a| = x12 + x22 + x32 + ... + xn2 (5) This proposed algorithm is based on dividing the squares of vector lengths into groups based on certain rules. After the objects are converted to numerical values, clustering based on length values makes it even possible to separate two objects that are very close to each other into other clusters. This makes it possible to equalize the sizes of clusters during clustering. The proposed clustering algorithm based on the square of the vector modulus is implemented by performing the following steps: Step 1. Read the data (objects); Step 2. Transfer the specific characteristics of each data to vector coordinates in digital form; Step 3. Calculate the square of the modules of the resulting vectors; Step 4. Sort the squares of the modules of the vectors in ascending order; Step 5. Write the squares of the modules of the vectors sorted in ascending order into the array A; Step 6. Input k which number of clusters; Step 7. Divide array A into k equal parts; Step 8. Determine the boundaries of the pieces; Step 9. Find the boundaries of the cluster through the boundaries of the slices; Step 10. Separate into clusters the data using defined thresholds. As a result of the execution of this algorithm, the elements of k clusters separated have their own geometric meaning. The elements of the cluster 1 consist of the points of a circle with a radius R, the elements of clusters 2, 3,... k − 1 are the points of the rings
The Same Size Distribution of Data Based on Unsupervised Clustering
443
Fig. 5. Graphical representation of clustering based on the square of the vector modulus
with thicknesses d2 , d3 , ... dk−1 , and the elements of the cluster k are the points lying outside the circle with the radius R1 . Here R1 = R + d1 + d2 + ... + dk−1 equality will be appropriate. The results of the proposed algorithm can be seen in Fig. 5. It can be seen from Fig. 5 that in the clustering based on the square of the vector module, the points that are very close to each other also belong to two other clusters. On the contrary, the points located far from each other belong to one cluster. This feature is another indicator that distinguishes the proposed algorithm from other clustering algorithms. But the main feature of this algorithm is that the sizes of the clusters formed as a result of clustering are almost equal.
5 Results In order to check the ability of clustering algorithms to divide into clusters of equal size, 3 different data sets containing 1000 elements were used. Each data set was divided into 2, 3, 4 clusters. The elements within these clusters were enumerated and the results compared.
Fig. 6. 3 different data sets used in the experiments
3 types of data shown in Fig. 6 were used in the experiments. The data used in the experiment was generated by the random function and does not have any special meaning. But these data were created using the make_blobs data generation function of the sklearn.datasets library of the python programming language, and data were generated around 2, 6, and 30 central points, respectively.
444
A. Rashidov et al. Table 1. The results of the distribution of a, b, and c data into 2 groups using clustering
The name of the algorithm
Clusters №
Clusters size a
b
c
1
2
2
999
2
998
998
1
1
521
364
436
2
479
636
564
Bisecting K-means clustering
1
508
364
596
2
492
636
404
The proposed clustering
1
500
500
500
2
500
500
500
Hierarchical clustering K-means clustering
Table 2. The results of the distribution of a, b, and c data into 3 groups using clustering The name of the algorithm Hierarchical clustering
K-means clustering
Bisecting K-means clustering
The proposed clustering
Clusters №
Clusters size a
b
c
1
2
2
997
2
997
4
2
3
1
994
1
1
272
430
380
2
257
329
339
3
471
241
281
1
250
213
468
2
259
423
211
3
491
364
321
1
333
333
333
2
333
333
333
3
334
334
334
The Same Size Distribution of Data Based on Unsupervised Clustering
445
Table 3. The results of the distribution of a, b, and c data into 4 groups using clustering The name of the algorithm Hierarchical clustering
K-means clustering
Bisecting K-means clustering
The proposed clustering
Clusters №
Clusters size a
b
c
1
2
2
997
2
6
4
1
3
991
993
1
4
1
1
1
1
245
264
215
2
246
232
146
3
269
186
223
4
240
318
416
1
216
364
365
2
275
272
233
3
250
151
202
4
259
213
200
1
250
250
250
2
250
250
250
3
250
250
250
4
250
250
250
As can be concluded from Tables 1, 2 and 3, the proposed vector length-based clustering algorithm recorded a good result in terms of the characteristic of equal-sized clustering of most data for 3 types of data. This situation can be seen in all 2, 3, and 4 clusters. The worst indicator belongs to the Hierarchical clustering algorithm, in all cases, the sizes of the clusters were very different from each other. In all cases, a large part of the data belonged to a certain cluster, and a small part belonged to other clusters. K-means and Bisecting K-means algorithms performed relatively well as expected, but in no case did they perform better than the proposed approach. The result of the proposed algorithm for solving the given problem can be seen in Fig. 7. Figure 7 shows the distribution of the data a, b, and c into 4 clusters based on the proposed algorithm, and the 4 clusters are shown in 4 different colors.
446
A. Rashidov et al.
Fig. 7. The result of the distribution of the data set a, b, and c into 4 clusters based on the proposed algorithm
6 Conclusion In conclusion, this research work was devoted to studying the possibility of clustering algorithms to divide data into clusters of equal size. For this purpose, this possibility of existing unsupervised clustering algorithms Hierarchical, K-means, Bisecting K-means and DBSCAN clustering algorithms was studied. The results of the research showed that existing unsupervised clustering algorithms have strong capabilities in identifying hidden content in data, but limited capabilities in dividing clusters of equal size. In order to overcome these shortcomings, a clustering algorithm based on the square of the length of the proposed vector was proposed and the steps of the algorithm were presented. The research results showed that this algorithm is very effective in solving the given problem.
References 1. Maheshwari, A.K.: Business Intelligence and Data Mining. Business Expert Press, LLC, 222 East 46th Street, New York, NY 10017, p. 162 (2015) 2. Akhatov, A., Nazarov, F., Rashidov, A.: Mechanisms of information reliability in big data and blockchain technologies. In: ICISCT 2021: Applications, Trends and Opportunities, 35.11.2021 (2021). https://doi.org/10.1109/ICISCT52966.2021.9670052 3. Akhatov, A., Nazarov, F., Rashidov, A.: Increasing data reliability by using bigdata parallelization mechanisms. In: ICISCT 2021: Applications, Trends and Opportunities, 3-5.11.2021 (2021). https://doi.org/10.1109/ICISCT52966.2021.9670387 4. Jumanov, I., Djumanov, O., Xolmonov, S.: Mechanisms of image recovery optimization in the system for recognition and classification of micro-objects. AIP Conf. Proc. 2686, 020009 (2022). https://doi.org/10.1063/5.0113052 5. Tan, P.-N., Steinbach, M., Karpatne, A., Kumar, V.: Introduction to Data Mining, 2nd edn. Pearson Education, New York, NY (2019) 6. Akhatov, A., Sabharwal, M., Nazarov, F., Rashidov, A.: Application of cryptographic methods to blockchain technology to increase data reliability. In: 2nd ICACITE (2022). https://doi. org/10.1109/ICACITE53722.2022.9823674 7. Jumanov, I.I., Xolmonov, S.M.: Optimization of identification of non-stationary objects due to information properties and features of models. IOP Conf. Ser. Mater. Sci. Eng. (2021). https://doi.org/10.1088/1757-899X/1047/1/012064 8. Akhatov, A., Renavikar, A., Rashidov, A., Nazarov, F.: Development of the Big Data processing architecture based on distributed computing systems. Uzbek Journal of the Problems of Informatics and Energetics (1), 71–79 (2022)
The Same Size Distribution of Data Based on Unsupervised Clustering
447
9. Rashidov, A., Akhatov, A.: Big data and its application in various fields. Descendants of Muham-mad Al-Khwarizmi 4(18), 135–144 (2021) 10. Rashidov, A., Akhatov, A., Renavikar, A.: Optimization of the database structure based on Machine Learning algorithms in case of increased data flow. In: ICABCS (2023) 11. Höppner, F., Klawonn, F.: Clustering with size constraints. In: Jain, L.C., Sato-Ilic, M., Virvou, M., Tsihrintzis, G.A., Balas, V.E., Abeynayake, C. (eds.) Computational Intelligence Paradigms, pp. 167–180. Springer Berlin Heidelberg, Berlin, Heidelberg (2008). https://doi. org/10.1007/978-3-540-79474-5_8 12. Ganganath, N., Cheng, C.T., Chi, K.T.: Data clustering with cluster size constraints using a modified k-means algorithm. Institute of Electrical and Electronics Engineers (IEEE) (2016). https://doi.org/10.1109/CyberC.2014.36 13. Kumari, P.L., Jeeva, M., Satyanarayana, C.: A novel hierarchical document clustering framework on large TREC biomedical documents. Int. J. Inf. Technol. Comput. Sci. (IJITCS) 14(3), 16–22 (2022). https://doi.org/10.5815/ijitcs 14. Zhao, Y., Karypis, G.: Hierarchical clustering algorithms for document datasets. Data Min. Knowl. Disc. (2005). https://doi.org/10.1007/s10618-005-0361-3 15. Sinaga, K.P., Yang, M.-S.: Unsupervised K-means clustering algorithm. IEEE Access 8, 80716–80727 (2020). https://doi.org/10.1109/ACCESS.2020.2988796 16. Jumanov, I.I., Safarov, R.A., Xurramov, L.Y.: Optimization of micro-object identification based on detection and correction of distorted image points. AIP Conf. Proc. 2402, 070041 (2021). https://doi.org/10.1063/5.0074018 17. Fahim, A.: Finding the number of clusters in data and better initial centers for K-means algorithm. Int. J. Intell. Syst. Appl. (IJISA) 12, 1–20 (2020). https://doi.org/10.5815/ijisa 18. Abuaiadah, D.: Using bisect K-means clustering technique in the analysis of Arabic documents. ACM Trans. Asian Low-Resource Lang. Inf. Process. 15, 1–13 (2016). https://doi.org/ 10.1145/2812809 19. Maithri, C., Chandramouli, H.: Parallel DBSCAN clustering algorithm using hadoop mapreduce framework for spatial data. Int. J. Inf. Technol. Comput. Sci. (IJITCS) (2022). https:// doi.org/10.5815/ijitcs 20. Laskhmaiah, K., Murali Krishna, S., Eswara Reddy, B.: An optimized K-means with density and distance-based clustering algorithm for multidimensional spatial databases. Int. J. Comput. Netw. Inf. Secur. 13(6), 70–82 (2021). https://doi.org/10.5815/ijcnis.2021.06.06
Investigation of Microclimate Parameters in the Industrial Environments Solomiya Liaskovska1(B) , Olena Gumen2 , Yevgen Martyn3 , and Vasyl Zhelykh4 1 Department of Artificial Intelligence, Lviv Polytechnic National University, Kniazia Romana
Street, 5, Lviv 79905, Ukraine [email protected] 2 Department of Descriptive Geometry, Engineering and Computer Graphics Peremohy pr, National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”, 37, Kyiv 03056, Ukraine 3 Department of Project Management, Information Technologies and Telecommunications, Lviv State University of Life Safety, Kleparivska Street, 35, Lviv 79007, Ukraine 4 Department of Heat and Gas Supply and Ventilation St. Bandery, Lviv Polytechnic National University, 12, Lviv-13 79013, Ukraine
Abstract. Researches on the organization and implementation of the project management of selection processes and the development of adequate models of interaction of many parameters of the microclimate of the technological process of raising the poultry meat breeds in industrial premises have been carried out. Actuality of the researches has been confirmed, the analysis and a choice of interaction of parameters research methods are proved. It is shown that the use of physical modeling with the involvement of the corresponding experimental equipment and the use of the experimental results obtained for geometric modeling on the basis of applied multidimensional geometry are effective. The significance of the study is in the established features of the nature of the heat exchange of air in the premises in view of the effectiveness of growing products and in various types of farming. The method of measuring the temperature in the room chosen in the process of geometric modeling allows to cover with lines of the same temperature all points of the living space of the production premises in which the bird is located.The proposed method of research can be practically adapted for the processing of the results of studies on the microclimate parameters of agro-industrial objects of various intended uses. Keywords: Industrial premises · microclimate · parameters · physical and geometric modeling
1 Introduction The effective implementation of technological processes of production in the industrial and agrarian sectors has big impact in development of plant. With regard to the food production, in particular, poultry, an important additional condition is the observance of the comfortable microclimate in the zone of its cultivation. For this, there are industrial © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 448–457, 2023. https://doi.org/10.1007/978-3-031-36115-9_41
Investigation of Microclimate Parameters in the Industrial Environments
449
buildings that contain poultry. The main elements of poultry keeping are such components of comfort as the air temperature and its purity. Heaters and ventilation systems are used to provide regulatory requirements. Their effective work is possible when rational modes of operation of the corresponding technological equipment are used. The development of appropriate approaches to the creation of industrial heating and ventilation systems for industrial buildings with the massive holding of poultry, the choice of means and the study of the parameters of the process constitute one of the promising directions of development of the agricultural sector for the holding of meat breed poultry. Scientific research, however, is scattered on a particular aspect of the study of the comfortable stay of the poultry in the poultry house [1–14]. A free scientific niche is the question of a reasonable choice based on the principles of project management of selection and development of tools for processing graphical experimental data with the involvement of the apparatus of geometric modeling [15, 16]. The selection of previously unsettled parts of the general problem points to a limited number of scientific publications in its separate direction concerning the provision of the comfortable microclimate in industrial premises with the massive holding of poultry. In addition, the recent studies and publications have been analyzed. They testify to the absence of a unified methodological approach to the organization and analysis of experimental studies, taking into account the project approach. 1.1 Defining of the Peculiarities of Resource Support for the Technological Process of Poultry Breeding In many countries, a project to grow agricultural products under hothouse conditions is successfully implemented and improved. This applies to both vegetable and livestock breeding. In particular, projects for the implementation of appropriate technological processes are being carried out for the breeding of meat breed poultry. In the process of the project initiating, fundamental decisions are taken on the creation and arrangement of an industrial building (poultry house). Project planning defines the purpose of the project and the peculiarities of resource support for the technological process of poultry breeding. At this stage of the project management, it is important to develop appropriate measures and to take measures to ensure proper conditions for holding of poultry, which is one of the important moments in the implementation of the technological process of poultry breeding. The aim of the study is to find out on the basis of the project approach the main components of the project on the research of the microclimate of industrial premises, to substantiate, develop and apply means of geometric analysis of the graphic dependencies of its parameters. The process realization of research has been carried out on the example of determining the amount of heat utilized from the premises, the work of local tidal and exhaust ventilation. In accordance with the research objectives, a universal laboratory installation that can serve as a tool for conducting research is installed (Fig. 1).
450
S. Liaskovska et al.
Fig. 1. The graphic visualization of the laboratory installation, 1—room; 2—infrared heater; 3—exhaust outlet; 4—air duct; 5—ventilator; 6—thermometer; 7—coordinate grid; 8—floor
The laboratory installation, located in the room 1, contains an infrared heater 2 over which there is an exhaust outlet 3 connected by an air duct 4 with a ventilator 5. To measure the temperature at an arbitrary point of space 1, a thermometer 6 with a coordinate grid 7 installed on the floor 8 is provided. In the scheme (Fig. 1) there are no commuting devices, conventionally not depicted, needed for the formation of the given equipment the scheme of intended use.
Fig. 2. Isotherms of the cutting plane 1–1
The work of local exhaust ventilation the temperature fields in the room space 1 1—room; 2—infrared heater; 3—exhaust outlet; 4—air duct; 5—ventilator; 6—thermometer; 7—coordinate grid; 8—floor. The temperature of the air we can measure with the thermometer 6 and use the coordinate meter 7. When studying the work of
Investigation of Microclimate Parameters in the Industrial Environments
451
local exhaust ventilation the temperature fields in the room space 1 are compared with the ventilator 5 switched off and on. The results of measurements were given by the isotherms in the cutting plane with the measurements of the height of the room h and the coordinate x of the room 1 with a constant value of the coordinate y, that is the width of the room (Fig. 2 a, b).
2 The Comparative Analysis of the Insulation Isotherms Experimental Results This density is characterized by h (the maximum value of exceeding the height of the isotherm) at a certain interval x (Fig. 2 a) and (Fig. 2 b). The interval Δx = 0.3 0.8 m is common to the isotherms. Table 1 shows Δh in the section plane 1–1. 2.1 Solving Model Based on Graphic Dependencies The values of h (some isotherms) plotted in the secant plane 1–1 are given in Table 1. Table 1. The value of the slope of isotherms h Isotherm’s meaning
16.5
without a fan h, m
0.45
with a fan h, m
> 1.4
16.6 0.11 0.3
17 0.1 0.1
18 0.09 0.11
19 0.08 0.15
Graphic dependencies, constructed according to the data of Table 1, give an opportunity to determine the effectiveness of exhaust ventilation (Fig. 3).
Fig. 3. Determination of the field of effective influence of the work of local exhaust ventilation: the work of local exhaust ventilation; without with the work of local exhaust ventilation
452
S. Liaskovska et al.
From analysis of graphic dependencies Δh = f (t °C) we have that the area of the effective influence of the work of the local exhaust ventilation starts at a higher altitude then h > 1.0m (isotherm of 17°C). Almost each of the isotherms has a minimum. In the case of local exhaust ventilation, there is a slight change in the height h of the location of the minimum height value of each isotherm (Table 2), in addition to the isotherm 17°C, as well as their condensation in the upper part of the room. Height interval Δhb of the location of isotherms in the upper part of the room: – without the work of local exhaust ventilation for isotherms of interval17zC… 19°C value Δhb = 0.42m; – with the work of local exhaust ventilation for the same isotherm interval of 17zC… 19°C value Δhb = 0.28m.
Table 2. Minimum height values of the isotherm location Isotherm
16.5
16.6
17
18
19
h, m without a fan
0
0.51
0.83
1.18
1.25
h, m with a fan
0
0
1.08
1.19
1.3
Height interval Δhb of the location of isotherms in the lower part of the room: – without the work of local exhaust ventilation for isotherms of interval 0zC 16.7°C value Δhb = 0.68m; – with the work of local exhaust ventilation for the same isotherm interval of 0zC 16.8°C value Δhb = 0.23m. Thus, there is the decrease in the height interval in the upper part of the room in relation to the lower part of the room without the work of local exhaust ventilation 0.6 times and the increase in the height interval in the upper part of the room in relation to the lower part of the room with the work of local exhaust ventilation 1.2 times.
3 Processing Results by Geometric Method Using Complex Drawing The research methodology consists in the deep and more thorough use of geometric modeling tools in the process of analyzing the results of experimental research through an effective project approach to the implementation of the described stages of research. In the process of scientific research of the thermal field using experimentally obtained isotherms (Fig. 2), it becomes necessary to use those isotherms that are not on the graph. It is possible to construct the necessary isotherms, if we consider such lines as the result of intersection of the surface of the temperature field with the horizontal cutting plane with a given required temperature value.
Investigation of Microclimate Parameters in the Industrial Environments
453
Let’s show the sequence of construction of an isotherm at a temperature value, for example, 18.5°C in the case of local exhaust ventilation system work. We isolate two neighboring isotherms, in this case the isotherms are with temperature values 18.0°C and 19.0°C (Fig. 4). We construct a complex drawing of the section of the temperature field with values of isotherms 18.0°C and 19.0°C (Fig. 5). Typically, the required lines of level of surface of the temperature field are constructed by passing the cutting plane 18.5 between the planes of level with traces 18 and 19 (Fig. 5).
Fig. 4. Determination of the field of effective influence of the work of local exhaust ventilation
Fig. 5. Determination of the field of effective influence of the work of local exhaust ventilation
Taking into account that the surface of the temperature field is smooth, the line of intersection of the temperature field with the cutting plane of level 18.5 is equidistant to the lines, isotherms, with traces 18 and 19 (in Fig. 5 it is indicated by a dashed line). The proposed complex drawing allows determining the temperature of the air at an arbitrary point on the surface of the thermal field.
454
S. Liaskovska et al.
Let’s construct projections of a point of the temperature field with the value of temperature, for example, t = 18.7°C at x = 0.3m (Fig. 6). The construction is carried out in the following sequence: 1. First we conduct a trace of the cutting plane at 18.7 (in Fig. 6 it is indicated by a dotted-dashed line). 2. Then we conduct a projection link line at x = 0.3m perpendicular to the axis Ox. At the intersection of both lines we obtain point T1, which is a horizontal projection of the temperature value 18.7°C of surface of the temperature field. 3. Through the point T1 we pass a section a1 of an arbitrary straight line which intersects the traces of the cutting planes 18.5 and 19.0 at points 11 and 21 respectively.
Fig. 6. Constructing projections of a point of the temperature field
4. Their projections 12 and 22 are found on isotherms 18.5 and 19.0, respectively. By connecting the points 12 and 22, we obtain the projection a2 of the section of an arbitrary line a. 5. Point T2 of the intersection of the projection lines and section a2 is the required point of the temperature field surface with the temperature value of 18.7°C. The complex drawing in Fig. 5 gives two projections of the temperature field in Oxht0 space. According to such projections, a 3D model of the temperature field, for example, in the working range 18… 19°C of temperature changes in the room can be constructed by means of graphic information technologies (Fig. 7).
Investigation of Microclimate Parameters in the Industrial Environments
455
A visual computer model of temperature distribution in the room is also constructed when the coordinate is changed (Fig. 8). The researches of the air temperature at change of coordinate value show constancy of the pattern of temperature distribution in cutting plane 1–1. Effective project implementation of the technological process of raising poultry in the direction of conducting comprehensive scientific research also involves identifying the features of the influence of tidal ventilation on the microclimate in the production premises. In the process of conducting research on the influence of tidal ventilation, the ventilator 5 of the laboratory installation provided air from the air channel 4 and the exhaust outlet 3 to the room 1. The overall picture of the location of the isotherms is the same: all lines have a minimum located within 3 < x < 4 m. As with exhaust ventilation, there is an extension of all isotherms without exception. An increase in the speed of supply of inflow air causes a decrease in the height of the maximum value of the temperature in the room. The trends in the influence of the exhaust ventilation on the nature of the distribution of heat in the room will be considered by the example of the isotherm displacement, in
Fig. 7. Constructing projections of a point of the temperature field
When Δh exceeds the height of the 24zC isotherm throughout the gap Δx = 0… 7 m is equal to 0.18m for the speed of the supply air flow v = 0.1 m/s; 0.45m for v = 0.2 m/s; 0.46 m for v = 0.35m/s particular 24°C (Fig. 9). With an increase in the velocity of the flow of tidal air from 0.1 m/s to 0.2 m/s, the isotherm of 24°C dropped from a height h = 1.5 m to a height of h = 0.9 m, that is 60%. With an increase in the velocity v of the flow of tidal air from 0.2 m/s to 0.35 m/s it dropped to a height of 0.55 m. Exceeding h height of isotherm 24°C across the entire gap x = 0… 7 m is 0.18 m for the velocity of the flow of tidal air v = 0.1 m/s; 0.45 m for v = 0.2 m/s; 0.46 m for v = 0.35 m/s.
456
S. Liaskovska et al.
Fig. 8. 3D model of temperature distribution along the axis
Fig. 9. Dependence of incoming air velocity and location of isotherms
4 Summary and Conclusion In this paper, we analyzed that the involvement of that the involvement of geometric modeling allows to effectively study the parameters of the microclimate of industrial buildings. An example of the technological process of poultry breeding is considered. A geometric method of processing experimental parameters has been implemented. Quantities that are difficult or impossible to determine under experimental conditions were also investigated.
Investigation of Microclimate Parameters in the Industrial Environments
457
The obtained scientific results demonstrate the use geometric modeling using multidimensional geometry. This method will allow research of the agro-industrial complex for various purposes.
References 1. Made, R.I., et al.: Experimental characterization and modeling of the mechanical properties of Cu–Cu thermocompression bonds for three-dimensional integrated circuits. Acta Mater. 60(2), 578–587 (2012) 2. Zhou, H., Li, Q., Lin, T.: Power system disturbance and operation identification based on WAMS electric power. Automation Equip. 31(02):,7-11 (2011) 3. Liu, X., Yang, M.: Simultaneous curve registration and clustering for functional data. Comput. Stat. Data Analysis 53(4), 1361-1376 (2019) 4. Rama, A., Lekyasri, N., Rajani, K.: PID control design for second order systems. IJEM 9(4), 45–56 (2019) 5. Qin, X., Bi, T., Yang, Q.: A new method for hybrid nonlinear state estimation with PMU. Automation of Power Syst. 31(4), 28–32 (2017, in Chinese) 6. Pijush, D., Madhurima, M., Asok, K.: Parametric optimization of liquid flow process by ANOVA optimized DE, PSO & GA algorithms. IJEM 11(5), 14–24 (2021) 7. Karimui, R.Y.: A new approach to measure the fractal dimension of a trajectory in the highdimensional phase space. Solitons & Fractals 151, 111–239 (2021) 8. Zarei, M., et al.: Employing phase trajectory length concept as performance index in linear power oscillation damping controllers. IEPE 98, 442–454 (2018) 9. Syed, K., Ali, S., Waqas, M.: Smart home automation using IOT and its low cost implementation. IJEM 10(5), 28–37 (2019) 10. Wang, L., Sharkh, S., Chipperfield, A.: Optimal decentralized coordination of electric vehicles and renewable generators. IJEPES 98, 474–487 (2018) 11. Ali, S., Xie, Y.: The impact of industry 4.0 implementation on organizational behavior and corporate culture: the case of pakistan’s retail industry. IJEM 10(6), 20–31 (2020) 12. Zarei, M., et al.: Oscillation damping of nonlinear control systems based on the phase trajectory length concept: an experimental case study on a cable-driven parallel robot. Mech. Mach. Theory 126, 377–396 (2018) 13. Yu, Q., Wang, X., You, J., et al.: Equality constraints two-step state estimation model based on phasor measurements. Power System Technol. 31(10), 84-88 (2007, in Chinese) 14. Savelyev, A.V., Stepanyan, I.V.: Spiral flows at the cardiovascular system as the experimental base of new cardiac-gadgets design. IJEM 8(6), 1–12 (2018) 15. Hovorushchenko, T., et al.: Development of an intelligent agent for analysis of nonfunctional characteristics in specifications of software requirements. Eastern-European J. Enterprise Technol. 1(2), 6–17 (2019) 16. Liaskovska, S., Izonin, I., Martyn, Y.: Investigation of anomalous situations in the machinebuilding industry using phase trajectories method. ISEM 463, 49–59 (2021)
Detection of Defects in PCB Images by Separation and Intensity Measurement of Chains on the Board Roman Melnyk and Ruslan Tushnytskyy(B) Lviv Polytechnic National University, Lviv 79013, Ukraine [email protected], [email protected]
Abstract. The subtraction requires constant comparison with the input images. The developed approach automatically demonstrates defects and chains containing them. Algorithms for K-means clustering, flood-filling the image traces with color, thinning of binary images are used to find the coordinates of the contacts and defects on the images of the printed circuit board. The clustering algorithm is used to reduce the number of colors, obtain uniform colors, and highlight the objects of the image. The thinning algorithm is used to construct skeletons and find special points indicating the contact pixels. Pixels of contacts are accepted as start positions to mark pixels of traces in the test image. The flood-fill algorithm is used to mark contacts, tracks, and background. The developed approach detects PCB defects of connectivity, and internal defects of traces. Keywords: Printed circuit board · defect detection · chain · contact · pixel · switch · ending · flood-fill · thinning · K-means clustering
1 Instruction There are many modern approaches for PCB defects detection which are in the published articles. They conditionally could be grouped into three classes. The first one unites works describing techniques for fault diagnostics, and artificial networks for image processing. One representant of this group is the works [1] in which the PCB surface inspection is realized by a powerful Deep Learning tool. Works of the second class are based on comparing a standard PCB image with a PCB image to be inspected. Approaches use different subtraction algorithms. For example, the work [2] consider a normalized histogram to select components of the PCB image and then separates defects by the subtraction algorithm. As a result defects are classified. The paper [3] presents an approach for defects detection based on the connected table of a reference image. In many cases this table is unknown. And coordinates of contacts are to be determined. In the publications [5–8] to compare two PCB images a subtraction algorithm to detect the defected regions was used. In the works [2, 9, 10] different methods of PCB defect detection are presented. In this article detection of contacts coordinates in a PCB image is realized by segmentation with the K-means clustering algorithm, flood-filling of PCB chains, and thinning © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 458–467, 2023. https://doi.org/10.1007/978-3-031-36115-9_42
Detection of Defects in PCB Images by Separation and Intensity
459
algorithm to find ending and switches in skeletons. The found pixels of contacts are input data for checking procedures and algorithms to find defects of connectivity, errors in contacts and traces and extra material on a PCB background.
2 PCB Image Preprocessing Algorithms The main part of the defects of printed circuit boards is caused by incorrect shape and filling of contacts and traces. Defects in the image of a printed circuit board can be conventionally grouped into three classes: the contact surface, the surface of traces, and metal residues in the background space. Defects are associated with the wrong amount of conductive material and the wrong shape of contacts and traces. PCB images show defects as background colors or traces compared to a reference image. In one case, the defect is marked by background pixels. Otherwise, the defect is indicated by additional pixels of the trace (Fig. 1).
Fig. 1. PCB image with marked two defects: open and short
The number of pixel defects is very small and it is very difficult to find a tool to measure it. A very sensitive approach is used to illustrate this fact. The method discussed below constructs two images: DCH(r) and DCH(m). They are distributed cumulative histograms of the reference and manufactured PCB image. Then a special logical subtraction operation is applied to them: DCH (r − m) = DCH (r) − DCH (m). This operation is called special because it is applied to grayscale images and different pixels are marked with red or blue. The resulting image is shown in Fig. 2. The DCH(r - m) image contains two bands indicating places where DCH(r) and DCH(m) differ by 3–5 intensity units. The red bar indicates increased background pixels and the blue bar indicates additional trace pixels. Their OX coordinates correspond to the OX coordinates of the defects on the printed circuit board image. Thus, it is possible to measure the integral intensity of defects and determine the OX coordinates. Two more distributed cumulative histograms for the OY axis are required to detect the full defect positions. This approach does not allow us to specify the type of defects, recording only irregularities in the distribution of pixels. The previous image of a PCB is taken for discussion about used preprocessing algorithms. Their goal is to prepare the PCB image to be processed for reliable results. And then to apply algorithms for the detection of defects and measurement their intensity.
460
R. Melnyk and R. Tushnytskyy
Fig. 2. Difference of two distributed cumulative histograms
The reference image is shown in Fig. 2a. It is the image from Fig. 1, but without defects. Also, for illustration in Fig. 3b the distributed cumulative histogram of the input image is shown. It takes part in the special operation of subtraction.
a
b
Fig. 3. PCB image (a) and its distributed cumulative histogram (b)
2.1 Distributed Cumulative Histogram To build the image of a distributed cumulative histogram two sets of ordinary histograms of numbers N for columns and M for rows are calculated: Vi (c) = Vij (c) , j = 0, 255, i = 1, N (1) Vj (r) = Vji (r) , i = 0, 255, j = 1, M Then two distributed cumulative histograms as sets of frequencies sums: Vj (cc) = Vj (cr) =
l
l=0 i
Vli (c), i = 0, 255 , j = 1, N Vli (r), i = 0, 255 , j = 1, M
(2)
l=0
where V i (cr), V j (cc) – are cumulative histograms in rows and columns, V ij (c), V li (r) – are an intensity frequencies in column and row, N, M – are numbers of columns and rows. The last equations are the mathematical models of distributed cumulative histograms of an image.
Detection of Defects in PCB Images by Separation and Intensity
461
2.2 K-Means Clustering Slgorithm The K-Means clustering algorithm approximates the PCB image by K segments of pixels with the same intensity. This algorithm assigns pixels into different clusters to minimize a sum of distances between the centroids and pixels within a cluster. The criterion function is as follows: S=
K m i=1 k=0
wik |I − I k |, S =
K m
wik [I − I k ]2
(3)
i=1 k=0
where wik = 1 if a pixel of intensity I ir belongs to the cluster k; otherwise, wik = 0; I k is the intensity of the i-th centroid. The input image has three colors. So, the clustering is processed for K = 3. The clustered image is shown in Fig. 4a. Its DCH(r) image is shown in Fig. 4b.
a
b
Fig. 4. Clustered PCB image (a) and its distributed cumulative histogram (b)
The DCH(r) image distinctly shows three groups of pixels corresponding to contacts, traces, and background. Segmenting the PCB image by intensity levels marked with red in Fig. 4.b allows us to separate contacts and traces. After darkening for better visibility, they are shown in Fig. 5a and Fig. 5b.
a
b
Fig. 5. Segmented contacts (a) and traces (b) in the clustered PCB image
Separation of contacts and traces is not the best solution because many connectivity defects arose in places of their touching. Often these defects are very thin and are of poor visibility.
462
R. Melnyk and R. Tushnytskyy
2.3 Flood-Fill Algorithm The developed flood-fill algorithm realizes four special functionalities: 1–2) a start pixel can be appointed by coordinates or by the intensity value, 3–4) colors can be distributed within a closed area or among all pixels with a target color. The last case is demonstrated in Fig. 6 by marking pixels of one level (a), or of a few levels.
a
b
Fig. 6. Two strategies of flood-filling: one level (a) several levels (b)
In Fig. 7a the clustered PCB image presents some blue traces and red contacts. Then these elements are segmented and shown in Fig. 7b.
a
b Fig. 7. Clustered image (a) and segmented contacts and traces (b)
These segmented elements are ready for further processing by algorithms of defects detection.
Detection of Defects in PCB Images by Separation and Intensity
463
2.4 Thinning Algorithm The Hilditch’s Thinning Algorithm [4] works with binary (black and white) images. When applying the flood-fill algorithm to the clustered PCB image (white for background, black for contacts and traces) the binary PCB image in Fig. 8a is ready for further processing. The thinning algorithm builds a skeleton shown in Fig. 8b. The algorithm realizes two additional functionalities: 1) it finds endings for lines, 2) it finds switches for nodes. The first and second elements are shown in Fig. 8b. The last tool is very important because it finds coordinates of pixels which are accepted as contacts of chains and assigned in fact to identify chains. Also, they are used as starting points for the flood-fill algorithm to build and select chains in the reference and real PCB images.
a
b Fig. 8. Binary image (a) and its skeleton with endings and switches (b)
3 Selection and Separation of Chains 3.1 Connectivity Defects After finding coordinates of endings and switches in the skeleton image the following step is to bind chains in the reference PCB image with their specific points: ending and switches. It is realized by a help of the flood filling algorithm. Arbitrary ending or switch pixels are taken as start points. The resulting PCB image is shown in Fig. 9. Note that not all chains are not flood-filled. An order in which chains are flood-filled gives us their ID numbers after flood-filling all endings and switches are with corresponding colors and ID numbers of chains to which they belong. Thus, a set of pixels as specific points for the reference PCB image Ee = {E1 , E2 , . . . , En } is formed, where E i – is a set of pixels belonging to the i-th chain.
464
R. Melnyk and R. Tushnytskyy
All endings and switches are placed in middle geometrical positions of contacts and traces. So, their positions are not very sensitive from the manufacturing process. They are taken as starting points for the manufactured PCB to select chains in the image: De = Ee, {D1 , D2 , . . . , Dn } = {E1 , E2 , . . . , En } where De is a set of specific points for the manufactured PCB image and Di is a set for the i-th chain.
Fig. 9. The reference PCB image after flood filling of four chains
3.2 Separation of Chains and Connectivity Defect Detection Taking elements from the set Di as starting points for the flood-fill algorithm chains from the manufactured PCB image are selected and separated. Figure 10a shows two sets of starting points D1 = {B, C, D, H }, D2 = {A, E, F, G}, and two flood-filled chains in the reference image. Figure 10b shows one correct chain and one chain with open defect. Figure 10c shows two chains as one chain caused by short defect. It is in the manufactured PCB image. To select the red chain with the open defect two iterations of flooding are used: for the point B and for the point H. Two iterations indicate available open defect. One color for two and more sets of starting points indicate available short defect. Two blue chains with the short defect are flooded and selected by one starting point and one iteration.
Fig. 10. Illustration of flood-filling and selection of of chains: chains in the reference (a), open (b), short (c)
The resulting PCB image after three iterations is shown in Fig. 11.
Detection of Defects in PCB Images by Separation and Intensity
465
When defective chains are selected and separated on a basis of starting points the places of defects can be found by new application of the thinning algorithm. In Fig. 12a two chains from the reference image and its skeleton are shown. The image of selected defective chains and its skeleton are shown in Fig. 12b.
Fig. 11. The manufactured PCB image after flood filling of two chains
Fig. 12. Chains after flood-filling and thinning: of the reference PCB (a), of the manufactured PCB (b)
The coordinates of the found specific points directly indicate the places of defects: endings indicate a circuit break, and switches indicate a short circuit. 3.3 Measurement of Intensity Defects To measure the chain inaccuracy a mean intensity function in the W columns and H rows of the image matrix is used: ⎛ ⎞ W I (i) = 1/W ⎝ I (i, j)⎠, i = 1, 2, ..., H , (7) j=1
I (j) = 1/H
H
I (i, j) , j = 1, 2, ..., W ,
(8)
i=1
where I (i, j) is pixel intensity in the i-th row and the j-th column (1 ≤ i ≤ H, 1 ≤ j ≤ W). The mean intensity functions of the reference and real components in Fig. 13 illustrate the different numbers of pixels in two defect places.
466
R. Melnyk and R. Tushnytskyy
3.4 Evaluation by Distributed Cumulative Histogram To visualize defects a distributed cumulative histogram is more productive. It is possible to notice changes of a difference between pixels on the etalon and real samples.
Fig. 13. Mean intensity functions of two chains
Figure 14 contains two DCH of the etalon and defective chains, and their difference overlayed with a chain.
Fig. 14. Difference of DCH images and an overlay
4 Summary and Conclusion Developed approach allows us to inspect all open and short defects visible for a camera and innoticable for the user. The K-Means clustering algorithm, flood-filling, thinning, and distributed cumulative histograms are used to detect PCB defects and to visualize them.
Detection of Defects in PCB Images by Separation and Intensity
467
The developed approach automatically demonstrates defects and chains containing them. The developed software on the basis of the created mathematical apparatus confirmed the expediency and effectiveness of the developed processing concept.
References 1. Jungsuk, K., Jungbeom, K., Hojong, C., Hyunchul, K.: Printed circuit board defect detection using deep learning via a skip-connected convolutional autoencoder. Sensors 21(15), 4968 (2021) 2. Chaudhary, V., Dave, I., Upla, K.: Automatic visual inspection of printed circuit board for defect detection and classification. International Conference on Wireless Communications, Signal Processing and Networking (WiSPNET), pp. 732–737 (2017) 3. Tatibana, M.H., Lotufo, R.A.: Novel automatic PCB inspection technique based on connectivity. In: Proceedings X Brazilian Symposium on Computer Graphics and Image Processing, pp. 187–194 (1997) 4. Hilditch’s Thinning Algorithm (2023). http://cgm.cs.mcgill.ca/~godfried/teaching/projec ts97/azar/skeleton.html 5. Chauhan, A.P., Bhardwaj, S.C.: Detection of bare PCB defects by image subtraction method using machine vision. In: Proceedings of the World Congress on Engineering, vol. II, WCE, pp. 68 (2011) 6. Bond. An Efficient and Versatile Flood Fill Algorithm for Raster Scan Displays (2011). www. crbond.com 7. Kamalpreet, K., Beant, K.: PCB defect detection and classification using image processing. Int. J. Emerging Res. Manage. Technol. 3(8), 42–46 (2014) 8. Moganti, M., Ercal, F., Dagli, C.H., Tzumekawa, S.: Automatic PCB inspection algorithms: a review. Comput. Vis. Image Underst. 63(2), 287–313 (1996) 9. Sarath, A.K., Kumar, N.S.: A review of PCB defect detection using image processing. Int. J. Eng. Innovative Technol. (IJEIT) 4(11), 188–192 (2015) 10. Putera, S.H., Ibrahim, Z.: Printed circuit board defect detection using mathematical morphology and MATLAB image processing tools. In: 2nd International Conference on Education Technology and Computer, 5, pp. 359–363 (2010)
Tropical Cyclone Genesis Forecasting Using LightGBM Sabbir Rahman(B) , Nusrat Sharmin, Md. Mahbubur Rahman, and Md. Mokhlesur Rahman Military Institute of Science and Technology, Mirpur, Dhaka 1216, Bangladesh [email protected]
Abstract. Cyclones, or tropical storms, are a major threat to coastal communities and industries around the world. Accurate and timely forecasts of cyclone strength and trajectory are critical for minimizing their impact. In recent years, machine learning algorithms have shown promise in predicting cyclone genesis by analyzing large amounts of data and identifying patterns and correlations. In this paper, we propose a new machine learning approach using the LightGBM technique for cyclone genesis forecasting. Our experiments were conducted using the Best Track dataset, which includes wind speed, sea level pressure, and surface moisture flux data. We compare our method to the Random Forest algorithm and provide a detailed analysis of the performance of both approaches. Our findings demonstrate that the LightGBM method outperforms Random Forest and offers improved accuracy and efficiency in cyclone forecasting. Our study contributes to the field of cyclone prediction by offering a novel machine learning approach that can help mitigate the impact of these extreme weather events on society. Keywords: Machine Learning · Weather Prediction · LightGBM · Random Forest
1 Introduction Cyclones are dreadful weather phenomena that can kill people and seriously harm infrastructure. To predict cyclones’ potential impact and reduce the potential damage they can do, accurate forecasting is necessary. [1] Presently, the Saffir-Simpson scale, which has categories 1 through 5, is used to categorize cyclones according to wind speed. [2] For better catastrophe management and prediction, more precise classification systems might be needed because this system might not always be able to capture the complexity of cyclone data. Machine learning algorithms have shown promise in cyclone genesis forecasting, with various algorithms such as decision trees, logistic regression, random forest, AdaBoost, support vector machine, support vector regression, and multiple linear regression applied for tropical cyclone genesis forecasting. However, previous studies have not explored the use of the LightGBM algorithm in this context, and there is no concrete conclusion on the use of different parameters from various datasets. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 468–477, 2023. https://doi.org/10.1007/978-3-031-36115-9_43
Tropical Cyclone Genesis Forecasting Using LightGBM
469
In this study, we propose a new cyclone categorization system that classifies cyclones into six categories, ranging from 0 to 1. This new system aims to provide a more nuanced understanding of cyclones, with improved accuracy and relevance to disaster management. Categories 2 to 6 are further categorized according to the Saffir-Simpson scale. In order to categorize cyclones, we employed the LightGBM algorithm, a well-liked machine learning technique that has shown effectiveness in a number of classification applications. We found that LightGBM performed better than the Random Forest algorithm in terms of precision and accuracy, demonstrating that it is a better algorithm for this particular application. In the following sections, we provide an overview of related work on cyclone categorization and machine learning algorithms. We describe our dataset and the preprocessing steps taken to prepare it for the machine learning models. We present our experimental results, including a detailed analysis of the performance of LightGBM and Random Forest. Finally, we conclude our study and discuss future directions for research on cyclone categorization. Our study aims to provide a more accurate and nuanced approach to cyclone categorization, which can help improve disaster management efforts.
2 Literature Review Several studies have explored the use of machine learning algorithms for weather prediction, including LightGBM and Random Forest models. In this section, we describe relevant works related to cyclone prediction and evaluate their relevance to our research objectives. A model based on the LightGBM algorithm was created by Xinwei Liu et al. [3] to identify and predict severe convective events, such as hail, brief periods of very heavy rain, and convective gusts. The authors classified various types of severe weather using ground data and C-band radar echo outputs. This study shows that LightGBM is excellent in forecasting extreme weather even if it focuses on diverse kinds of weather occurrences. To improve forecasting performance and rectify numerical prediction findings, Rongnian Tang et al. [4] suggested a spatial LightGBM model. By utilizing a single-station, single-time approach and capturing the local spatial information of stations, the authors were able to give high-performance correction of medium-range forecasts. Although cyclone prediction is not the primary focus of this study, the proposed model’s capacity to capture local spatial information may be helpful in predicting cyclone movement. To estimate dominating wave periods in oceanic waters, Pujan Pokhrel et al. [5] suggested a forecasting model based on the LightGBM algorithm. To build the model, the authors used a dataset of oceanic wave data collected from five separate locations. Despite the fact that this work primarily focuses on oceanic waves, cyclone behavior could be predicted by using LightGBM to forecast wave patterns. Pingping Wang et al. [6] proposed a novel intensity applicable conformal prediction framework using the Random Forest model as the underlying algorithm. The authors used satellite infrared images of tropical cyclones to extract 71 intensity-related features that may be divided into four categories: eye features, circle features, texture features, and time-series features. This study is directly relevant to our research objectives, as it
470
S. Rahman et al.
demonstrates the effectiveness of Random Forest in predicting the intensity of tropical cyclones. Gregory R. Herman et al. [7] used historical forecasts from NOAA’s Second Generation Ensemble Forecast System Reforecast ensemble to provide probabilistic forecasts of severe weather in the contiguous United States. The authors found that SPC outlooks performed somewhat better on Day 1 than RF outlooks, but much better on Days 2 and 3. Although this study focuses on severe weather in general, the use of Random Forest in probabilistic forecasting could be useful in predicting the path of cyclones. In Tamil Nadu, India, Rajasekaran Meenal et al. [8] investigated the application of machine learning to estimate global sun radiation and wind speed. The authors tested the Random Forest machine learning model with statistical regression and SVM models using data on recorded wind and sun radiation from IMD, Pune. Although cyclone prediction is not the main emphasis of this study, using Random Forest to anticipate wind speed may help in predicting cyclone behavior. These studies demonstrate the accuracy of the LightGBM and Random Forest models at forecasting wind speed, dominant wave durations, and severe weather events. The proposed models can capture local geographical data, extract features linked to intensity, and make probabilistic projections. Although some studies don’t specifically address cyclone prediction, they can be changed to do so by changing the machine learning algorithms and forecasting techniques they employ.
3 Methodology The proposed methodology aims to address the research objective of accurately predicting the category of a new, unseen cyclone in the North Indian Ocean region. To achieve this objective, the study uses machine learning techniques, specifically feature selection and a mapping function to predict the cyclone category based on input features. 3.1 Problem Formulation We can represent the input features of a cyclone as a matrix X, where each When analyzing data related to cyclones, we can organize the various factors that make up a cyclone into a matrix called X. Each column in this matrix indicates a distinct characteristic of the cyclones in each row, such as wind speed or temperature. As an illustration, Cyclone A’s wind speed would be listed in one column, while Cyclone B’s temperature would be listed in another column. We may make use of a vector called Y to keep track of the cyclone types. Each component of this vector represents a particular cyclone’s categorization. For instance, if Cyclone C is designated as a Tropical Storm, the vector would include an element to reflect that information. The issue of cyclone categorization can be described as follows: Given a dataset D = (X , Y ) containing n cyclones, where X is an nxm matrix of input features and Y is an nx1 vector of output categories, we seek to learn a mapping f : X → Y that can accurately predict the category of a new, unseen cyclone. We can
Tropical Cyclone Genesis Forecasting Using LightGBM
471
represent this mapping as a function that takes as input a vector of input features xj for a given cyclone j, and outputs a predicted category y j . This function can be defined as: yj = f xj
3.2 Dataset The North Indian Ocean 5-day predictor dataset contains historical data on cyclones in the North Indian Ocean region from 1990 to 2017. The dataset is used to anticipate the track and intensity of upcoming cyclones based on a variety of meteorological factors. The variables in the dataset are likely to contain details about the location, wind direction and speed, pressure, and surface moisture flux. The data also include information on the cyclone’s timing, length, and impact on coastal economy and communities. The dataset’s structure is likely in the form of a time-series, with each row representing a single observation of a cyclone at a specific moment. Most likely, the observations are arranged in chronological order, with the earliest observations coming first and the most recent observations coming last in the dataset. Very likely a large and varied collection of data regarding cyclones in the North Indian Ocean region makes up the dataset’s content. The information could be used to look for patterns and connections between different meteorological traits and cyclone intensity and course. Also, it can be used to evaluate the socioeconomic effects of cyclones on coastal economies and communities. 3.3 Feature Selection Choosing features is an important step in any machine-learning activity, especially when working with large and complex datasets. In the context of cyclone prediction, feature selection is crucial in defining the model’s accuracy and efficiency. We will explore the feature selection process for cyclone prediction using different methods. This study’s dataset contains four main features: VMAX, MSLP, CFLX, and TWAC. VMAX is the maximum sustained wind speed, MSLP is the minimum central pressure, CFLX is the net heat exchange of the cyclonic storm with the ocean, and TWAC is the total water vapor transport into the storm. These four features were selected because they are the main determinants of tropical storm behavior and intensity. We seek to uncover the most relevant and critical features that have a significant influence on cyclone prediction throughout the feature selection phase. To do this, we employed both the Pearson correlation coefficient and SelectKBest method from sklearn, a python based machine learning framework to assess the relevance of each characteristic. The relevance of a feature is measured by its ability to predict the target variable using each feature. The greater the feature priority score, the more vital the feature. The results of the feature selection procedure revealed that VMAX was selected as the most essential feature by both the Pearson correlation coefficient and SelectKBest module. This suggests that the maximum sustained wind speed is the most important feature in determining tropical storm strength and behavior followed by MSLP, CFLX, and TWAC. This finding is consistent with prior research that demonstrated a substantial link between wind speed and tropical storm intensity.
472
S. Rahman et al.
3.4 LightGBM Algorithm LightGBM is a framework for gradient boosting that use tree-based learning techniques. [9] It is optimized for large-scale, high-dimensional data and is meant to be efficient and scalable. LightGBM’s main concept is to develop a model by iteratively adding decision trees to capture complicated, non-linear connections in data. The approach fits a decision tree to the negative gradient of a loss function, which evaluates the difference between predicted and actual values for the target variable, at each iteration. Each tree’s output is then integrated to generate a final forecast for the target variable. The tree-building approach, which employs a histogram-based representation of the data to effectively split the feature space and limit the number of candidates for split points, lies at the heart of LightGBM’s efficiency. This histogram-based approach also enables LightGBM to efficiently and robustly handle categorical features and missing data. LightGBM uses the gradient boosting framework to minimize a loss function, L(y,f(x)), where y is the goal variable and f(x) represents the prediction provided by the model. The purpose of gradient boosting is to obtain the best prediction function, f(x), by including decision trees into the model, T1 , T2_ ,…,Tm , so that: m f (x) = f (x)Ti (x) i=1
where each decision tree, Ti , is fit to the negative gradient of the loss function. Ti (x) = −αi hi (x) where αi and hi (x) are the learning rate and decision tree fit to the negative gradient of the loss function, respectively. 3.5 Train with LightGBM We can train a LightGBM or Random Forest model to learn this mapping by minimizing a loss function that measures the discrepancy between the predicted categories yj and the true categories yj for the training set. We can define a loss function L as: n 1 L Y,Y = l yi , yi n k=1
where L is a loss function that measures the discrepancy between the true category yi and the predicted category yj for a single cyclone i. The goal is to find a set of model parameters that minimize this loss function, which can be done using an optimization algorithm such as gradient descent. Once the model is trained, we can use it to make predictions on new, unseen cyclones by applying the learned function f to their input features. Figure 1 provides us with the overview of our working Model.
Tropical Cyclone Genesis Forecasting Using LightGBM
473
Fig. 1. Working Diagram of Proposed Model
4 Result and Discussion In this study, LightGBM outperformed Random Forest in terms of classification accuracy, precision, recall, and f1-score. LightGBM achieved an accuracy of 1.00, precision of 1.00, recall of 1.00, and f1-score of 1.00, while Random Forest had an accuracy of 0.94, precision of 0.92, recall of 0.93, and f1-score of 0.92. These results suggest that LightGBM may be a better choice for this classification task. Using LightGBM we were able to predict all the six cyclone categories correctly. From the confusion matrix in Fig. 5 and 6 we observe that LightGBM predicted category 0 as 156 out of 156. On the other hand, Random forest was able to predict category 0 as a number of 107 out of 156 in the provided dataset and the same is applicable for all other categories as well. In Fig. 2, the plot displays the variation of multi_logloss over the course of the training process, with the number of iterations on the x-axis and the loss on the y-axis. Both the train loss and the validation loss exhibit a steady decrease from an initial value of 1.2 to a final value of 0.1, within a span of 100 iterations. This trend suggests that the LightGBM algorithm is effective at reducing the multi_logloss metric for the training and validation datasets, implying that the model has been trained to generalize well and perform accurately on unseen data. In this study, we analyzed the performance of the model in terms of the train score and cross-validation score as the training set size increased. In Fig. 3, the train score, which represents the model’s accuracy on the training set, began at 0.860 and gradually increased as the training set size increased from 600 to 1000. This indicates that the model can learn more from the data as the size of the training set grows. Additionally, the cross-validation score, which represents the model’s accuracy on a validation set,
474
S. Rahman et al.
increased from 0.860 to 0.920 as the training set size increased from 600 to 1000. This suggests that the model is able to generalize better to new data as the training set size increases. LightGBM and Random Forest are both commonly used algorithms for classification tasks, and they both have advantages and disadvantages. LightGBM outperformed Random Forest in this study, implying that it may be a better choice for similar classification tasks, at least under the conditions of this study. The accuracy, precision, recall, and f1-score of the models were used to assess their performance. The percentage of correct predictions made by the model is known as accuracy, while the percentage of correct positive predictions is known as precision, and the percentage of correctly identified positive examples is known as recall. The F1-score is the harmonic mean of precision and recall. According to the study’s findings, LightGBM was more effective than Random Forest at correctly recognizing both positive and negative situations, with higher values for accuracy, precision, and recall. Furthermore, LightGBM outperformed Random Forest in probability prediction.. Overall, the results indicate that LightGBM is a more precise and accurate algorithm for this specific classification task, with superior ability to differentiate between positive and negative examples and forecast probability. However, keep in mind that the algorithm of choice can vary depending on the specifics of the data and the task at hand. This is why it’s always a good idea to compare multiple models and choose the one that best serves our needs. Table 1 depicts the comparison results between LightGBM and Random Forest in terms of accuracy, precision, recall, and f1-score. As the problem is based on classification tasks, we used cross-entropy as the loss function. Figure 2 and 3 shows us the pictorial view of training and validation loss with learning curve of LightGBM algorithm respectively. Figure 4 resembles the root of the tree that represents the entire dataset, and the tree branches out from there. A decision point based on a particular feature value is represented by each split. The final prediction or choice is made at each leaf node, or the end of the tree branch. The majority class of the training instances that reached that leaf node provides the basis for the prediction at each leaf node. Table 1. Performance metrics of two machine learning model Ml Models
Accuracy
Precision
Recall
f1-Score
LightGBM
1.00
1.00
1.00
1.00
Random Forest
0.94
0.92
0.93
0.92
The confusion matrices from the two models, LightGBM and Random Forest, are shown in Figs. 5 and 6, respectively. Through comparing the instance of the models, it is observed that the LIghtGBM model performs better on the specified categorization job. Additionally, accuracy, precision, recall, and F1 score are metrics that can be obtained
Tropical Cyclone Genesis Forecasting Using LightGBM
475
Fig. 2. Train and validation loss
Fig. 3. Learning Curve LightGBM
Fig. 4. Tree Generated by LIghtGBM
from the confusion matrix and used to assess how well the model performs in terms of true positives, true negatives, false positives, and false negatives.
476
S. Rahman et al.
Fig. 5. Confusion Matrix for LightGBM
Fig. 6. Confusion Matrix for RF
5 Conclusion The research focused on cyclone categorization, which is critical for forecasting their potential impact and minimizing damage. By comparing the performance of two machine learning algorithms, LightGBM and Random Forest, the study discovered that LightGBM outperformed Random Forest in terms of accuracy and precision in this specific task. This finding is significant because it suggests that LightGBM may be a better choice for similar classification tasks, at least under the conditions of this study. From a scientific standpoint, this study contributes to the field of cyclone research by providing a new approach to categorizing cyclones that may be more accurate and effective than existing methods. This is important because accurately categorizing cyclones can help emergency responders and disaster management teams prepare for and respond to cyclone-related events more effectively.
Tropical Cyclone Genesis Forecasting Using LightGBM
477
In terms of future research, there are several avenues that could be explored. One possibility is to investigate the use of other machine learning algorithms or ensembles to further improve cyclone categorization. Additionally, with the potential impacts of climate change on cyclones, it would be interesting to explore how the categorization of cyclones may change in the future and how machine learning algorithms can be adapted to these changes. Finally, incorporating real-time data could improve the accuracy and timeliness of cyclone categorization, which would have important implications for emergency response and disaster management. Acknowledgment. This research is supported by Military Institute of Science and Technology, Dhaka, Bangladesh.
References 1. https://public.wmo.int/en/our-mandate/focus-areas/natural-hazards-and-disaster-risk-reduct ion/tropical-cyclones 2. https://www.nhc.noaa.gov/aboutsshws.php 3. Classified Early Warning and Forecast of Severe Convective Weather Based on LightGBM Algorithm. written by Xinwei Liu, Haixia Duan, Wubin Huang, Runxia Guo, Bolong Duan, published by Atmospheric and Climate Sciences 11(2) (2021) 4. Tang, R., Ning, Y., Li, C., Feng, W., Chen, Y., Xie, X.: Numerical forecast correction of temperature and wind using a single station single time spatial LightGBMMethod. Sensors 22, 193 (2022).https://doi.org/10.3390/s22010193 5. Pokhrel, P., Ioup, E., Hoque, M., Abdelguerfi, M., Simeonov, J.: A LightGBM based Forecasting of Dominant Wave Periods in Oceanic Waters (2021) 6. Wang, P., Wang, P., Wang, D., Xue, B.: A conformal regressor with random forests for tropical cyclone intensity estimation. IEEE Trans. Geoscience Remote Sensing, p. 1 (2021). https:// doi.org/10.1109/TGRS.2021.3139930 7. Hill, A., Herman, G., Schumacher, R.: Forecasting severe weather with random forests. Mon. Weather Rev. 148, 2135–2161 (2020). https://doi.org/10.1175/MWR-D-19-0344.1 8. Meenal, R., Michael, P., Pamela, D., Rajasekaran, E.: Weather prediction using random forest machine learning model. Indonesian J. Electrical Eng. Comput. Sci. 22, 1208 (2021). https:// doi.org/10.11591/ijeecs.v22.i2.pp1208-1215 9. Ke, G., et al.: Lightgbm: A highly efficient gradient boosting decision tree. Advances in Neural Inf. Processing Syst. 30 (2017) 10. Nti, I., Nyarko-Boateng, O., Aning, J.: Performance of machine learning algorithms with different K values in K-fold cross-validation. Int. J. Inf. Technol. Comput. Sci. 6, 61–71 (2021). https://doi.org/10.5815/ijitcs.2021.06.05 11. Saber, M., et al.: Examining lightGBM and catboost models for wadi flash flood susceptibility prediction Geocarto Int. 37 (2021). https://doi.org/10.1080/10106049.2021.1974959 12. Meng, F., et al.: IEEE international geoscience and remote sensing symposium IGARSS. Brussels, Belgium 2021, 8476–8479 (2021). https://doi.org/10.1109/IGARSS47720.2021. 9555156 13. Eisenstein, L., Schulz, B., Qadir, G.A., Pinto, J.G., Knippertz, P.: Identification of high-wind features within extratropical cyclones using a probabilistic random forest – Part 1: method and case studies. Weather Clim. Dynam. 3, 1157–1182 (2022) 14. Acula, D.: Classification of disaster risks in the philippines using adaptive boosting algorithm with decision trees and support vector machine as based estimators. J. Modeling Simulation Materials 4, 7–18 (2021). https://doi.org/10.21467/jmsm.4.1.7-18
Optimization of Identification and Recognition of Micro-objects Based on the Use of Specific Image Characteristics Isroil I. Jumanov and Rustam A. Safarov(B) Samarkand State University, Samarkand 140104, Uzbekistan [email protected]
Abstract. Models and algorithms for optimizing the identification and recognition of micro-objects have been developed based on the combination of dynamic models with neural networks and mechanisms for extracting statistical, dynamic, specific characteristics of images, extracting and segmenting the contour, selecting reference points and reducing redundant points, as well as setting model variables. Identification optimization mechanisms are implemented that use the correlation and dynamic characteristics of a sequence of points on the image contour. A dynamic model has been implemented that performs the functions of forming a matrix with the coordinates of distorted points, approximating image segments, taking into account its deformation. The effectiveness of image pre-processing tools, recognition and classification algorithms was studied using examples of a large amount of images presented about wheat pollen grains. When solving the problems of identification and recognition of pollen grains, the mechanism of non-linearity of the influence of factors, as well as the conditions of a priori insufficiency and uncertainty, were used. A software package for visualization, recognition, classification of micro-objects based on the use of an interpolation spline-function 7, a three-layer neural network with learning algorithms for forward and back propagation of errors, training with and without a teacher, mechanisms for forming a training sample with vector quantization, segmentation, and the formation of a “sliding window”» and tracking control points along the contour of the image segment. Keywords: Micro-object · Identification · Recognition · Classifications · Detection of a Distorted Point · Characteristics and Properties · Image
1 Instruction Research and development of methods for identification, recognition and classification of micro-objects are relevant and in demand in the fields of palynology, medicine, environmental protection and ecology, etc. [1, 3]. Solutions to the problem of recognition and classification of micro-objects are being carried out by specialists from a number of foreign countries such as England, France, Germany, Italy, Spain, Austria, New Zealand, etc. [2, 6]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 478–487, 2023. https://doi.org/10.1007/978-3-031-36115-9_44
Optimization of Identification and Recognition
479
Recognition and classification systems have been created to solve the problems of identifying and counting pollen grains in air samples [4, 8]. Nineteen European countries are connected to the EAN/EPI European Airborne Allergy Network, which collects information on pollen. Pollen scientists currently operate only with the available conventional standards [3, 5, 6]. However, the standards used are not sufficient to take them as a basis for research on the creation of an automated image visualization system for recognition, classification and accounting of micro-objects [7–9]. Recognition and classification of micro-objects are preceded by the tasks of removing foreign particles, etc. [10–12]. The problem of detecting and correcting distorted image points due to interference, noise, and blur is becoming increasingly important [13–15]. In classical algorithms for identifying micro-objects, little attention is paid to the use of mechanisms for detecting distorted points in image identification [16–18]. It is necessary to improve and develop methods for identifying micro-objects based on the use of mechanisms for detecting and correcting distorted points against the background of noise in conditions of a priori insufficiency, uncertainty and low data reliability [19–21]. This study is devoted to the creation and application of microobject image identifiers, which use a multilayer neural network, radial-basic networks, as well as mechanisms for extracting redundant information structures - histological, morphological, fractal, geometric and other specific characteristics of images [22–24].
2 Methodology for Identification and Recognition of Microobjects Based on the Use of Redundant Information Structures Let the input of the mechanism of identification, recognition and classification of microobjects be given the implementation of a signal with noise, the distortion of the points of the image in which is given in the form [25, 26] ξ (t) = Qs(t) + n(t), where s(t) - a signal of a known form; n(t) - Gaussian noise with a spectral density of N0 ; E = T T 2 2 N0 0 ξ (t)s(t)dt = 0 s (t)dt - spectrum power, ξ (t) = s(t); Q - the value that takes the value “1” if the distorted point is present, “0” if the distorted point is absent. The control of distorted points on the image contour is evaluated by the criterion of maximum √ = 1 − F 2z , where z = 2E/N0 - signal-tolikelihood [24, 27] Perr (z) = α(z)+β(z) 2 noise ratio. If the sequence of image points is characterized by stationary behavior, then the problem of detecting distorted points is not difficult. If the sequence of image points is characterized by a dynamic and non-stationary process, then adaptive control mechanisms are required that use the main characteristics of the signals. The simulation of the process was carried out using multi-stage correlators combined with the NN, which are aimed at performing mathematical operations similar to the operations of algorithms for the statistical detection of distorted image points.
480
I. I. Jumanov and R. A. Safarov
In a modified version of this mechanism, the synaptic weight coefficient of neurons wi is used. The weighted sum of neuron N weights of the correlator built on the basis of a multilayer NN is given as U = i=0 wi ξi , where ξi - input realization counts; N - number of inputs. At the second stage of operation of the mechanisms, the sum of the products of the input implementation ξi and its weight coefficients wi is compared with a given threshold. The term “0” of the coefficient and the imaginary “1” input are distinguished. A weighting factor with a zero index acts as a decision threshold, the value of which is compared with the weighted sum. Weight coefficients are used for fixing and identifying “reference” (informative) points of micro-object images. An algorithm for learning a supervised NN is studied, at the output of which the network reactions are determined in the form wi = wi +η(d − y)ξi , where η - coefficient selected in [0, 1]; d - expected network reactions [25, 28]. Implementations of the correlator based on the multilayer NN have been tested in the MATLAB environment. 10 NNs are investigated, each having 11 inputs, the weight coefficients are set randomly. The coefficients are normalized with respect to w 0. 10 1 The weight coefficients are calculated by wi cp = 10 j=1 wij , where wij is the weight coefficient of the j-th network (Fig. 1).
Fig. 1. Weight coefficients of the NN correlator of the image signal
It is determined that the required stability of the neural network learning algorithm is achieved when noise with a power 5–7 times less than the power of the signal itself is superimposed on the training sample. The dependencies of the probability of detecting distorted points on the parameter signal-to-noise for all given three signals are investigated. Estimates of the error function of the neural network correlator of image signals are obtained e(U ) = d − y(U ) = d − H (U ),
; H (U ) - Heaviside function.
Optimization of Identification and Recognition
481
Figure 2 shows the dependence of the probability of detecting distorted points for signals 1 (Fig. 2. a)) and signal 2 (Fig. 2. b)): 1 - solid line, corresponds to the traditional algorithm; 2 – dash-dotted line, corresponds to the algorithm of the modified mechanism.
Fig. 2. Distorted Point Detection Efficiency
It is determined that the value of the probability of detecting distorted image points varies depending on the training steps of the NN. The error function H (U ) has a discontinuity of the first kind at the threshold point of the neuronal activation function. It is determined that it is impossible for the weighting coefficients to achieve an exact repetition of the waveform. Graphs of the probability of detecting distorted points depending on the parameter signal/noise almost converge with the curve of the modified mechanism. The efficiency of the mechanism was analyzed by the maximum likelihood criterion. The main features of the mechanism are superimposed on training samples, size variations that depend on the initial values of the NN weights.
3 Optimal Image Identification with Detection and Correction of Distorted Points Based on Wavelet Filtering To improve the processes of detecting and correcting distorted image points, it is proposed to use a generalized filtering mechanism based on wavelet-transform algorithms and a multi-parameter threshold function under the influence of additive noise. The useful signal summation mechanism S(t) with noise n(t) includes sampling, binarization, coding, and wavelet filtering procedures. At its output, a filtered signal S (t) is issued. The wavelet function in a certain finite time and frequency domain is represented by an infinitely oscillating function, which distinguishes this approach from the classical Fourier transform and encoded information is used. Another feature of the mechanism is that it works with orthogonal functions and has the ability to use a smaller number of expansion coefficients. Analysis of the results of the implementation of such a mechanism allows us to establish the following positive aspects of the study: – a greater gain in the speed of information processing is achieved with less time spent;
482
I. I. Jumanov and R. A. Safarov
– Discrete Wavelet Transformer (DWT) is obtained from a continuous time scaled with a factor of a and a factor of b, like a = 2m , b = k · 2m , where m and k are some integers; – the continuous plane a and b is divided into a certain grid and the DWT characterizes the shift of the next function in time t−b 1 1 = √ ψ 2−m t − k (1) ψmk (t) = √ ψ m a a 2 +∞ Direct DWT has the form cmk = (S(t) · ψmk (t)) = −∞ S(t)· ψmk (t)dt, where cmk conversion coefficients. The inverse DWT is defined as S(t) = m,k cmk · ψmk (t). Implementations of the generalized mechanism have been tested. Signal S(t) with an amplitude of Am , with a duration of Ts is considered. Analytical expressions for them are given in the form: Am (1 − exp(−αt)), at 0 ≤ t ≤ Ts /2; (2) S(t) = Am (1 − exp(α(t − Ts ))), at Ts /2 ≤ t ≤ Ts . For DWT, it is necessary to represent a continuous signal S(t) in the form of sample Sd (t) = {Si }, where i = 0, 1, ..., N − 1 with a frequency N −1 of fm and with time steps of s(it)δ(t − it). t = 2f1m . The discrete signal is given as Sd (t) = i=1 The efficiency of the E(.) - fractal-correlation functional of the correlator of the identification mechanism is studied based on the following calculation E(x, y, n) = (1 − (Dx + Dy )/d (n))1/3 ≈ r(x, y), where E(.)- fractal-correlation functional of the correlator; d (n) - scaling factor depending on the size of the test sample -n; Dx , Dy sums of modules of the difference of adjacent data x and y. Table 1. Results of the fractal-correlation functional n
d(n)
σ(Fr)
σ(r)
6
5.05
0.513
0.365
8
5.61
0.492
0.360
10
6.12
0.501
0.349
12
6.75
0.452
0.345
14
7.16
0.471
0.317
16
7.52
0.487
0.314
18
8.43
0.462
0.295
Table 1 shows the values of normalizing coefficients and standard deviations for fractal-correlation functionals, with different test sample sizes and zero correlation of the tested data pairs. Standard deviations of the distribution of errors, obtained by calculations using the standard formula. It has been established that DWT allows one to obtain effective filtering results in an area with distorted image points.
Optimization of Identification and Recognition
483
The efficiency of the generalized mechanism for selecting objects (points, features, fragments) belonging to rectangular areas with the procedures of wave segmentation and image identification has been studied. Each point p of the original image I is taken as the origin of the local coordinate system and it is checked whether its value belongs to a rectangular area. For this, a rectangular area dp centered at the selected point is defined. Area dp is characterized by a set width of one raster unit, a given length of l, and a rotation angle of α between dp and the selected x-axis. The rotation angle α is changed in the range of values [0; π ) in increments of φ. The length l of area dp is selected taking into account the scale of the original image I . Thus, dp is a rectangular segment in the original image I. The mean-square error (MSE) of identification, represented by a function of f (.) points belonging to dp of all possible dp relative to point p, was studied. If at a certain angle of rotation, dp takes a positive value, then with a certain probability it can be argued that point p belongs to the rectangular area of the image. The mean-square error (MSE) of identification, represented by a function of f (.) points belonging to dp of all possible dp relative to point p, was studied. The uniformity of the distribution of the brightness function in the area of the extension and the sharpness of the contours of the object is checked.
4 Analysis of the Results of the Study Symmetric spectral density functions are implemented, in which dp is measured along the abscissa. Its calculation is made for the first half dp , which includes points with positive offsets. And the second half is dp with negative offsets of points along the ordinate axis. For each value of the rotation angle α of area dp , the first and second half dp , respectively, σ1 and σ2 are calculated. Of all the values σ1 , the minimum σ1 min and the corresponding value of the rotation angle α1 are determined. The average values of α2 , and a2 and mean square deviations (MSD) of σ 1 , and σ2 min are calculated. The belonging of points to a rectangular area is checked as (|σ1 − σ1 min | > Tp ) ∧ (|σ2 − σ2 min | > Tp ). Threshold Tp is set by the operator, taking into account the brightness of the dots and the contrast of the original image. The uniformity of the brightness distribution function on a rectangular area of the image is checked as (σ1 min → 0) ∧ (σ2 min → 0). An additional check is due to the division of each half of area dp in the form (|σ1 min − σ2 min | → 0) ∧ (α1 ≈ α2 ). The MSD value must correspond to one position dp or an adjacent position, which is performed at a low value of φ of the angle of rotation. The minimum values of the MSD brightness of the points correspond to the position of area dp , coinciding with the stretch of the object. The presence of clear indices of image points reflects the transition of the brightness level of the points on the border of the fragments of the object. If at a certain value of the rotation angle α, area dp is a spread, then the brightness value of point dp will become high. If at point p it takes a positive value, then for p the values σmin and the corresponding rotation angle ασ are calculated. Approximately the
484
I. I. Jumanov and R. A. Safarov
same value of angles α1 and α2 are replaced by a threshold value of |α1 − α2 | < Tα , where Tα is a threshold value that depends on the rotation angle step. When area dp is a straight line segment, then it becomes necessary to calculate the position coordinates and brightness values of points in area dp at the nodes of the discrete grid. To calculate the MSD, it is necessary to determine the brightness f (.) at points belonging to area dp centered at p. The decompression of a compressed grayscale image is also performed in two parts: the first, a two-level contour of the image, and the second, the wavelet-transform coefficients, the inverse discrete wavelet-transform of the masked image. The masked halftone image was reconstructed. A generalized mechanism for selecting image points, checking whether they belong to rectangular areas allows you to detect distorted points. The following designations have been applied: – X = {Xi } is a set of image point vectors, each of which is given as Xi =< xij >=< xi1 , ..., xiN >, where i = 1, ..., M ; – M - power set X , j = 1, ..., N = 5, and j = 1 which corresponds to the value of the concentration (optical density) of the background component of the image: j = 2, j = 3, j =
4; – A = Ap - set of standards Ap images p = 1, ...Np .
To check the quality of image identification, the following are set: A = Apg is a set of standards of gradations Apg in each of the fragments Ap of the image, Ng p where Apg = N1g i=1 xig is the standard of the g-th gradation of the fragment Ap ; Ng - number of historical data, g = 1, ..., 7. The cross-correlation coefficient ρxy is defined as the ratio of the covariance cov(Xi , Yi ) of two vectors Xi and Yi . The i , Yi ) means μx and μy , as well as the RMS σx and σy , are given as: ρxy = cov(X σx ·σy ; cov(Xi , Yi ) = 1n ni=1 (xi − μx )(yi − μy ), where x and y is respectively xkj and ylj . The values of membership functions of image points are calculated by ρ(A, B) =
n
|μA (xi ) − μB (xi )|
(3)
i=1
xkj − xlj , where xkj and xlj are compared The distance is interpreted as hkl = analyzes of parameters k and l, i = 1, ..., M , n = M . The software package (PC) for visualization, identification, recognition and classification of images of micro-objects includes the following functional modules: detection, selection of contours in the image, segmentation of the contour, determination of the boundaries of segments; morphological processing of the contour of the image, dilatation of the contour two-level image; compression of one-dimensional, two-level contours of the image and layout of the compressed halftone image file; reconstruction of the masked halftone image based on the scanned masked image using a contour image; contour bilinear interpolation with a bidirectional weighting of control points; masking of image contour points, line-by-line representation, and interpolation based on wavelet transformations and biorthogonal polynomials.
Optimization of Identification and Recognition
485
Compressed halftone image files are generated in two parts. The first, is a twolevel contour of the image, and the second, is wavelet coefficients - transformations. Functional modules are tested in a Matlab environment.
5 Conclusion A methodology has been developed for optimizing identification, recognizing the classification of micro-objects based on combining dynamic models with neural networks, extracting statistical, dynamic, and specific characteristics. Methods and algorithms use adaptive correlators, synthesized with a Fourier transform, a Gaussian filter, mechanisms for detecting and correcting distorted points on the contour of an image of a micro-object and allow you to design transparent software tools that complement existing standards for pollen recognition in palynology, environmental protection and others. A software package based on the use of an interpolation spline - function 7, a threelayer neural network, algorithms for direct and inverse learning, learning with and without a teacher has been implemented. The identification mechanisms are modified based on the synthesis of algorithms for forming a training sample with vector quantization, segmentation, the formation of a “sliding window” and tracking of control points. It is proved that when solving the problems of selection and seed production of wheat grain, medical diagnosis of patients with tuberculosis, the effectiveness of the implemented models for identifying micro-objects increases based on the mechanisms of using histological, morphological, fractal, geometric characteristics. The required stability of the neural network learning algorithm is achieved when noise with a power of 5–7 times less than the power of the signal itself is applied to the training sample, as well as with mechanisms for detecting and correcting distorted points, filtering based on wavelet transforms and a multi-parameter threshold function.
References 1. Gonzalez, R., Woods, R.: Digital Image Processing, M.: Technosfera. (2005) 2. Kuleshov, S.V., Aksenov, Y.A., Zaitseva, A.A.: An approach to identifying the source of images from digital cameras. Innovative Sci. 5, 82–86 (2015) 3. Jumanov, I.I., Djumanov, O.I., Safarov, R.A.: Methodology of optimization of identification of the contour and brightness-color picture of images of micro-objects. International Russian Automation Conference, pp. 190–195 (2021). https://doi.org/10.1109/RusAutoCon52 004.2021.9537567 4. Khaikin, S.: Neural Networks: Full Course, p. 1104. Williams, Moscow (2006) 5. Philist, S.A., Tomakova, R.A., Zar, D.Y.: Universal network models for problems of classification of biomedical data, SWGU. Part 2, 4(43), 44–50 (2012) 6. Jumanov, I., Safarov, R.: Optimization of the processing of images of pollen grains based on the use of their specific characteristics and geometric features. AIP Conference Proceedings, 2637, 040016 (2022). https://doi.org/10.1063/5.0118857 7. Kohonen, T.: Self-Organizing Maps. Kohonen, T. BINOM. Knowledge Lab, p. 655 (2013)
486
I. I. Jumanov and R. A. Safarov
8. Jumanov, I.I., Djumanov, O.I., Safarov, R.A.: Optimization of identification of images of micro-objects taking into account systematic error based on neural networks. International Russian Automation Conference, pp. 626–631 (2020). https://doi.org/10.1109/RusAutoCo n49822.2020.9208164 9. Osovsky, S.: Neural Networks for Information Processing. Osovsky, S. M.: Finance and statistics, p. 344 (2002) 10. Jumanov, I.I., Djumanov, O.I., Safarov, R.A.: Recognition of micro-objects with adaptive models of image processing in a parallel computing environment. Journal of Physics: Conference Series, (1791). Omsk, Russia (2020). https://doi.org/10.1088/1742-6596/1791/1/ 012099 11. Bouguet, J.Y.: Pyramidal implementation of the Lucas Kanade feature tracker. Intel Corporation, Microprocessor Research Labs 9 (2000) 12. Jumanov, I.I., Djumanov, O.I., Safarov, R.A.: Mechanisms for optimizing the error control of micro-object images based on hybrid neural network models. AIP Conf. Proc. 2402, 030018 (2021). https://doi.org/10.1063/5.0074019 13. Bovik, A.C., Clark, M., Geisler, W.S.: Multichannel texture analysis using localized spatial filters. IEEE Trans. PAMI. 12(1), 55–73 (1990) 14. Dash, R., Majhi, B.: Motion blur parameters estimation for image restoration. Optik 125(5), 1634–1640 (2014) 15. Ibragimovich, J.I., Isroilovich, D.O., Abdullayevich, S.R.: Optimization of identification of micro-objects based on the use of characteristics of images and properties of models. International Conference on Information Science and Communications Technologies, ICISCT, 9351483 (2020). https://doi.org/10.1109/ICISCT50599.2020.9351483 16. Dunn, D., Higgins, W.E.: Optimal Gabor filters for texture segmentation. IEEE Trans. Image Processing 4, 947–964 (1995) 17. Haralick, R.M., Shapiro, L.G.: Computer and Robot Vision. 2. Addison-Wesley, Reading, MA (1993) 18. Jumanov, I., Djumanov, O., Safarov, R.: Improving the quality of identification and filtering of micro-object images based on neural networks. E3S Web of Conferences, 304, 01007 (2021). https://doi.org/10.1051/e3sconf/202130401007 19. Hoang, M.A., Gcuscbrock, J.M., Smoulders, A.W.M.: Color texture measurement and segmentation. Signal Process 85(2), 265–275 (2005) 20. Maltoni, D., Maio, D., Jain, A.K., Prabhakar, S.: Handbook of Fingerprint Recognition. Springer-Verlag, N. Y. 496 (2009). 21. Ibragimovich, J.I., Isroilovich, D.O., Abdullayevich, S.R.: Recognition and classification of pollen grains based on the use of statistical, dynamic image characteristics, and unique properties of neural networks. Adv. Intell. Syst. Comput. 1323, 170–179 (2021). https://doi. org/10.1007/978-3-030-68004-6_22 22. Sakthi Bharathi, D., Manimegalai, A.: 3D digital reconstruction of brain tumor from MRI scans using Delaunay triangulation and patches. ARPN J. Eng. Applied Sci. 10(20), 9227– 9232 (2015) 23. Singh, M., Kumar, S., Singh, S., Shrivastava, M.: Various image compression techniques: lossy and lossless. Int. J. Comput. Appl. 142(6), 23–26 (2016) 24. Jumanov, I.I., Safarov, R.A., Djumanov, O.I.: Mechanisms for using image properties and neural networks in identification of micro-objects. In: The 16th IEEE International Conference, Application of Information and Communication Technologies, 22541907.| Washington DC (2022). https://doi.org/10.1109/AICT55583.2022.10013633 25. Jumanov, I.I., Safarov, R.A.: Optimization of recognition of micro-objects based on reducing excessive information structures of images. J. Phys.: Conf. Ser. 2373, 072030 (2022). https:// doi.org/10.1088/1742-6596/2373/7/072030
Optimization of Identification and Recognition
487
26. Bramesh, S.M., Anil Kumar, K.M.: An efficient and scalable technique for clustering comorbidity patterns of diabetic patients from clinical datasets. I.J. Modern Education Comput. Sci. 14(6), 35–52 (2022) 27. Panda, M.: Developing an efficient text pre-processing method with sparse generative naive bayes for text mining. Modern Educ. Comput. Sci. 10(9), 11-19 (2018) 28. Jadhav, A.N., Dharwadkar, N.V.: A speaker recognition system using gaussian mixture model, EM algorithm and K-means clustering. I.J. Modern Educ. Comput. Sci. 10(11), 19–28 (2018)
Development and Comparative Analysis of Path Loss Models Using Hybrid Wavelet-Genetic Algorithm Approach Ikechi Risi1 , Clement Ogbonda2 , and Isabona Joseph3(B) 1 Department of Physics, River State University, Port Harcourt, Nigeria 2 Department of Physics, Ignatius Ajuru University of Education, Port Harcourt, Nigeria 3 Department of Physics, Federal University, Lokoja PMB 1154, Nigeria
[email protected]
Abstracts. The received signal strength of any cellular network system at the user equipment terminal is dependent on the propagation environment and the path loss model used during the network planning stage. One of the most commonly used empirical model for predictive path loss analysis and estimation is the COST231 model. Despite the popularity of this model, its precision performance is usually poor when engaged for propagation prediction modelling outside the intended environment wherein it was originally developed. To cater for such limitation, an adaptive path loss model for optimal path loss prediction is developed in this paper using a hybrid Wavelet-Genetic genetic algorithm (Wavelet-GA). For the purpose of comparative analysis between wavelet-GA, GA, and COST231 models, three evaluation indicators based on root mean square error (RMSE), mean absolute error (MAE), correlation coefficient (R), were engaged. The results showed that the estimated RMSEs attained with the proposed wavelet-GA are 2.896, 4.715, 1.945, and 3.498, whereas those of GA and COST231 models are 3.47, 5.49, 3.69, 4.55 and 78.13, 74.74, 84.30, 76.54 respectively. Also, the estimated MAE of wavelet-GA, GA, and COST231 models are 2.38, 3.92, 1.45, 2.95; 2.83, 4.66, 2.84, 3.84 and 78.04, 74.37, 84.36, 76.39 respectively. The analyzed results also showed that wavelet-GA, GA, and COST231 models correlate with the measured data by 93.3%, 90.8%, 90.8%; 83.9%, 79.6%, 79.6%; 92.1%, 76.9%, 76.9%; and 86.3%, 78.4%, 78.4% for site 1,2,3 and 4 respectively. Again, the analysed validation results also proved that the developed wavelet-GA model is 99.2% and suitable to be applied in Awka, Nigeria. The developed wavelet-GA based path loss model has the capacity to care of future planning and development cellular wireless systems in similar environments. Keywords: Wavelet · Genetic Algorithm · Hybrid Wavelet-GA · Path loss · COST231 model
1 Instruction Predicting the path loss has been a crucial task in the design and planning of mobile communication networks. This is as a result of various physical mechanisms such as reflection, diffraction and scattering of the signal during propagation, as well as the © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 488–500, 2023. https://doi.org/10.1007/978-3-031-36115-9_45
Development and Comparative Analysis of Path Loss Models
489
phenomenon of multipath and the constant movement of mobile communication users [1]. Therefore, the prediction of the electromagnetic signal power loss with minimal error during signal propagation from transmitter to receiver is an important topic in planning and optimization of telecommunications networks. When deploying a new mobile technology such as a Long Term Evolution (LTE) network with the goal of increasing network capacity and speed with broader coverage and strong Quality of Service (QoS), it is important to identify these factors that affect the network quality and capacity of the mobile network [2]. Signal path loss estimation is an influential factor in network planning because it allows network engineers to perform various configuration tests before the changes are physically implemented. However, path loss prediction for radio coverage is a complex task, hence the need for an accurate and computationally efficient prediction tool [3]. Accurate and reliable models are crucially important for the prediction of radio channel parameters in the cellular network systems. Path loss models are generally the empirical mathematical formulation of the signal propagation behavior of an environment. Empirical models essentially explain the relationship between environment and path loss. It does not require large amounts of computation and it’s easy to implement, but are less sensitive to the physical and geometric configuration of the environment. The need for proper cellular network planning, determination of base stations (BS) locations and proper operating frequency during upgrade or deployment of a new communication network such as LTE to ensure improved QoS is stressed in many previous works [4–13]. However, understanding and building these networks relies on the knowledge of signal path loss over distance in a pragmatic environment. Path loss is one of the most important characteristics of communication networks. It is the received signal power relative to transmitted [14]. The drop in electromagnetic signal power density as it travels through space can be as a result of diffraction, reflection, scattering, etc., and depends on different environments [15]. Factors such as topography, urban planning, population density, rainfall, vegetation, etc. contribute to path loss. Variations in height of the transmitter and receiver antennas also create losses. An accurate estimate of the path loss provides a good basis for correct BS positioning and appropriate frequency plan determination [13]. There is also a need for network engineers to have a suitable method for mapping the extent of coverage of both existing and planned networks.
2 Literature Review Eichie et al. (2017) [16], developed ANN model that estimated the path loss within suburban and rural paths in Minna. Different parameters such as transmitter and receiver antennas distance, transmitter and receiver powers, and the potential height of the measurement locations. Comparative analysis were made between path loss estimated through the developed ANN model achieved by geometric algorithm and path loss calculated from existing predicted models among the existing predicted models considered, Hata model proved the best achievement along the rural paths where the RMSE estimated varied from 5.05 to 9.30 dB, whereas, in the suburban paths, the RMSE calculated were
490
I. Risi et al.
from 9.50 to 82.14 dB. The best performed existing predicted model in the suburban paths was Egli’s model with an estimated RMSE that ranged from 3.81 to 8.18 dB. However, Ericsson and COST 2.31 models showed that highest path loss calculated in the rural and suburban paths. The ANN paths loss model developed, showed better performance than the existing predicted models in the environments were the RMSE achieved ranged from 3.96 to 7.07 dB and 1.22 to 4.82 dB in rural and suburban paths respectively. The formulated ANN path loss model proved to be suitable for the estimation at GSM signals in both rural and suburban areas within Minna and its surroundings. The authors in [17, 18], analyzed and presented commonly used propagation loss models. The analyzed and simulated results were performed to identify the propagation loss by varying base transmitter station antenna height, mobile station antenna height and distance between transmitter and receiver. The relationship between propagation loss and other wireless propagation parameters, such as transmitter–receiver antenna spacing, antenna heights, and operating frequency were presented to enhance cellular network performance optimization. The data provided were analyzed in the MATLAB software to predict signal propagation losses; radio coverage estimation; avoid interference; and determine the received power level. The propagation loss decreases due to the increase in BTS antenna height for all models. These interpretations obviously showed the impact of propagation loss on cellular network. The models selected simulated were; freespace, Okumura, Okumura-Hata, and COST-231 propagation loss models in suburban areas. The demonstrations apparently confirmed that the Okumura model performed better than Okumura-Hata and COST-231 in terms of propagation loss reduction. The Okumura model could be used to study 5G radio network planning in cellular network. 2.1 COST231 Model COST231 model is an enhancement of Okumura-Hata model and it is mostly used for path loss determination and can be applicable in range of frequency between 1.8 GHz to 2 GHz. PL (dB) = 46.3 + 33.9 log(f ) − 13.82 log(ht ) − α(hr ) + 44.9 − 6.55 log(ht ) log(d ) + Cm where F is the transmitting frequency. ht is the transmitter height. hr is the receiver height. d is the distance 2 α(hr ) = 3.2 log(11.75hm ) − 4.97 for large cities, f ≥ 300 MHz 2 α(hr ) = 8.29 log(1.54hm ) − 1.1 for large cities, f < 300 MHz α(hr ) = (1.1 log f − 0.7)hm − (1.56 log f − 0.8) for medium to small cities 0 dB; medium sized cities and suburban areas Cm = 3 dB; metropolitan areas
(1)
Development and Comparative Analysis of Path Loss Models
491
3 Methodology In this research, the locations in which field works are conducted were first visited to survey the area (that is, have good knowledge of the terrain) and know which the drive test routes to take when collecting signal strength data. It also provided means of knowing the number of eNodeBs antennas the LTE cellular systems network provider have in the study locations (Fig. 1).
Fig. 1. The flowchart of path loss models development
3.1 Method of Data Collection With the aid of professional and reliable TEMS equipped labtop and TEMS pocket Sony Ericson phone with map info software, field measurements were conducted over a popular commercial LTE cellular networks air interface, propagating on the 2600 MHz band in Port Harcourt, in River State and Awka in Anambra State, Nigeria. The building clusters in the area are a mixture of residential/commercial bungalows, two-story or three-story buildings encompassed with medium density user and vehicular traffics as earlier described. Precisely, the measurement routes were selected along the main streets and sideway of the roads of the area, where the LTE eNodeB transceivers are deployed. One of the main LTE radio networks data collected during measurement is RSRP (i.e., Reference Signal received Power) data. Technically, the RSRP is an indicator of signal power level at the UE terminal in LTE networks. Generally, the stronger RSRP level received at UE, better signal coverage quality can be achieved in the radio network. There exist sundry factors that can impact the RSRP levels at the UE terminals, among which are transmitter–receiver (Tx-Rx) communication distance, RF channel conditions, signal propagation loss, UE location, total radiated eNodeB power, etc.
492
I. Risi et al.
3.2 Proposed Hybrid Wavelet-GA Method Contained in Fig. 1 is the summarized flowchart employed to develop and implement the proposed hybrid Wavelet-GA method. It started by conducting field signal data ass described above. This followed by routing the measured signals through wavelet processing platform. The resultant processed signal is then engaged to calculated the propagation loss, which in turn supplied to GA optimization scheme to tune the the existing log-distance model in correspondence with measured loss data. The log-distance loss model is defined as [6]. PL (dB) = a1 + a2 log f + a3 log d
(2)
where a1 , a1 and a1 are the parameters to be optimized using the proposed wavelet-GA method.
4 Results Table 1 is the measured RSRP from four commercial cell sites deployed in Awka and Table 2 contain the developed mean wavelet-GA and GA path loss models of the four cell sites. Table 1. Measured Signal Strength from Four Sites in Awka eNodeB Site 1 Dist (m) RSRP(dBm)
eNodeB Site 2 Dist (m) RSRP(dBm)
eNodeB Site 3 Dist (m) RSRP(dBm)
eNodeB Site 4 Dist (m) RSRP(dBm)
100
−76.01
100
−80.19
200
−82.44
100
−72.3
120
−74.31
130
−76.81
220
−74.38
120
−71.21
140
−75.24
160
−71.81
240
−71.81
130
−73.33
150
−69.5
190
−75.38
260
−71.94
150
−81.13
160
−64.56
200
−77.88
270
−73.38
170
−76.56
170
−71.38
230
−73.63
300
−72.63
190
−76.38
180
−76.63
250
−70.81
310
−74.13
200
−76.75
190
−72.81
270
−77
320
−75
210
−77.94
200
−81.69
300
−78.94
340
−76
220
−85.25
210
−77.81
320
−80.69
360
−77.31
230
−85.06
220
−77.94
340
−82.81
380
−74.94
260
−82.44
230
−74.19
360
−80.13
400
−75.38
280
−81.13
250
−78.38
390
−81.56
410
−79.69
300
−79.25
260
−77.94
400
−84.56
420
−84.25
310
−77.69
280
−77.31
410
−86
440
−85.25
340
−73.56
300
−77.44
420
−91.63
460
−90
360
−77.5
(continued)
Development and Comparative Analysis of Path Loss Models
493
Table 1. (continued) eNodeB Site 1 Dist (m) RSRP(dBm)
eNodeB Site 2 Dist (m) RSRP(dBm)
eNodeB Site 3 Dist (m) RSRP(dBm)
eNodeB Site 4 Dist (m) RSRP(dBm)
310
−76.56
440
−90.38
480
−78.06
400
−86.44
320
−77.56
460
−95.88
500
−80.69
420
−93.25
330
−74.63
480
−87.38
520
−79.56
450
−92.19
350
−83.88
490
−91.69
540
−86.69
480
−94.06
360
−78.88
500
−87.38
560
−80.94
500
−86.13
380
−84.25
510
−97.06
570
−76.94
520
−92.38
400
−89.88
520
−101.06
590
−81.94
550
−85.75
420
−91.44
540
-96.25
600
−86.81
570
−87.19
440
−87.75
550
−97.81
610
−90.31
600
−97.06
460
−86.5
560
−93.88
630
−83.69
620
−90.56
480
−87.75
580
−90.13
660
−84.31
640
−91.88
500
−91.25
600
−94
680
−79.19
670
−96.52
530
−88.44
609
−78.13
700
−81.44
700
−95.69
550
−84.44
647
−77.25
720
−83.94
720
−94.32
570
−85.06
685
−78.31
740
−84.38
760
−86.56
600
−86.94
724
−77.88
760
−84.19
780
−90.5
610
−87.5
762
−18.31
780
−82.44
800
−87.25
620
−90.19
792
−78.31
800
−89.81
830
−86.31
630
−91.38
832
−79.13
820
−85.88
860
−87.19
660
−90.13
870
−79.19
840
−88.25
900
−87.75
670
−86.63
900
−80.06
860
−93.94
904
−70.88
700
−88.88
929
−79.00
880
−83
908
−68.69
710
−92.44
959
−75.44
900
−85.63
913
−70.88
The propagation loss data used in this paper were obtained from the measured RSRP values by subtracting the total transmit power of the various eNodeB antennas from the measured RSRP values [17, 18] (Figs. 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13 and 14). Table 2. The Developed Propagation loss Model using the proposed hybrid Wavelet-GA approach and the existing standard GA approach Cell sites
Proposed Hybrid Wavelet-GA Model
Standard GA Model
Awka 1
−8.14 + 24.46log(d) + 24.56log(f)
9.92 + 26.38log(d) + 117.89log(f)
Awka 2
−2.59 + 30.0log(d) + 19.64 log(f)
−0.59 + 23.63log(d) + 23.83 log(f)
Awka 3
4.98 + 26.45log(d) + 17.74log(f)
-3.45 + 23.79log(d) + 22.29 log(f)
Awka 4
0.48 + 21.53log(d) + 24.68log(f)
9.81 + 21.81log(d) + 21.74log(f)
Mean
-1.32 + 25.6 log(d) + 21.66og(f)
3.92 + 23.27log(d) + 21.36 log(f)
494
I. Risi et al.
Fig. 2. Predicted RMSE Difference using hybrid Wavelet-GA model, GA Model, and COST231Hata, in eNode Site 1 of Awka
Fig. 3. Predicted RMSE Difference using hybrid Wavelet-GA model, GA Model, and COST231Hata, in eNode Site 2 of Awka
Fig. 4. Predicted RMSE Difference using hybrid Wavelet-GA model, GA Model, and COST231Hata, in eNode Site 3 of Location 2
Development and Comparative Analysis of Path Loss Models
495
Fig. 5. Predicted RMSE Difference using hybrid Wavelet-GA model, GA Model, and COST231Hata, in eNode Site 4 of Awka
Fig. 6. Predicted MAE difference using the hybrid Wavelet-GA method, GA Method, and the COST231-Hata model in Site 1 of Location 2
Fig. 7. Predicted MAE difference using the hybrid Wavelet-GA method, GA Method, and the COST231-Hata model in Site 2 of Awka
496
I. Risi et al.
Fig. 8. Predicted MAE difference using the hybrid Wavelet-GA method, GA Method, and the COST231-Hata model in Site 3 of Location2
Fig. 9. Predicted MAE difference using the hybrid Wavelet-GA method, GA Method, and the COST231-Hata model in Site 4 of Awka
Fig. 10. Correlation of wavelet-GA, GA and COST231-Hata models to measured data in Site 1 of Awka
Development and Comparative Analysis of Path Loss Models
497
Fig. 11. Correlation of wavelet-GA, GA and COST231-Hata models to measured data in Site 2 of Awka
Fig. 12. Correlation of wavelet-GA, GA and COST231-Hata models to measured data in Site 3 of Awka
Fig. 13. Correlation of wavelet-GA, GA and COST231-Hata models to measured data in Site 4 of Awka
498
I. Risi et al.
Fig. 14. Validation of proposed hybrid wavelet-GA model using another Site in Awka
5 Discussion COST231 model attained path loss values of 220-250 dB, 210-248 dB, 210-238 dB, 220241 dB as compared to measured path loss whose values are 138-184 dB, 120-156 dB, 126-155 dB, and 130-142 dB, respectively. From the results, it is noticeably clear that path loss values attained by the COST231 model is quite too high than the ones obtained through field measurements. The higher path loss values produced by the existing model may be ascribed to the physical terrain and the topographical differences between locations where the measurement loss data were conducted or the terrain characterization where the COST231 model was developed. Thus, the needs to fine tune the existing model to reliably fit the measured path loss data. The results attained by applying measured path loss with hybrid Wavelet-GA, GA model achieved without denoising the signal, and COST 231 model are provided. The performance of the proposed hybrid wavelet-GA model over the standard approaches are evaluated and revealed using three key performance indicators. The indicators include RMSE, MAE, and correlation coefficient (R). The figures display path loss prediction accuracies attained using the Wavelet-GA optimization model in comparison with the ones attained using COST 231 model and GA. From the results, the proposed adaptive hybrid wavelet-GA model attained better prediction performance with lower RMSE values, which are 2.89 dB, 4.71 dB, 1.94 dB, and 3.49 dB. But for the developed GA optimization model, without denoising, the measured signal, the prediction errors are quite higher in terms of RMSE values. The RMSE values are 3.93 dB, 4.62 dB, 4.56 dB and 7.31 dB. The analysed result showed that COST231 model has the highest RMSEs. The values of COST231 model are 78.13 dB, 74.74 dB, 84.30 dB, 76.54 dB respectively. The enhanced path loss prediction performance attained using the proposed wavelet-GA model over standard models are also presented in the figures and tables respectively, using MAE, and Coefficient of correlation (R) performance. It showed that the developed hybrid wavelet-GA method proved to estimate the lowest MAE, whereas the COST 231 model achieved the highest value.
Development and Comparative Analysis of Path Loss Models
499
The MAE achieved are 2.38 dB, 3.92 dB, 1.45 dB, and 2.958 dB. The COST231 model achieved the highest values of MAE as 78.04 dB, 74.37 dB, 84.36 dB, 76.39 dB respectively. Lastly, the measured path loss where also compared with the developed hybrid wavelet-GA model, GA model, and COST231 model in terms of the correlation coefficient (R). It showed that the developed hybrid wavelet-GA model achieved the highest values of R, which means that there is correlation between the measured path loss and the developed hybrid wavelet-GA optimization model. The values of R on wavelet-GA are 93.32%, 83.93%, 92.07% for site 1, 3, 4 of respectively. To validate a developed path loss model means to test the model prediction performance using another cell site data other than the one used for wavelet-GA model developed. Thus, validation provides means of assessing the performance of any newly developed path loss model in another eNodeB cell in order to establish its prediction dynamism and efficacy. The respective mean resultant developed model using the proposed Wavelet-GA. It showed that the validation prediction performances of proposed Wavelet-GA model 0.99 (99%) correlate with the measured data and as such valid. This again validates the efficacy of the hybrid signal path loss predictive modeling approach, which is proposed in this paper.
6 Conclusion The proposed adaptive hybrid path loss prediction modeling performance is compared with the standard genetic algorithm and other theoretical methods in literature such as COST 231 model. The results show that the proposed adaptive hybrid path loss prediction modeling method gives better prediction accuracy in terms of RMSE, MAE, and correlation coefficient (R). For instance, the proposed Wavelet-GA optimization model attained better prediction performance with lower RMSE values of 2.89 dB, 4.71 dB, 1.94 dB, and 3.49 dB. But for standard GA optimization model, the prediction errors are quite higher in terms of RMSE values, which are 3.93 dB, 4.62 dB, 4.56 dB and 7.31 dB. From the results, the proposed Wavelet-GA model for better path loss prediction over other existing techniques is clearly justified. Particularly, the ability of wavelet transform to extract both local spectral and temporal real signal information component from the noisy one assisted the GA to adaptive model and predicts the signal path loss values more effectively, than using only the GA optimization model. The power of genetic algorithms comes from their ability to combine both exploration and exploitation in an optimal way.
References 1. Japertas, S., Grimaila, V:. Mobile signal path losses in microcells behind buildings. Radio Eng. 26(1), 191–197 (2017). https://doi.org/10.13164/re.2017.019. 2. Coinchon, M., Salovaara, A. P., Wagen, J. F.: The impact of radio propagation predictions on urban UMTS planning. In: 2002 International Zurich Seminar on Broadband Communications Access - Transmission - Networking: Meeting the Challenge of High-Speed Communications, IZS 2002 - Proceedings, vol. 1, no. 1, pp. 321–326 (2001). https://doi.org/10.1109/IZSBC. 2002.991775
500
I. Risi et al.
3. Nationale, E., Sa-bedja, K., Nationale, E.: Propagation models calibration in mobile cellular networks : a case study in togo. Future Technollogies Conf. 2(11), 923–927 (2017) 4. Joseph, I., Divine, O.O.: Application of Levenberg-Marguardt algorithm for prime radio propagation wave attenuation modelling in typical urban, suburban and rural terrains. Int.J. Intell. Syst. Appl. 7, 35–42 (202) 5. Ojuh, D.O., Isabona, J.: Field electromagnetic strength variability measurement and adaptive prognostic approximation with weighed least regression approach in the Ultra-high radio frequency band. J. Intell. Syst. Appl. 2021(4), 14–23 (2021) 6. Isabona, J., Ojuh, D.O.: Adaptation of propagation model parameters toward efficient cellular network planning using robust LAD algorithm. Int.J. Wireless Microwave Technol. 5, 13–24 (2020) 7. Isabona J., Konyeha C.C.: Urban area path loss propagation prediction and optimisation using Hata Model at 800MHz. IOSR J. Appl. Phys. (IOSR-JAP) 3 (4), 8–18, 2013. 8. Isabona, J., Kehinde, R.: Multi-resolution based discrete wavelet transform for enhanced signal coverage processing and prediction analysis. FUDMA J. Sci. 3(1), 6–15 (2019) 9. Ebhota, V.C., Isabona, J., Srivastava, V.M.: Environment-Adaptation Based Hybrid Neural Network Predictor for Signal Propagation Loss Prediction in Cluttered and Open Urban Microcells. Wireless Pers. Commun. 104(3), 935–948 (2019) 10. Isabona, J., Imoize, A.L.: Terrain-based adaption of propagation model loss parameters using Non-linear square Regression. J. Eng. Appl. Sci. 68(1), 1–19 (2021). https://doi.org/10.1186/ s44147-021-00035-7 11. Almalki, F. A., Sciences, P.: Optimisation of a Propagation Model for Last Mile Connectivity with low Altitude Platforms Using Machine Learning Brunel University London, 1–129 (2017) 12. Oguejiofor, O.S.: Pathloss Prediction for a typical mobile communication system in Nigeria using empirical models. Int. J. Comput. Netw. Wireless Commun. 3(2), 207–211 (2015) 13. Khare, A., Saxena, M., Tiwari, S.: Multimedia networks based dynamic WCDMA system proposal for QoS. Int. J. Eng. Adv. Technol. 1(1), 52–55 (2011) 14. Ebhota, C., Isabona, J., Srivastava, V.M.: Improved adaptive signal power loss prediction using combined vector statistics based smoothing and neural network approach. Progress Electromagnet. Res. C 82(1), 155–169 (2018). https://doi.org/10.2528/pierc18011203 15. Jakborvornphan, S.: Analysis of path loss propagation models In mobile communication. J. Theor. Appl. Inf. Technol. 98(4), 725–730 (2020) 16. Echie, J.O., Oyedum, O.D., Ajewole, M.O., Aibinu, A.M.: Comparative analysis of basic models and artificial neural network based model for path loss prediction. Progress Electromagnet. Res. 61(9), 133–146 (2017) 17. Isabona, J., Zhimwang, J.T., Risi, I.: Cascade forward neural networks-based adaptive model for real-time adaptive learning of stochastic signal power datasets. Int. J. Comput. Netw. Inf. Secur. 2(3), 63–74 (2022). https://doi.org/10.5815/ijcnis.2022.03.05 18. Joseph, I., Ituabhor, O., Timothy, J., zhimwang, Risi Ikechi,: Achievable throughput over mMobile broadband network protocol layers: practical measurements and performance analysis. Int. J. Adv. Netw. Appl. 13(04), 5037–5044 (2022). https://doi.org/10.35444/IJANA. 2022.13404
Mathematical Advances and Modeling in Logistics Engineering
Risk Assessment of Navigation Cost in Flood Season of the Upper Reaches of the Yangtze River Based on Entropy Weight Extension Decision Model Jun Yuan1,2(B) , Peilin Zhang1 , Lulu Wang2 , and Yao Zhang3 1 School of Transportation and Logistics Engineering, Wuhan University of Technology,
Wuhan 430063, China [email protected] 2 School of Logistics, Wuhan Technology and Business University, Wuhan 430065, China 3 Technical University of Munich Asia Campus, Technical University of Munich (TUM), Singapore, Singapore
Abstract. Inland navigation is greatly affected by climate and hydrological changes, of which the flood and dry season have the greatest impact on inland navigation. Changes in water level and flow rate during the flood and dry season will have various impacts on such aspects as navigation speed, navigation ship loading rate, navigation time efficiency, etc., many of which will cause changes in navigation costs. Therefore, how to predict and evaluate the cost risk of navigation during the flood season has become an important topic. This paper takes the navigation cost risk in the flood period of the upper reaches of the Yangtze River as the research object, establishes the navigation cost risk assessment index system in the flood period of the upper reaches of the Yangtze River and the navigation cost risk assessment model in the flood period of the upper reaches of the Yangtze River based on the entropy weight extension decision model. On this basis, the feasibility of risk assessment and prevention of navigation cost in flood season of inland river shipping is discussed by using the case of relevant projects to test the evaluation model. Keywords: Inland River shipping · Risk assessment · Entropy weight extension model
1 Introduction Inland river shipping is one of the important components of the national comprehensive transportation system, which has the advantages of large transportation capacity, small land occupation, small energy consumption and light pollution. Vigorously developing inland shipping is of great significance to the reduction of comprehensive logistics costs and the improvement of logistics benefits, and thus to the improvement of the overall national economic level. As the most important part of China’s inland river shipping system, Yangtze River shipping is of self-evident importance [1, 2]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 503–513, 2023. https://doi.org/10.1007/978-3-031-36115-9_46
504
J. Yuan et al.
Inland navigation is greatly affected by climate and hydrological changes, of which the flood and dry season have the greatest impact on inland navigation. Changes in water level and flow rate during the flood and dry season will have various impacts on such aspects as navigation speed, navigation ship loading rate, navigation time efficiency, etc., many of which will cause changes in navigation costs. Transportation cost is the most important factor in the production and operation of inland shipping enterprises. It will directly determine the benefits of shipping enterprises and then affect the benefits of the entire industry. Therefore, how to predict and evaluate the cost risk of navigation during the flood season has become an important topic [3–5]. This paper takes the navigation cost risk in the flood period of the upper reaches of the Yangtze River as the research object, establishes the navigation cost risk assessment index system in the flood period of the upper reaches of the Yangtze River and the navigation cost risk assessment model in the flood period of the upper reaches of the Yangtze River based on the entropy weight extension decision model. On this basis, the feasibility of risk assessment and prevention of navigation cost in flood season of inland river shipping is discussed by using the case of relevant projects to test the evaluation model.
2 Risk Assessment Index System of Navigation Cost in Flood Season in the Upper Reaches of the Yangtze River 2.1 Impact of Flood Period on the Navigation Cost of Ships in the Upper Reaches of the Yangtze River The upstream channel of the Yangtze River is a typical inland channel. To evaluate the navigation cost risk of the upstream of the Yangtze River during the flood period, it is necessary to establish the corresponding evaluation index system. Before that, first analyze the impact of the flood period on the navigation cost of the upstream of the Yangtze River [6, 7]. The flood period of the inland waterway shows the characteristics of the overall rise of the water level of the waterway, the sharp daily fluctuation, and the increase of the flow velocity. According to this analysis, the main impact of the flood period on the navigation cost of the upper reaches of the Yangtze River is as follows. 2.1.1 Impact on Navigation Speed of Navigable Ships When the flood season comes, the water level in the inland waterway rises, the river flow and flow rate increase in varying degrees, the speed of downstream ships increases, and the speed of upstream ships decreases. Since the upper reaches of the Yangtze River are mainly downstream channels, the shipping speed of ships in the upper reaches of the Yangtze River increases during the flood season, which improves the efficiency of ship operation and has a certain impact on the navigation cost [8, 9]. 2.1.2 Impact on the Loading Rate of Navigable Ships The ship loading rate is related to the ship type and channel water depth, of which large ships are greatly affected by the channel water depth, and small and medium-sized ships
Risk Assessment of Navigation Cost in Flood Season of the Upper Reaches
505
are relatively small. During flood period, the water level in the upper reaches of the Yangtze River changes sharply and rises as a whole. The changes in ship loading rate caused by these water level changes will have an indirect impact on ship navigation costs [10, 11]. 2.1.3 Impact on Navigation Obstruction Time Due to the changes in the navigation environment of the channel caused by the changes in the flood season, a large number of ships may be blocked and the waiting time may occur. The time cost caused by the blocking and waiting time will affect the overall cost of ship navigation [12, 13]. 2.1.4 Other Impacts Due to the special navigation environment during the flood season, it may cause certain additional losses to navigation ships. In order to adapt to the navigation environment in flood season, ships need to additionally improve the navigation adaptability of ships in flood season, which will cause changes in navigation costs [14, 15]. 2.2 The Risk Assessment System of Ship Navigation Cost in Flood Season in the Upper Reaches of the Yangtze River According to the above analysis, the factors that affect the risk of ship navigation costs in the flood season in the upper reaches of the Yangtze River can be divided into five categories: Factories of Ship, Factories of Channel, Factories of Crew, Factories of Natural Environment, Factories of Relevant Department management [16]. Factors of Ship includes: Index1 Tonnage of ship; Index2 Age of ship; Index3 Navigation adaptability of ships in flood season. Factors of Channel includes: Index4 Affluence depth of channel; Index5 Width of channel; Index6 Navigation density of channel; Index7 Perfection of channel supporting facilities. Factors of Crew includes: Index8 Basic professional level of crew; Index9 Safety navigation training level of crew; Index10 Crew’s awareness of safe navigation; Index11 Average age of crew. Factors of Natural environmental includes: Index12 Flow velocity of navigable channel; Index13 Wind speed of navigable channel; Index14 Visibility of navigation channel. Factors of Department management includes: Index15 Management level of maritime departments; Index16 Management level of channel departments; Index17 Management level of shipping company [17]. The Risk assessment index system of navigation cost in flood season in the upper reaches of the Yangtze Rivers is showed as Table 1.
506
J. Yuan et al.
Table 1. The Risk assessment index system of navigation cost in flood season in the upper reaches of the Yangtze River Factor type
Index list
Factors of Ship
Index1 Tonnage of ship Index2 Age of ship Index3 Navigation adaptability of ships in flood season
Factors of Channel
Index4 Affluence depth of channel Index5 Width of channel Index6 Navigation density of channel Index7 Perfection of channel supporting facilities
Factors of Crew
Index8 Basic professional level of crew Index9 Safety navigation training level of crew Index10 Crew’s awareness of safe navigation Index11 Average age of crew
Factors of Natural environmental
Index12 Flow velocity of navigable channel Index13 Wind speed of navigable channel Index14 Visibility of navigation channel
Factors of Management
Index15 Management level of maritime departments Index16 Management level of channel departments Index17 Management level of shipping company
3 Extension Evaluation Model of Risk Entropy Weight for Navigation Cost in Flood Season in the Upper Reaches of the Yangtze River Based on the construction of the evaluation index system of navigation cost risk in flood period in the upper reaches of the Yangtze River, the entropy weight extension evaluation model of navigation cost risk in flood period in the upper reaches of the Yangtze River is constructed as follows [18–21]: 3.1 Evaluation Object, Classical Domain and Section Domain of the Model According to the basic theory of the entropy weight extension decision model, the matrix of the composite matter element Rxi in the classical domain of the navigation cost evaluation model in the flood season of the upper reaches of the Yangtze River is: ⎤ ⎡ Nxj , c1 , vxj1 ⎢ ⎥ .. (1) Rxj = Nxj , ci , vxji = ⎣ ⎦ . cn , vxjn Assume that there are a total of i(i = 1, 2, · · · , n) indicators in the navigation cost risk assessment system in the upper reaches of the Yangtze River during the flood period,
Risk Assessment of Navigation Cost in Flood Season of the Upper Reaches
507
and there are a total of j(j = 1, 2, · · · , m) categories of risk ratings of the assessment objects. The navigation cost risk is a matter element Rxj representing the navigation cost risk in the flood period of the upper reaches of the Yangtze River; Nxj represents the j risk rating in the navigation cost risk assessment system of the upper reaches of the Yangtze River during Ci represents the evaluation index i, and its classical domain flood season; is vxyi = axjn , bxjn . Construct the corresponding node Rp Rp ⊃ Rj for the above classical domain, Assume that the value range of the ith characteristic ci of matter-element Rp is api , bpi . Then Rp is recorded as: ⎤ ⎤ ⎡ P, c1 , ap1 , bp1 P, c1 , vp1 ⎥ ⎢ .. .. ⎥ = ⎢ .. . Rp = (p0 , ci , vi ) = ⎣ . . ⎦ ⎣ . .. ⎦ cn vpn cn apn , bpn ⎡
(2)
The matter element of the risk assessment object of navigation cost in flood period of the upper reaches of the Yangtze River to be assessed is recorded as Rx , Let the actual value of the ith index ci of matter element Rx be vi . Then Rp is recorded as: ⎤ px , c1 , v1 ⎢ .. .. ⎥ Rx = (px , ci , v) = ⎣ . . ⎦ ⎡
(3)
cn vn
3.2 Correlation Function and Correlation Degree of the Model According to the basic theory of the entropy weight extension decision model, the correlation between the ith evaluation index of the evaluation object matter-element R and the jth rating grade of the entropy weight extension evaluation model of the navigation cost risk in the flood period in the upper reaches of the Yangtze River is: ⎧ γ (vi ,vxji ) ⎨ ,v ∈ / vxji γ (vi ,vpi )−γ (vi ,vxji ) i (4) Kxj (vi ) = γ v ,v ( ) i xji ⎩− ∈ v , v i xji a −b | xji xji | In this formula:
1 1
γ vi , vxji = vi − axji + bxji
− bxji − axji 2 2
1 1 γ vi , vpi =
vi − api + bpi
− bpi − api 2 2
(5) (6)
508
J. Yuan et al.
3.3 Comprehensive Correlation Degree and Risk Level of the Model Assume that the correlation degree between the evaluation index vi of the subject matter element Rx and the risk rating grade j is Kxj (vi ), then take the weighted value of the correlation degree between each index and each evaluation grade as the comprehensive correlation degree Kxj (Rx ), and record it as: Kxj (Rx ) =
n
ωi Kxj (vi )
(7)
i=1
Calculate the maximum value of the comprehensive correlation degree Kxj (Rx ) between all the material elements Rx and the evaluation level to determine the final cost risk value level of the current case. The calculation formula is expressed as follows: Kxj (Rx ) =
max
j=1,2,··· ,m
Kxj (Rx )
(8)
4 Test of Navigation Cost Risk Assessment Model in Flood Period of Upper Yangtze River Based on Entropy Weight Extension Decision Model 4.1 Data Source of Test Case Based on relevant projects, this paper studies the risk of ship navigation cost of a ship company A passing through a section B in the upper reaches of the Yangtze River during flood season. The project investigated the natural environment of the relevant channel, the basic condition of the channel, the basic condition of the navigable ships and the crew. The investigation method is field investigation, and the relevant information is collected by inviting the management personnel of relevant channel management departments, shipping companies and port companies to have a discussion. The information involves all indicators in the navigation cost risk assessment system of the upper reaches of the Yangtze River in flood season in this paper. 4.2 Optimization of Evaluation Index Date The data collected in this paper relate to the parameters of the ship, the age and training status of the crew, the weather data of the route section, the basic condition of the channel, etc., while 8 of which are qualitative indicators and 9 are quantitative indicators, and the actual value is directly taken. Invite relevant water transport cost risk assessment experts and internal management personnel of the shipping company to form a 10person assessment team to score each risk assessment index in turn. The score adopts a five-point system. The higher the score, the less likely the index will cause cost risk. According to the characteristics of inland river shipping, this paper divides the risk of navigation costs in the flood season of the upper reaches of the Yangtze River into five levels, which are respectively recorded as high risk (Level A), high risk (Level B), general risk (Level C), low risk (Level D), and very low risk (Level E), corresponding to expert scores of 1, 2, 3, 4, and 5.
Risk Assessment of Navigation Cost in Flood Season of the Upper Reaches
509
4.3 Determination of Case Evaluation Index Weight In risk assessment research, there are many methods to determine the weight of evaluation indicators. In order to conform to the characteristics of entropy weight extension evaluation research, this paper adopts the information entropy method to determine the weight of various indicators. The specific calculation method is as follows: If the evaluation index system has i index and j evaluation levels, the evaluation matrix is: ⎤ ⎡ X11 · · · X1y ⎥ ⎢ (9) X = ⎣ ... ... ... ⎦ Xi · · · Xij
The calculation formula of entropy H (Xn ) of the index η (n = 1, 2, · · · , i) is H (Xn ) = −
i
piy × ln piy
(10)
n=1
In this formula: piy =
Xij y
, i = 1, 2, · · · , n
(11)
Xiy
x=1
The calculation formula of the difference coefficient h of the index is hn = 1 − H (Xn )
(12)
Then the calculation formula of the corresponding index weight ωn is ωn =
hn i hn
(13)
n=1
According to the above formula and the collected data, it can be calculated that the weight of the cost risk assessment index of the navigation section B of the ship company A during the flood period is showed as Table 2. Table 2. The Weight of case navigation cost risk assessment index Index
1
2
3
4
5
6
7
8
9
Weight
0.071
0.067
0.061
0.052
0.044
0.048
0.086
0.072
0.054
Index
10
11
12
13
14
15
16
17
Weight
0.058
0.038
0.066
0.062
0.064
0.044
0.051
0.062
510
J. Yuan et al.
4.4 Determine of the Classical Domain and Matter Element to be Evaluated According to the constructed entropy weight extension evaluation model of navigation cost risk in the flood period of the upper reaches of the Yangtze River, the classical domain included in the evaluation system in this case is:
A Tonnage of ship (0, 1)
.. .. (14) Rxd = . .
Management level of shipping company (0, 1) RxB
RxC
RxD
RxE
B Tonnage of ship
.. = .
Management level of shipping company
C Tonnage of ship
.. = .
Management level of shipping company
D Tonnage of ship
.. = .
Management level of shipping company
E Tonnage of ship
.. = .
Management level of shipping company
The object element to be evaluated in this case is ⎡ Tonnage of ship, px , ⎢ .. Rx = (px , cn , vn ) = ⎣ .
(1, 2)
.. .
(1, 2)
(15)
(2, 3)
.. .
(2, 3)
(16)
(3, 4)
.. .
(3, 4)
(17)
(4, 5)
.. .
(4, 5)
(18)
⎤ v1 .. ⎥ . ⎦
(19)
Management level of shipping company vn According to the data brought in by formulas (4), (5), (6) and (7), it can be calculated that the comprehensive correlation between all indicators of the case and all evaluation levels is showed as Table 3. Table 3. Comprehensive correlation between each evaluation index and each risk level of the case Risk value
1
2
3
Index 1
−0.0395
−0.5245
−0.0563
4
Index 2
−0.2523
−0.0656
Index 3
−0.0026
−0.0876
5 0.0556
−0.3151
−0.0875
−0.2351
−0.8622
0.0862
−0.1253
0.0314
(continued)
Risk Assessment of Navigation Cost in Flood Season of the Upper Reaches
511
Table 3. (continued) Risk value
1
2
3
4
5
Index 4
−0.0532
−0.3251
−0.0723
−0.0321
−0.0882
Index 5
−0.0499
0.0209
−0.0351
−0.0831
−0.0542
Index 6
−0.0323
0.0624
−0.0251
−0.0885
−0.0231
Index 7
−0.0872
−0.0731
−0.0031
−0.0724
−0.0312
Index 8
−0.0621
0.0213
−0.0241
−0.0631
−0.0821
Index 9
−0.0012
−0.0063
−0.0321
−0.0516
−0.0851
Index 10
0.0231
−0.0826
−0.0351
−0.0420
−0.0671
Index 11
−0.0301
−0.0821
−0.0842
−0.0316
−0.0696
Index 12
−0.0216
0.0251
−0.0061
−0.0698
−0.0286
Index 13
−0.0761
−0.0071
−0.0761
−0.0781
−0.0862
Index 14
−0.0871
−0.0316
−0.0224
−0.0361
−0.0372
Index 15
−0.0051
−0.0871
−0.0774
−0.0765
−0.0351
Index 16
0.0324
−0.0368
−0.0875
−0.0614
−0.0881
Index 17
−0.0031
−0.3159
−0.0616
−0.0375
−0.0135
According to the above table, the maximum correlation degree max Kxj (c1 ) = 0.0556 between index1 (Tonnage of ship) and each risk level can be calculated, Calculate the risk grade correlation degree and final evaluation grade of all indicators is showed as Table 4. Table 4. Maximum correlation degree and final risk level between case evaluation index and each risk level Index
1
2
3
4
5
6
7
8
max Kxj (cn )
−0.0395
−0.0655
0.0862
−0.0321
0.0209
0.0624
−0.0031
0.0213
−0.0012
Risk level
4
2
3
4
2
2
3
2
1
17
Index
10
11
12
13
14
15
16
max Kxj (cn )
0.0231
−0.0301
0.0251
−0.0071
−0.0224
−0.0051
0.0324
−0.0031
Risk level
1
1
2
2
3
1
1
1
9
In this case, evaluation indicators 1 and 4 are at high risk; Evaluation indicators 3, 7 and 14 are at general risk; Evaluation indicators 2, 5, 6, 8, 12 and 13 are at low risk; The evaluation indicators 9, 10, 11, 15, 16 and 17 are at very low risk. Analyzing the distribution of main risk levels in this case, we find that the assessment index 1, ship tonnage, and the assessment index 4, channel affluence risk is at a higher risk level. The indicators 1 and 2 are mainly related to the uncertainty risk caused by the increase of water flow speed due to the rise of water level during the flood period, which needs to be improved. The evaluation index 3 Navigation adaptability of ships in flood season, index 7 Perfection of channel supporting facilities, and index 14 Visibility of
512
J. Yuan et al.
navigation channel are at general risk. Although there is no major risk for the time being, it still needs the attention of relevant shipping companies and management departments. For index 3, shipping companies can improve the ship’s navigation adaptability according to the navigation characteristics in flood season; For index 7, the channel management department should strengthen the construction of channel supporting facilities to reduce the probability of navigation risk accidents in flood season; For index 14, it is necessary for the shipping company and the channel management department to take measures to deal with the low visibility of the navigable river section in advance. Other indicators are at low risk or very low risk, and there is no need for obvious improvement for the time being.
5 Conclusion Based on the analysis of the main factors affecting the navigation cost in the flood period of the upper reaches of the Yangtze River, this paper constructs the navigation risk evaluation index system in the flood period of the upper reaches of the Yangtze River, and establishes the entropy weight extension evaluation model of the navigation cost in the flood period of the upper reaches of the Yangtze River. Using this model, the cost risk of a ship company’s navigation in a certain section of the upper reaches of the Yangtze River during the flood period is evaluated, and the main indicator factors that may cause cost risk in this case are evaluated, and the corresponding improvement suggestions are given. The navigation risk assessment model of the upper reaches of the Yangtze River constructed in this paper can also be applied to the assessment of navigation risk in dry season or the navigation risk in the upper, middle and lower reaches of other inland waterways after modifying the scoring criteria of some evaluation indicators. For example, the impact of ship tonnage on the cost in dry season is just opposite to that in flood season, and the relevant research is not repeated due to space limitation. The evaluation model built in this paper has certain pertinence. If it needs to be widely used, it needs to further optimize the evaluation indicators and models, and relevant research needs to be further carried out in the future.
References 1. Huang, Q.: Yangtze river shipping development report 2020. Changjiang Navigation Administration, pp. 6364 (2021, in Chinese) 2. Liu, N.: Theory, method and application of transportation project operation cost evaluation. Shanxi People’s Publishing House, pp. 1821 (2003, in Chinese) 3. Lei, X., Wu, Y., Ye, S.: Regional food security early warning based on entropy weight extension decision model. J. Agri. Eng. 28(5), 233-239 (2012, in Chinese) 4. Shi, K., Sun, X.: Evaluation of Water Resources Carrying Capacity in Chongqing Three Gorges Reservoir Area Based on Entropy Weight Extension Decision Model 33(2), 609-616 (2013, in Chinese) 5. Hu, E.: Practical techniques and methods for environmental risk assessment. Beijing: China Environmental Science Press, pp. 3334 (2000, in Chinese)
Risk Assessment of Navigation Cost in Flood Season of the Upper Reaches
513
6. Peng, Z., Perch, S.: Transportation and Climate Change. McGraw-Hill, New York, U.S. A, pp. 5455 (2011) 7. Zhang, H., Zhao, X.: Analysis of organizational behavior of container shipping in the upper and middle reaches of the Yangtze River based on hub-and-spoke network. J. Coastal Res. 73, 119–125 (2015) 8. Bernstein, L., Roy, J., Delhotel, K.C.: Contribution of Working Group Three To Fourth Assessment Report of the Intergovernmental Panel on Climate Change, pp. 1–22. Cambridge University Press, UK (2007) 9. Greater London Authority. Climate Change and London’s Transport Systems. Greater London Authority, London, UK, pp. 4446 (2005) 10. Cai, W., Yang, C., He, B.: Preliminary Extension Logic. Science and Technology Literature Press, Beijing, pp. 20–21 (2015, in Chinese) 11. Luo, Z.: Research on logistics security system. Logistics Technol. 12(10), 8-10 (2005, in Chinese) 12. Liu, T.: Study on the Influence Mechanism and Model of Water Level Change on Shipping Logistics Cost in the Middle Reaches of the Yangtze River. Wuhan University of Technology, Wuhan, pp. 5458 (2015, in Chinese) 13. Zhang, H., He, Q.-M., Zhao, X.: Balancing herding and congestion in service systems: A queueing perspective. Inf. Syst. Operations Res. 58(3), 511–536 (2020) 14. Vanem, E., Skjong, R.: Designing for safety in passenger ships utilizing advanced evacuation analyses: a risk based approach. Saf. Sci. 44(2), 111–135 (2006) 15. Wu, Z.: Marine Traffic Engineering. Dalian Maritime University Press, Dalian, pp. 66-68 (2014) 16. Pillay, A., Wang, J.: Technology for Safety of Marine Systems. Elservier Ocean Engineering Book Series, pp. 8185 (2013) 17. Cai, W.: Matter-Element Model and its Application. Science Press, Beijing, pp. 89 (2014, in Chinese) 18. Abdullah, L., Otheman, A.: A new entropy weight for sub-criteria in interval type-2 fuzzy topsis and its application. Int. J. Intelligent Systems and Appl. (IJISA) 5(2), 2533 (2013) 19. He, Q.-M., Zhang, H., Ye, Q.: An M/PH/K queue with constant impatient time. Mathematical Methods of Operations Res. 18(11), 139-168 (2018) 20. Hadiwijaya, N.A., Hamdani, H., Syafrianto, A.: The decision model for selection of tourism site using analytic network process method. Int. J. Intell. Syst. Appl. (IJISA) 10(9), 2331 (2018) 21. Thayananthan, V., Shaikh, R.A.: Contextual risk-based decision modeling for vehicular networks. Int. J. Comput. Network Inf. Security (IJCNIS) 8(9), 19 (2016)
Influencing Factors and System Dynamics Analysis of Urban Public Bicycle Projects in China Based on Urban Size and Demographic Characteristics Xiujuan Wang1 , Yong Du1 , Xiaoyang Qi1 , and Chuntao Bai2(B) 1 Fundamental Science Section in Department of Basic Courses, Chinese People’s Armed
Police Force Logistics College, Tianjin 300309, China 2 Party and Government Office, Tianjin Beichen Science and Technology Park Management
Co., Ltd„ Tianjin 300405, China [email protected]
Abstract. The public bicycle system can be seen as a product of the development and progress of a low-carbon society. With the continuous development of urban scale, the problem of traffic congestion is becoming increasingly prominent. In order to reduce carbon emissions, “bicycle priority” has gradually become an important concept of urban transportation and travel. In some major cities in China, such as Hangzhou, the growth rate of public bicycles is extremely fast. However, at the same time, the research on China’s urban public bicycle system is relatively insufficient. In response to this issue, this paper analyzes the main characteristics of Chinese cities, and uses system dynamics methods to analyze the factors that affect the advancement of the public bicycle system, such as the size of Chinese cities and the main characteristics of the urbanization process, the division of transportation modes, population characteristics, bicycle infrastructure and right-of-way, and institutional structure. The results of these analyses is capable to help planners design feasible public bicycle development strategies. The paper summarizes the existing problems and solutions of the public bicycle rental system, providing a certain reference for further improving the public bicycle rental system. Keywords: Public bike · Characteristics of Chinese cities · Transportation system · Bicycle · System dynamics · Engineering of communication
1 Introduction Public bike system is a short term leasing system, in which user can pick the bike up and return it in any bike station, and do not need to bear the cost of buying one nor worry about its safety or parking [1]. In 1965, the first generation of public bike system began to operate in Holland, Amsterdam, which has experienced three generations after over 50 years’ development. At present, the vast majority of public bicycle projects in the world belong to the third generation public bicycle system. By 16 September 2018, 1780 cities worldwide have hosted © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 514–523, 2023. https://doi.org/10.1007/978-3-031-36115-9_47
Influencing Factors and System Dynamics Analysis of Urban
515
the advanced bike-sharing programs (Meddin and DeMaio, 2018) [2]. Once known as “the kingdom of bicycles”, China has a long development history of bicycle, which has served as the one of the most important means of transportation for urban and rural residents and is widely used [3]. Even today, in most of the cities, bicycle is still one of the most important means of transportation. Even so the bicycle traffic in China has been neglected for a long time, so has the relevant research, not to mention related research on public bicycle Public bicycle has a long development time in foreign countries, and the related research is also relatively rich. According to the diffusion theory, its experience and policy are transferable [4]. In China, the history of public bicycle project is only 13 years. The actual projects basically learn from and even copy the experience of foreign countries’. There is no denying that the developed countries in the process of development of public bicycle project accumulated a lot of successful experience, but Chinese cities have their own characteristics and need to combine the experiences with the characteristics of local development to develop public bicycle projects.
2 Relevant Research on Public Bike Along with the extensive development of public bicycle at home and abroad, the research of public bicycle begins to emerge. Take published public bicycle related articles in Transportation Research Record (TRR), the vane in traffic field, for example, before 2009 there is no article in this field; after 2009, it increases year by year, 2 in 2010, 4 in 2011, 5 in 2012, and 5 in 2013. As of May, 2018, a total of 345 articles were presented by Web of Science, which indicates that public bicycle has attracted many researchers’ attention and is becoming a hot research topic. In general, the study of public bicycle shows a similar pattern, which has been in the stage of rapid development in the last more than ten years [5, 6]. From 2002, public bicycle research literature began to appear abroad, which can be divided into the following aspects: a. the introduction of public bicycle development [7]; b. feasibility analysis on public bicycle project [8]; c. public bicycle helmet use and safety of users [9]; d. research on public bicycle station layout optimization and public bicycle vehicle balancing and scheduling [10]; e. public bicycle project replacing other modes of transportation [11]; f. evaluation after the public bicycle project has been carried out [10]. Compared with foreign countries, the time of China’s public bicycle project is relatively late. In May 2008, Hangzhou launched China’s first public bicycle project Hangzhou public bicycle system. Since then the public bike development in China’s cities went onto the fast lane. By April 24, 2015, a total of 237 public bicycle projects had been launched in 29 provinces and municipalities in mainland China [2]. Public bicycle has been paid more attention to in China, which has experienced a faster development process. However, China’s existing public bicycle study focused on introduction of public bicycle concept, technology, experience, etc., while analysis on problems and countermeasures of in the process of the development of public bicycle, public bicycle project investigation, public bicycle project operation evaluation, and systematic study based on the characteristics of China’s urban public bicycle have not yet been carried out. Compared with cities in developed countries, Chinese cities have many characteristics, which will affect and even determine whether public bicycle project can be successfully
516
X. Wang et al.
implemented. Therefore, this paper will analyze the impact on the development of public bicycle key factors basing on the characteristics of Chinese cities.
3 Analysis on Characteristics of Chinese Cities Factors affecting the development of urban public bicycle project can be divided into internal ones and external ones. Internal factors refer to the factors that can be adjusted according to the external environment in the short term, which can be divided into institutional design and physical design factors. The external factors are the factors that are related to certain cities and not easy to adjust in the short term. Specific division and contains of internal and external factors are in Table 1. Table 1. Internal and factors influencing the development of urban public bicycle project Internal factors
Physical design
Hardware and technology Service design
institutional design
Operator category Contract and ownership Sources of funds Job opportunities
External factors
City scale climate Traffic mode division Population density Demographic characteristics Economic development level Geography and terrain Existing infrastructure conditions Financial situation Political conditions
Resource: According to Büttner J, Petersen T. Optimising Bike Sharing in European Cities-A Handbook.
Considering China’s vast territory and different cities have their own characteristics, this paper only selected general characteristics which can represent Chinese city but significantly different with foreign ones when discussing the external factors affecting he public bicycle. Specifically, it includes city size, process of urbanization, traffic patterns, demographic characteristics, existing infrastructure conditions, political conditions, etc.
Influencing Factors and System Dynamics Analysis of Urban
517
3.1 City Size and Process of Urbanization Usually, there are two kinds of indicators of measuring the size of the city--the demographic indicator and the geographical indicator. From the aspect of regional scale, the size of China’s cities continues to expand. From 1990 to 2000, the built-up area of China’s urban area increased from 12.2 thousand square kilometers to 21.8 thousand square kilometers, an increase of 78.3%. By 2010, this number increased to 40.5 square kilometers, and increased by another 85.5%. The number in 2010 is more than twice of that in 1990 [13]. With the constant enlargement of the city scale, people’s travel distance will be increased accordingly. The urban population spatial distribution is further concentrated, and the regional imbalance is further intensified. According to the Chinese city status report (2012– 2013), taking the three major cities--Beijing, Shanghai and Guangzhou for example, in 2010, the total population of three cities in the proportion of 268 prefecture level cities was 4.5%. The population density of the Yangtze River Delta, Pearl River Delta and Beijing-Tianjin-Hebei region were 739 people per square kilometer, 608 people per square kilometer and 481 people per square kilometer, respectively, far higher than the average population density of China (140 people/km2 ) and the western region (53 people/km2 ). Public bicycle project has scale benefit and network benefit, and with the increasing of the scale and scope of the network, the operation cost is lower and the service level is higher, namely, to provide services through the construction of the scale and network services; therefore, to carry out the project requires certain local urban traffic demand. If the local population and travel demand are small, in order to ensure the convenience of the public bicycle project, it is necessary to maintain a certain scale of facilities, it will not be economic and the benefit cannot be guaranteed. if the operation depends only on the government’s subsidy, it is difficult to ensure that economic projects can be sustained. Compared with foreign cities, Chinese cities have large scale, high population density; big cities, especially mega cities have objective needs to carry out the public bicycle project, and even larger public bicycle plan. 3.2 Traffic Mode Division As a new mode of traffic travel in the city, public bicycle is bound to have an impact on the structure of urban transportation system. At the same time, the structure of the urban transportation system will affect the development and operation of the project, so it is necessary to understand the structure of urban traffic system, and determine the proportion of different traffic modes. So far, bicycle traffic is still a very important travel method in many Chinese cities, its proportion is much higher than that in Germany and Holland where public bicycle project is carried out smoothly. At present, both production and export amount of Chinese bicycles ranked the or above 60% all over the world, and the consumption ranked the first. Taking Beijing for example, the proportion of traveling by bicycle has been dropping since1986, but by 2010 it is still the important method of traveling in the city traffic system, which is shown in Fig. 1.
518
X. Wang et al.
Data recourse˖Beijing traffic development research center. Note˖Walking not included Fig. 1. Traffic mode proportion of Beijing
3.3 Demographic Characteristics 3.3.1 Age Considering China’s rapid urbanization, there will be more and more senior citizens. Since public bicycle needs human power to drive and there is a certain requirement of physical strength and techniques. The quality of the individual and riding speed limit their travel distances. The elder’s physical is relatively poor, so travel ways consume too much physical strength are not applicable. Therefore, in order to achieve the sustainable development of public bicycle which is a green traffic mode in urban China, the aging problem must be taken into consideration. Compared with western countries, China is facing a more serious problem of population aging. In 2000, China’s over 60 years old population is 1.3 billion, accounting for 10.71% of the total population at that time. According to international aging society standard (population over 60 years old accounted for the proportion of the total population of 10%, or over the age of 65 population accounted for total population proportion reached 7%), China has become an aging country since 2000. 3.3.2 Incomes Income of urban residents continued to increase and the growth speed is relatively fast, but the low-income groups still account for a higher proportion. The per capital disposable income of urban residents increased from 1510 yuan in 2011 to 21810 yuan in 1990, an increase of 14.5 times. The proportion of lower income households, whose average income is less than 6566 yuan, is 40%. The middle-income households, whose average income is 13178 yuan, accounted for 20%. The proportion of low-income groups is larger. In China’s urban transport system, the bicycle traffic is a major way to travel.
Influencing Factors and System Dynamics Analysis of Urban
519
3.4 Bicycle Infrastructure and Right of Way Characteristics Bicycle infrastructure and security are very closely linked. A United States survey shows that bicycle traffic safety and travel rate increase with the improvement of bicycle infrastructure. But for the country where bicycle infrastructure is relatively common, the new facilities have less obvious influence on improving the performance of the level of bicycle travel. The existence of bicycle infrastructure cannot directly determine the proportion of bicycle travel, while a higher proportion of bicycle travel can stimulate the construction of bicycle infrastructure. Bicycle traffic has always been an important part of the urban transportation system in China, and perfect bicycle infrastructure has been built up in most cities. In recent years, for the purpose of developing sustainable transportation, many cities in China are planning and constructing bicycle traffic network system in all directions in the urban areas to meet the development of bicycle traffic. In recent years, the standard of many city’s bike lanes and road mileage has been improved, but there is still uneven allocation of urban road space, bicycle driving space constantly being squeezed and unguaranteed traffic safety, lacking effective laws and regulations to protect the rider’s right of way. 3.5 Regime Structure Public bicycle is a typical quasi-public goods, which has the characteristics of public welfare and economy, which needs the participation of the government to ensure its public welfare. In order to strengthen the efficiency of management, government is divided into several departments, each of whose responsibilities and rights are only limited to the scope of the higher authorities. There is no special and independent organization to manage the public bicycle, which is a new thing, so its operators need to deal with all departments. Taking the site selection of the public bike station as an example, it involves the different departments concerning traffic management, urban management, urban planning, urban construction, greening, electricity and others. For example, in Shenzhen, when developing public bicycle project, setting rental set near the bus station needs the approval of the city transportation department; setting rental set near the subway station requires the approval of the Shenzhen Metro Group; setting rental set on the pavement needs the approval of Urban Management Department.
4 System Dynamics Analysis on the Relationship Between Bicycle Travel and Public Bicycle In the short term, public bicycle development can ease the traffic problems caused by motor vehicles, but it also occupies public resources for the development of bicycle traffic, such as funds, land and so on. At present, the funding of the development of public bicycle project is not sustainable, which needs long-term capital from government and other investment funds to operate. When transport related investment is not sufficient, public bicycle will occupy funds used for other bicycle related projects, such as funds for bicycle related infrastructure construction, and cycling promoting related policies. In
520
X. Wang et al.
most Chinese cities, the bicycle is an important means of transportation, which attracts a large number of users; but resources input for improving bicycle safety and convenience is far from enough. If the resources are used to carry out public bicycle project, attracting the original bicycle users to the use of public bike, it will not be very useful to solve the urban traffic problems. Therefore, the effect of settling the urban transport problems by whether using capital, land and other public resources to carry out public bicycle project, or directly to improve the use of bicycles need to be examined in a long-term aspect. In this paper, this issue will be examined with the system dynamics method in urban traffic system. 4.1 Reason for Using the System Dynamics Method System dynamics method was first proposed by Forrester, which is based on feedback control theory and takes computer simulation technology as the main method. It is mainly used in the study of complex social economic system. It was first used in the urban areas, and was then quickly spread to economics, ecology and other fields. In the early stages of the traffic field was used to analyze the relationship between traffic and land use. Urban transportation system includes a large number of elements, and the feedback mechanism between elements is complex. There is a dynamic and complex nonlinear relationship and feedback between the subsystems of different levels and the variables. It is difficult to comprehensively grasp the characteristics of urban traffic system by using the common method of quantitative research, while the system dynamics method can be used to conduct systematic comprehensive quantitative analysis. The public bicycle is a systematic problem, whose related variables have a dynamic and complex nonlinear relationship and feedback effect, and it is suitable to be analyzed with systematic dynamic method. 4.2 System Dynamics Analysis of the Relationship Between Bicycle and Public Bicycle China’s rapid urbanization process has increased the size of the city, which in turn has expanded the average travel distance of people. Considering suitable trip distance of different urban travel mode, cycling ratio decreases with people’ s average travel distance increasing, and the proportion of motor vehicle travel increases with people’ s average travel distance increasing. The convenience of vehicle travel provides the conditions to further expanding the scale of cities. On the other hand, this issue can be considered from the perspective of characteristics of Chinese urban land use. Before the Reform and Opening-up, more than 90% of urban population belongs to a neighboring work unit, with suitable housing, education, health care and other institutions around, and community function is perfect. While after the Reform and Opening up, China has implemented the land administrative allocation system for a long time and due to the scarcity of urban land, Chinese cities improve the utilization efficiency of urban land by increasing intensity of land, and integrating a variety of different land types. High mixture of the multi-function area is still the main characteristic of Chinese city. Therefore, the function structure and land use type of Chinese cities are different from western industrialized countries, which have the characteristics of compact structure
Influencing Factors and System Dynamics Analysis of Urban
521
and land use. Chinese cities’ this characteristic shorten people’s travel distance, making bicycles and other modes suitable for short distance travel play a greater role in the transport, reducing the demand for motorized travel mode. Mixed land use, the process of urbanization and motor vehicle travel form a dynamic feedback loop. Motorized travel growth rate exceeds the service capacity of the urban road, and traffic congestion reduces the vehicle speed. The rush hour traffic speed of certain road sections is even lower than the bicycle speed. Taking the cost into account, people will turn to the use of bicycles, thereby reducing the amount of motor vehicles. The aging of urban population, the increase of people’s income level and the bicycle travel forms a dynamic feedback loop. For travelers, the biggest advantage of public bike is exempting bike purchase and maintenance cost, but bicycles cost is not the main factor deciding bicycle use level. In urban mixed traffic environment, the most major factor affecting people using bicycle is safety. When the rider’s safety is guaranteed, more travelers will use bicycle. The more riders, the more attention motor vehicle drivers will pay, which can further improve the rider’s safety. Public resources devoted to cycling safety and the use of bicycle form a dynamic feedback loop [13]. Decision makers understanding the importance of bicycle infrastructure will benefit the investment of related public resources, but even if public resources are devoted into the bicycle infrastructure construction, the construction need some time. So there will not immediately be positive effect. In most cities in China, with the rapid increase of motor vehicles, bicycles and motor vehicles compete fiercely on the limited road, the right of road, capital and other public resources. That results in a lack of cycling facilities investment, and further affects the safety of bicycle use. Once security becomes a significant problem, there will be widespread social concern and decision makers’ awareness of the importance of bicycle infrastructure will be improved. A dynamic feedback loop is formed between the structure of Chinese cities and the safety of bicycle. Experience of carrying out public bike project at home and abroad shows that public bicycle increases the number of riders in the short term, which in turn requires more bicycle related facilities. So it improves the decision makers’ awareness of the importance of bicycle facilities and increases investment of cycling facilities. Due to that in a certain period a city’s resources of developing bicycle traffic is limited, the bicycle infrastructure construction and public bicycle project competition in the aspect of investment of public resources. Increasing bicycle facilities investment means reducing investment for public bicycle project resources, which is not conducive to the improvement of public bicycle project expansion and the service level. The public bicycle project and bicycle infrastructure development form a dynamic feedback loop [13–15]. According to the shortage of investment and growth based model, there is time delay between public resources input and bicycle related facilities level improvement. When the demand for bicycle related facilities is low, the government should increase the public resources investment in case of use decline due to the lack of related infrastructure. In general, there is bicycle related facilities supply shortage in most Chinese cities, this not an appropriate time to promote the growth ring, that is not to take measures to continue to increase bicycle travel. The correct management policy should aim to improve the level of bicycle infrastructure.
522
X. Wang et al.
5 Conclusion Public bicycle can ease the traffic problems caused by motor vehicles in the short term, but it also occupies the public resources for the development of bicycle traffic. The development of public bicycle is generally in the city center, occupying the valuable land resources, caused the public bicycle project is non-sustainable. In most cities of China, bicycle is an important means of transportation with huge number of users, but resources for improving bicycle safety and convenient is far from enough. Therefore, whether public resources such as funds and land should be used to develop public bicycle projects or directly used to increase the utilization rate of bicycles needs to be examined from the perspective of long-term solution to urban traffic problems. Public bike is a systematic problem, with complex dynamic nonlinear relationship and feedback effect between its related variables. The characteristics of urban traffic system cannot be comprehensively grasped by the usual quantitative research methods, while systematic dynamics method can carry out systematic comprehensive quantitative analysis of it. Therefore, this paper, taking the characteristics of Chinese cities into consideration, uses the systematic dynamics method to analyze the relationship between bicycle and public bicycle. The results showed that characteristics of Chinese cities determine that when most Chinese cities solving urban traffic problems through alternative means of transportation, especially the bicycle travel, the most pressing matter of the moment is not immediately launching large-scale public bicycle project, but taking measures to create a good external environment for bicycle travel, introducing relevant laws and regulations to protect the bicycle rider’s right of way, ensure the safety and convenience of bicycle travel. Acknowledgment. This project is supported by Basic Research Project of Logistics College of Armed Police Force References (WHJ202101).
References 1. Shaheen, S.A., Guzman, S., Zhang, H.: Bikesharing in Europe, the Americas, and Asia past, present, and future. Transp. Res. Rec. 2143, 159–167 (2010) 2. Meddin, R., DeMaio, P.: The Bike-Sharing World Map [EB/OL]. http://www.bikesharingw orld.com. Accessed 26 Sept 2018 3. Lin, X., Wells, P., Sovacool, B.K.: The death of a transport regime? The future of electric bicycles and transportation pathways for sustainable mobility in China. Technol. Forecast. Soc. Chang. 132, 255–267 (2018) 4. Hua, Z.: Research on Consumer Adoption Behavior of Low Carbon Transportation Mode Innovation. Lanzhou University, Lanzhou (2011). (in Chinese) 5. Zhengke, Y., Kailing, D., Xuemei, Z.: Review of research on shared bicycles at home and abroad. J. Chengdu Univ. Soc. Sci. Ed. 2, 7 (2018). (in Chinese) 6. Zhang, S., Chen, L., Li, Y.: Shared bicycle distribution connected to subway line considering citizens’ morning peak social characteristics for urban low-carbon development. Sustainability 13, 9263 (2021). https://doi.org/10.3390/su13169263 7. Ji, Y., Fan, Y., Ermagun, A., et al.: Public bicycle as a feeder mode to rail transit in China: the role of gender, age, income, trip purpose, and bicycle theft experience. Int. J. Sustain. Transp. 11(4), 308–317 (2017)
Influencing Factors and System Dynamics Analysis of Urban
523
8. El-Assi, W., Mahmoud, M.S., Habib, K.N.: Effects of built environment and weather on bike sharing demand: a station level analysis of commercial bike sharing in Toronto. Transportation 44(3), 589–613 (2017) 9. Mooney, S.J., Lee, B., O’Connor, A.W.: Free-floating bikeshare and helmet use in Seattle, WA. J. Community Health 24, 1–3 (2018) 10. Haider, Z., Nikolaev, A., Kang, J.E., et al.: Inventory rebalancing through pricing in public bike sharing systems. Eur. J. Oper. Res. 270(1), S0377221718302030 (2018) 11. Campbell, K.B., Brakewood, C.: Sharing riders: how bikesharing impacts bus ridership in New York City. Transp. Res. Part A Policy Pract. 100, 264–282 (2017) 12. Yin, J., Qian, L., Singhapakdi, A.: Sharing sustainability: how values and ethics matter in consumers’ adoption of public bicycle-sharing scheme. J. Bus. Ethics 149(2), 313–332 (2018) 13. Porag, F., Hossain, S.: Intelligent tour planning system using crowd sourced data. Int. J. Educ. Manage. Eng. (IJEME) 01(08), 22–29 (2018) 14. Fadel, M.A., Elrefaei, L.A.: Adjustments of methodology planning and assessment activities of senior projects in the computer science program. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 10(2), 16–25 (2018) 15. Biswas, D., Samsuddoha, M.: Determining proficient time quantum to improve the performance of round robin scheduling algorithm. Int. J. Mod. Educ. Comput. Sci. 11(10), 33–40 (2019). https://doi.org/10.5815/ijmecs.2019.10.04
Cross-border Logistics Model Design e-Tower Based on Blockchain Consensus Algorithm Shujun Li1(B) , Anqi He1 , Bin Li1 , Fang Ye1 , and Bin Cui2 1 School of Economics and Management, Wuhan Railway Vocational College of Technology,
Wuhan 430205, China [email protected] 2 Faculty of Art, Dongguk University, Seoul 100715, South Korea
Abstract. Logistics plays an important role in the development of cross-border e-commerce. Cross border e-commerce enterprises have to choose appropriate logistics channels. This paper discusses the typical application models of blockchain technology in cross-border logistics, and proposes specific implementation paths to improve the performance of cross-border logistics management. This paper first analyzes the structure framework and core technology (common consensus algorithms) of the blockchain. Taking Shenzhen ZK cross-border logistics company as an example, a cross-border logistics mode ZK e-Tower based on cross-border blockchain has been proposed. The key issues in the system design, such as access control and identity management, consensus algorithm, blockchain based transaction process, its application in cross-border customs clearance, security and privacy protection have been analyzed in detail, and a prototype system was built for simulation experiments thus solving the problems of traditional information systems, such as excessive power of central organizations and information tracking difficulties. Keywords: Blockchain Technology · Cross-border Logistics · Application Mode
1 Introduction According to the report issued by the National Bureau of Statistics, China’s total import and export of goods in 2021 will be 391009.4 billion yuan, an increase of 21.4% year on year. Looking at China’s foreign economic data in the past five years, the total import and export of goods has shown a sustained growth trend, with the total import and export exceeding 35000 billion yuan in 2021. However, there is a contradiction between the slow development of cross-border logistics and the rapid development of cross-border e-commerce. The low transparency of information, long logistics time and space cycle, high cost, numerous and miscellaneous logistics participants, low logistics standardization level and low logistics efficiency have seriously affected consumers’ shopping experience. The Business transformation and improvements in operation cost have contributed to the development of the Supply chain analytics [1]. Combining the © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 524–535, 2023. https://doi.org/10.1007/978-3-031-36115-9_48
Cross-border Logistics Model Design e-Tower Based on Blockchain
525
characteristics of blockchain technology, such as decentralization and traceability, to establish a new import and export cross-border logistics model -- building a blockchain logistics application platform will help solve the problems of opaque cross-border logistics information, difficult returns and exchanges, and low service level in transaction, and help improve the efficiency of import and export logistics [2]. Blockchain technology has been extended from the original Bitcoin to all walks of life, especially in logistics transportation, logistics supply chain and other service transactions [3]. It also becomes an important task for state controllers to discover, prevent and pre-control Emergency Logistics Risks (ELR) [4]. In essence, blockchain technology is based on peer-to-peer untrusted networks. Network nodes can communicate directly without a central node. It integrates new applications of computer technologies such as data encryption, storage, communication, consensus mechanism, and network structure [5]. Its characteristics of decentralization and trustworthiness overcome the insecurity of centralized networks. The traditional transaction model of logistics independent certification center has information security problems [6]. First, it is vulnerable to network attacks, which can lead to the disclosure of confidential information; Second, it is impossible to verify the authenticity of the user’s identity and review the traceability, and the unique identity signs of both parties to the transaction cannot be guaranteed [7, 8]; Third, credit rating is carried out through the past experience and qualification review of both parties to the transaction. Credit rating may require enterprises to spend corresponding costs, which allows the decision-maker to have knowledge of economic, ecological and social cost before making a decision [9]. And references [9, 10] have also discussed relevant blockchain issues on other areas. It is very difficult to collect comprehensive, complete and systematic logistics transaction information. Therefore, the information security of logistics trading activities and the consensus trust of upstream and downstream customers need to be solved urgently. Reference [11] have also pointed relevant security issues, and the algorithm used in the modes can be verified. Therefore, it is of great scientific significance and economic value to integrate cloud computing and blockchain technology and use the blockchain consensus algorithm to solve the problem of decentralization of logistics transactions and trust between users.
2 Common Consensus Algorithms in Blockchain Technology Now, there are many consensus algorithms used in the blockchain, such as: Proof of Work (PoW) algorithm, Proof of Stake (PoS) algorithm, Delegated Proof of Stake (DPoS) algorithm, and practical Byzantine Fault Tolerance (PBFT) algorithm [12]. Common consensus algorithms in blockchain technology are as follows. 2.1 PoW Algorithm PoW algorithm is a mechanism to prevent distributed service resources from being abused and denial of service attacks, which requires nodes to perform complex operations that consume time and resources in an appropriate amount, and its results can be quickly verified by other nodes to ensure that services and resources are used by real demand.
526
S. Li et al.
Hash algorithm is the basic technical principle of PoW algorithm. Suppose the hash value Hash (r) is calculated. If the original data is r, the operation result is R (Result). 1) R = Hash(r) For any input value r, the result R is obtained, and it cannot be inferred back from R. When the input original data r changes by 1 bit, the resulting R value changes completely. In the PoW algorithm of Bitcoin, the algorithm difficulty d and the random value n are introduced, and the following formula is obtained: 2) Rd = Hash(r + n) This formula requires that when the random value n is filled in, the first d bytes of the calculation result Rd must be 0. Due to the unknown result of the hash function, each node needs to do a lot of operations to get the correct result. After the calculated result is broadcast to the whole network, other nodes only need to perform a hash operation to verify. PoW algorithm uses this method to make computing consume resources, and the verification only needs one time. 2.2 PoS Algorithm The PoS algorithm requires that the node verifier must pledge a certain amount of funds to qualify for mining and packaging, and the regional chain system uses a random method when selecting packaging nodes. When the node pledges more funds, the greater the probability that it will be selected to package blocks. For example, if a node owns 5% of the entire blockchain system, it will have a 5% probability of packaging blocks in the next block out cycle. The process of node block out through PoS algorithm is as follows: To become a block out node, an ordinary node first needs to pledge its assets. When it is the node’s turn to block out, the block is packaged, and then broadcast to the whole network. Other verification nodes will verify the validity of the block. 2.3 DPoS Algorithm DPoS algorithm is similar to PoS algorithm, which also uses shares and equity pledge. The difference is that the DPoS algorithm uses the entrusted pledge method, which is similar to the method of electing representatives by the whole people to select large numbers of nodes for blocking. Voters cast their votes to a node. If a node is elected as an accounting node, the accounting node can return their votes to other voters in any way after obtaining a block of rewards. Those nodes will take turns to generate blocks, and the nodes will supervise each other. If they commit crimes, the pledge deposit will be deducted. 2.4 PBFT Algorithm We believe that PBFT algorithm can solve more problems with higher efficiency [13]. Figure 1 shows the basic consensus process of PBFT algorithm.
Cross-border Logistics Model Design e-Tower Based on Blockchain
527
In 1982, Leslie Lamport proved in the paper The Byzantine Generals Problem that when the total number of generals is more than 3f and the number of traitors is f or less, loyal generals can reach consensus on orders, that is, 3f + 1≤ n, and the algorithm complexity is O (nf + 1). In 1999, Miguel Castro and Barbara Liskov published a paper, Practical Byzantine Fault Tolerance, proposing the PBFT algorithm [14]. The number of Byzantine fault tolerance of the algorithm also meets 3f + 1≤ n, and the algorithm complexity is O (n2). This algorithm can provide high-performance computing, so that the system can process thousands of requests per second, which is faster than the old system.
Fig. 1. Consensus process of PBFT algorithm
The consensus process of the PBFT algorithm is as follows: the client initiates a message request, broadcasts and forwards it to each replica node (Replica), and one of the master nodes (Leader) initiates a pre-prepare proposal message and broadcasts it. Other nodes obtain the original message and send the prepare message after verification. Each node receives 2f + 1 prepare message, which means that the preparation is completed, and sends a commit message [15]. When the node receives 2f + 1 commit message, it is considered that the message has been confirmed to be completed (reply).
3 e-Tower Design for Shenzhen ZK Company Cross-border Logistics According to the actual situation of ZK company, this paper designs the cross-border logistics information platform ZK e-Tower based on blockchain technology, which mainly involves the following key issues: access control and identity management; security and privacy protection of users and transaction data in, as well as the consensus algorithm used in the blockchain network. The system adopts alliance chain architecture and introduces PKI system in the outer layer, so as to achieve the purpose of strengthening system identity management and access control. At the same time, BFT-SMART protocol based on fault-tolerant is introduced as the consensus algorithm [16, 17]. On the premise of ensuring that the system performance meets the actual needs, the consensus algorithm of conflict fault-tolerant commonly used in the existing chain system is
528
S. Li et al.
raised to the level that can tolerate 31% Byzantine errors. Although the system discards the completely decentralized characteristics of the traditional blockchain, it can meet the needs of information platform. On the contrary, the multi centralized architecture based on blockchain is conducive to giving full play to the advantages of blockchain technology in data tamper proof, traceability, multi-party verification and maintenance and other features. 3.1 Overall Architecture Design The e-Tower system architecture proposed in this paper is shown in Fig. 2, which is divided into two layers: the interaction and the blockchain consensus layer.
Fig. 2. Overall architecture design of e-Tower
The application layer is responsible for encapsulating user data and operations into standard digital assets, and then sending various requests to blockchain nodes in a predefined standard transaction form to complete user operations, and finally storing the data on the blockchain [17]. The blockchain consensus layer includes the P2P node network that maintains the blockchain and the only blockchain jointly maintained by the nodes in the system. The main participants are responsible for running the physical server as the node, verifying the transactions from the upper layer and composing the transactions into blocks, running the consensus algorithm and finally updating the data in the blockchain copy of each node. 3.2 Composition of Underlying P2P Network The e-Tower system adopts modular design and divides the nodes in the P2P network into three roles: 1) Endorsement Node: the endorsement node is responsible for verifying the legitimacy of the transaction and signing it. After the validation, it is responsible for executing the transaction.
Cross-border Logistics Model Design e-Tower Based on Blockchain
529
2) Organization Node: the organization node receives the transaction signed by the endorsement node, runs the consensus algorithm, and organizes the transaction into blocks to deliver to the submission node after ensuring consistency. 3) Submission node: submit the node to verify the block completed by the organization, and then update the blockchain with the block. 3.3 Application of PBFT Algorithm Integrating the cross-border logistics transaction process, mechanism and related theories of blockchain and cloud computing, and using the consensus mechanism of the practical fault tolerance algorithm (PBFT) of blockchain, the definition of logistics blockchain and cloud logistics blockchain is proposed, see Fig. 3.
Fig. 3. PBFT algorithm of e-Tower
The main steps of the algorithm’s application are as follows: 1) The client sends a request to the master node. 2) The master node broadcasts the request to other nodes, and the node executes the three-stage consensus process of the PBFT algorithm. 3) After the node processes the three-stage process, it returns a message to the client. 4) After the client receives the same message from the f + 1 node, the consensus has been completed correctly [17]. 3.3.1 Definitions For e-Tower, six definitions are proposed: Definition 1: There are at least 3f + 1 nodes in the logistics blockchain, which are adjacent to each other and form a basic Byzantine unit (UB), where f represents a normal logistics node and UB represents a consensus network unit. Definition 2: In the logistics blockchain, the PBFT mechanism is implemented according to the View, which is numbered and recorded as n. In a View with 3f + 1 nodes in total, there is only one master node, recorded as m, and the rest of the nodes are backup nodes called replicas. Each node is represented by an integer and is {0, 1, …3f + 1}in order, satisfying the following: m = n mod 3f + 1, where n represents consensus network initialization, m represents the origination block.
530
S. Li et al.
Definition 3: All nodes(Nodei) in the logistics blockchain use double chains to store blockchain information of other nodes. Nodei = {a, b}, a and b represent blockchain header index and blockchain content respectively. Definition 4: Blockchain information B-Info stores blockchain data information D, and records the node name N, data key K, and update time T. B-Info structure: B-Info = (BD, BN, BK, BT). Definition 5: On the basis of logistics blockchain, e-Tower virtual logistics network node mapping: Node: UB → UBC . C represents a collection of virtual nodes. Definition 6: Logistics blockchain achieves virtual node U through certificate transfer UBC Consensus process [18]. The certificate message format is: Message = (REPLY, b, v, i), where REPLY is the message type. Depending on the certificate type, b may be a block or a block hash, v represents the view number, and i represents the node number. 3.3.2 Recycle Bin With the state replication algorithm of PBFT, Replicas will record executed messages in the local log. To save memory, a mechanism is needed to clean the log. It is unwise to execute after each operation, because it consumes resources. It can be cleaned regularly, such as once every 100 times [19]. We call the post request execution status Checkpoint; whose checkpoint with proof is called stable certificate. When the node receives 2f + 1 checkpoint message, it can prove that the stable checkpoint is correct. Log messages before the stable checkpoint can be deleted. When clearing checkpoints, Replica broadcasts a checkpoint protocol to other replicas: < checkpoint, n, d , i >sig i
(1)
n is the serial number of the last correct execution request, and d is the summary of its current status. If each replica receives 2f + 1 checkpoint message with the same sequence number n and summary d, then each replica can clear the sequence number < n, d , v > whose log information less than or equal to n. Checkpoint protocols are also used to update logistics routes. The normal line h is equal to the serial number of the nearest stable logistics line, the abnormal line H = h + l, and l is the log size. 3.3.3 Information View Change When the master node of logistics is invalid, or some nodes receive 2f + 1 commits in the commit phase, and some nodes do not receive 2f + 1 commits, resulting in inconsistent states, the information view needs to be changed to provide system activity and security. When the request times out, the backup node enters view v + 1 and broadcasts the view change message: < new − change, v + 1, n, C, P, i >
(2)
n checks serial number, C checks certificate, P is a set that contains the relevant message set Pm for request m (the request serial number is greater than n). Pm contains 2f + 1 identical preparation message.
Cross-border Logistics Model Design e-Tower Based on Blockchain
531
When the master node of view v + 1 receives 2f change messages of the same view, it broadcasts new view messages to other replicas: < new − view, v + 1, V , O >, V is 2f + 1 view change message. The calculation rules of O are as follows: 1) Determine the serial numbers mins and maxs , where mins is equal to the stable checkpoint serial number in V , and maxs is equal to the maximum prepare message serial number in V . 2) The master node allocates pre-prepare messages for each sequence number n between mins and maxs . If the V contains the P combination corresponding to n, the corresponding pre-prepared message: < pre − prepare, v + 1, n, d >, that is, there are 2f + 1 prepare messages for the request corresponding to serial number n, and the request is still submitted in the new view). If V does not contain the P combination corresponding to n, the submitted null message: < pre − prepare, v + 1, n, d < null >>, that is, no processing is done. After receiving the new view message, the replica broadcasts a prepare message, enters v + 1, and the view change is completed [19, 20]. 3.4 Transaction Process In the e-Tower system, a standard transaction process consists of three stages: transaction endorsement, generation block and confirmation submission. The specific transaction info-flow process is shown in Fig. 4.
Fig. 4. Transaction info-flow process of e-Tower
The client sends a request to any legal endorsement node to start a transaction. The endorsement node simulates the smart contract that executes the transaction, reads the status of the current blockchain, signs it, and returns it to the client program. After receiving the data, the client program will package it with the original transaction data and broadcast it to all organization nodes [20]. The organization node cluster implements the
532
S. Li et al.
improved consensus algorithm based on BFT-SMART. After receiving a certain number of transactions or reaching a certain time, the transactions that have passed the consensus algorithm are organized into a block. When a block is completed, the organization node broadcasts the block to all submission nodes. The submission node verifies the timeliness and legitimacy of the block, and updates the local blockchain copy with the transaction data in the block after passing. Finally, all submission nodes separately notify the client whether the transaction is successfully submitted [21]. A complete transaction process is completed. 3.5 Application in Cross-Border Customs Clearance Customs clearance is one of the core links of cross-border logistics. It is not only necessary to strictly manage customs, but also to improve the speed of customs clearance and promote trade facilitation. These two aspects have been difficult problems for customs agencies [21]. The key and difficult point is to ensure the reliability of the information data of entry-exit goods. Generally, enterprises need to provide complete documents and materials to verify the accuracy of information, including on-site inspection of goods, which makes it difficult to improve the speed and experience of customs clearance [22]. As shown in Fig. 5, the combination of blockchain technology and cross-border customs clearance has achieved results.
Fig. 5. Blockchain support in cross-border customs clearance
4 Privacy and Security Verification of e-Tower The e-Tower system guarantees the security of transaction data and user information in two aspects of data storage and access control. The classic blockchain structure proposed by Satoshi Nakamoto is adopted for data storage. All data generated are encapsulated
Cross-border Logistics Model Design e-Tower Based on Blockchain
533
in various standardized transactions. After being regularly packaged into blocks by organization nodes, they are finally put into the only blockchain owned by the trading network and become an immutable part of the blockchain. At the system access control level, the system introduces the X.509 certificate specification and Merkle tree to manage the identity and authority in the system. Any participant who joins the network must first obtain a digital certificate from the authority as the only legal identity for activities in the network. As shown in Fig. 6, Merkle tree is introduced to ensure the privacy and security of the system based on cryptography principles, which is used to design and realize the confidentiality, integrity, authentication and non-repudiation of the blockchain.
Fig. 6. Merkle tree
5 Conclusion Based on the consensus algorithm of blockchain technology, a cross-border logistics blockchain model e-Tower has been designed for Shenzhen ZK cross-border logistics company. Especially the PBFT algorithm based on the practical Byzantine consensus algorithm realizes the decentralization and non-tampering of the system. At the same time, the distributed storage characteristics of cloud computing are applied to solve the computing power problem of large-scale consensus computing, providing a blockchain solution to a series of problems existing in the current cross-border logistics industry, such as opaque transactions, high security risks, and poor interactivity. Experiments show that the model has good performance in security, stability and throughput. However, the openness and transparency of the blockchain also means that cross-border logistics information is exposed in the logistics network. While users trace the source of information flow, some information that is not suitable for disclosure will also be exposed. Therefore, we need to further apply decentralized construction, optimization, and higher-level privacy security technology to better ensure the system operation focusing on users’ information security.
534
S. Li et al.
References 1. Anitha, P., Pati, M.M.: A review on data analytics for supply chain management: a case study. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 11(05), 30–39 (2018) 2. Zhang, Y.: Building China-EU cross-border e-commerce ecosystem with blockchain technology. China’s Circ. Econ. 32(2), 66–72 (2018). (in Chinese) 3. Wang, C., Yidi, W., Qin, Q.: Supply chain logistics information ecosystem model based on blockchain. Intell. Theor. Pract. 40(7), 115–120 (2017). (in Chinese) 4. Cheng, Q., Lei, Y.: Operational mechanism and evaluation system for emergency logistics risks. Int. J. Intell. Syst. Appl. (IJISA) 02, 25–32 (2010) 5. Hu, J., Ge, C., Sun, Y.: Research on logistics information management framework based on blockchain. Logistics Technology 41(10), 34–36 (2018). (in Chinese) 6. Li, X., Wang, Y., Wang, F.: Research on the application mode and implementation path of blockchain technology in the field of cross-border logistics. Contemp. Econ. Manage. 07, 1–2 (2020). (in Chinese) 7. Cavinato, J.L.: Supply chain logistics risks from the back room to the board room. Int. J. Phys. Distrib. Logist. Manag. 34(5), 383–387 (2004) 8. Gilad, Y., Hemo, R., Micali, S., Vlachos, G., Zeldovich, N.: Algorand: scaling byzantine agreements for cryptocurrencies. In: Proceedings of the 26th Symposium on Operating Systems Principles, pp. 51–68. ACM (2017) 9. Benotmane, Z., Belalem, G., Neki, A.: A cost measurement system of logistics process. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 05, 23–29 (2018) 10. Demestichas, K., Peppes, N., Alexakis, T., Adamopoulou, E.: Blockchain in agriculture traceability systems: a review. Appl. Sci. 12, 96–97 (2020) 11. Kiayias, A., Russell, A.: Ouroboros-BFT: a simple byzantine fault tolerant consensus protocol. Preprint Minor Revision 01, 1049 (2018) 12. Ning, Z., Li, M.: Logistics information platform LIP Chain based on alliance blockchain. Computer Technology and Development 08, 191–192 (2019). (in Chinese) 13. David, B., Gaži, P., Kiayias, A., Russell, A.: Ouroboros praos: an adaptively-secure, semisynchronous proof-of-stake blockchain. In: Nielsen, J.B., Rijmen, V. (eds.) Advances in Cryptology – EUROCRYPT 2018: 37th Annual International Conference on the Theory and Applications of Cryptographic Techniques, Tel Aviv, Israel, April 29 - May 3, 2018 Proceedings, Part II, pp. 66–98. Springer International Publishing, Cham (2018). https://doi.org/10.1007/ 978-3-319-78375-8_3 14. Zhou, J., Li, W.: Research on consensus algorithm of logistics blockchain based on cloud computing. Comput. Eng. Appl. 54(19), 239 (2018) 15. Ma, J.: Research on the new mode of blockchain and import logistics. Technol. Market 25, 23–24 (2018). (in Chinese) 16. Yao, L., Hao, D., Jianhua, M.: Optimization of logistics supply chain based on blockchain technology and genetic algorithm. Process Automation Instrumentation (07), 97 (2022). (in Chinese) 17. Zhao, G., Dong, J., Liu, M., Zhai, K.: Evolution game of banks and enterprises in digital supply chain finance driven by blockchain. In: The 34th Chinese Control and Decision Conference (CCDC), pp. 147–151. Hefei Press, Hefei (2022). (in Chinese) 18. Liang, W.: Research on service quality improvement of third-party logistics based on blockchain technology. Logistics Technol. Appl. 02, 101–102 (2020) 19. Saberi, S., Kouhizadeh, M.: Blockchain technology and its relationships to sustainable supply chain management. Int. J. Prod. Res. 57, 7–8 (2019) 20. Issaoui, Y., Khiat, A., Bahnasse, A., Ouajji, H.: Smart logistics: study of the application of blockchain technology. Procedia Comput. Sci. 160, 266–271 (2019)
Cross-border Logistics Model Design e-Tower Based on Blockchain
535
21. Wang, Y.: Strategies and suggestions for coordinated development of cross-border ecommerce and cross-border logistics in China. Economics Management 05, 14–15 (2019) 22. Zhu, G., Zhao, Y.: Application and development of blockchain technology in logistics. J. Suzhou Univ. Sci. Technol. (Eng. Technol.) 12, 34–35 (2020). (in Chinese)
Research on the Service Quality of JD Daojia’s Logistics Distribution Based on Kano Model Yajie Xu1,3(B) and Xinshun Tong1,2 1 Business School, Zhongyuan Institute of Science and Technology, Zhengzhou 45000, China
[email protected]
2 College of Economics and Management, Zhengzhou University of Light Industry,
Zhengzhou 45000, China 3 Henan Port Hub and Port Economic Research Center, Zhengzhou 45000, China
Abstract. As an important support for the development of new retail, logistics distribution is the closest logistics service format to consumers. Improving the quality of logistics distribution service, improving logistics efficiency and consumer satisfaction are the necessary conditions for the development of new retail. In order to effectively improve the service quality and customer satisfaction problems in the logistics and distribution of JD Daojia, based on Kano model, establish JD Daojia logistics service quality improvement model, determine the priority of the improvement of JD Daojia’s service quality elements, and guide enterprises to achieve maximum customer satisfaction with minimum investment. Through empirical research, the customer’s perceived importance and satisfaction with JD Daojia’s service quality elements were analyzed, and the service quality factors with the highest priority for improvement were obtained, which verified the feasibility and effectiveness of the model, It can provide some practical guidance for improving the service quality of JD Daojia. Keywords: Kano model · Service quality · JD Daojia
1 Introduction According to the Chinese New Year Consumption Report released by JD Daojia, quality replaces price to become shopping consumption first choice factor, “Self-pleasing consumption” is leading the new trend of mass online shopping. According to the consumption data released by JD Daojia, In the past few years, “One-Hour Shopping” is officially moving towards all categories, all scenes, and all customers, and in addition to the first-tier cities also obtained rapid development, The GMV growth rate of second- and third-tier cities has exceeded that of first-tier cities. At the same time of its rapid development, Consumers’ satisfaction with their service quality is becoming more and more important. Delivery service quality is an important indicator reflecting customer’s perceived satisfaction. This paper is based on the Kano model, established JD Daojia logistics service quality improvement model. And through empirical research, analyze customers’ perceived importance and satisfaction with the elements of JD Daojia’s logistics service quality, obtain the most prioritized stream service quality elements for improvement, it provides certain decision basis for enterprise logistics management. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 536–546, 2023. https://doi.org/10.1007/978-3-031-36115-9_49
Research on the Service Quality of JD Daojia’s Logistics Distribution
537
2 The Connotation of JD Daojia and Kano Model 2.1 JD Daojia JD Daojia is a brand-new business model developed by JD Group based on the traditional B2C business model to the higher frequency and sub-commodity service field. It is the O2O life service platform that JD.com focused on in 2015 and is an important improvement based on the traditional B2C model to the high frequency field [1]. It is based on the advantages of JD logistics system and logistics management, and at the same time, relying on the “Internet+” technology to vigorously develop “crowd sourcing logistics” under the promotion of the popular sharing economy, integrate various O2O life categories, cooperate with Dada, and provide consumers The distribution of fresh food and supermarket products is quickly delivered within 2 h based on LBS positioning, creating an integrated application platform for life services. Based on the change in consumption trends, in order to better meet consumer demand, JD Daojia proposed to promote the accelerated development of O2O instant consumption channels with “deep empowerment, deepening of channels, and deepening of products” as its core strategy. JD Daojia provides several types of home services, including supermarket home, takeout home, quality life, door-to-door service, and healthy home, etc. JD Daojia forms the prototype of “social e-commerce” in China through leveraging social resources [2]. In terms of commodity and service sources, JD Daojai provides logistics delivery through cooperation with large shopping malls. In terms of life services, JD Daojia can provide thousands of products and services such as supermarket products, fresh food, and takeout; In terms of the logistics distribution system of JD Daojia, there are not only JD’s self-operated logistics distribution, but also social logistics such as “JD Crowdsourcing” [3]. The following figure shows the information index of JD Daojia in the past three months, as shown in Fig. 1.
Fig. 1. JD Daojia Information Index
2.2 Kano Model In 1974, Herzberg put forward the two-factor theory when researching employee satisfaction. He criticized the traditional concept that “satisfaction and dissatisfaction are completely opposite”. He believed that even if the dissatisfaction factor is removed, employees will Not necessarily satisfied, and vice versa. Inspired by the two-factor theory, Noriaki Kano proposed the Kano demand model. As shown in Fig. 2, the abscissa
538
Y. Xu and X. Tong
means the quality characteristics of the product relative to the degree of satisfaction of consumers [4]. The more to the right, the better the quality characteristics of the product can meet the expectations of consumers; The ordinate represents the level of consumer satisfaction, and the higher it is, the higher the consumer’s satisfaction with the product [5]. Charm attribute belongs to the attribute of “icing on the cake”, which means that if the product provides this quality feature, the user’s feeling can get “surprise” and “pleasant” experience, and satisfaction will increase; However, if this quality feature is not provided, customers will not be disappointed and sad, they will feel “taken for granted” and will not reduce their satisfaction [6]. One-dimensional attributes are also called expected attributes, which means that if the product provides the quality characteristics, the user’s satisfaction will be very high, otherwise it will be very low. The necessary attributes are “must have” attributes. When the quality characteristics of the product are insufficient, it will greatly reduce the user’s experience and satisfaction; But if you find ways to optimize and improve this quality feature, user satisfaction will not increase significantly; The indifference attribute belongs to the attribute of “optional”, which means that no matter whether the product has the quality characteristic, the user’s feeling is not too strong, and the user satisfaction will not be affected by it [7]. The reverse attribute is a “backfire” attribute, which means that users have no demand for this quality characteristic. If the product provides this quality characteristic, it will make customers “very dislike” and “boring” and reduce user satisfaction [8].
Fig. 2. Kano two-dimensional attribute model
The Kano model is a useful tool to prioritize the classification of user needs. It is based on analyzing the impact of user needs on user satisfaction, and reflects the nonlinear relationship between product performance and user satisfaction [9]. In the Kano model, “M” means necessary demand, “A” means attractive demand, “O” means one-dimensional (expected) demand, “R” means reverse demand, “I” means indifferent
Research on the Service Quality of JD Daojia’s Logistics Distribution
539
demand, and “Q” Means that the surveyor’s answers are contradictory and are marked as “questionable answers”. See Table 1. Table 1. Classification of Kano model evaluation results Not providing service quality elements (negative problem)
Provide service quality elements (positive issues)
I like this
It must be like this
I do not mind
I can stand it
I hate that
I like this
Q
A
A
A
O
It must be like this
R
I
I
I
M
I do not mind
R
I
I
I
M
I can stand it
R
I
I
I
M
I hate that
R
R
R
R
Q
3 JD Daojia Service Quality Improvement Model Based on Kano Model 3.1 Questionnaire Design and Data Collection Firstly, design the questionnaire of the Kano model. For each service, the service quality elements of JD Daojia are shown in Table 2. Table 2. JD Daojia’s logistics service quality indicators Number
Content of service quality index
1
Delivered on time
2
Goods in good condition
3
Convenient query
4
Wide distribution range
5
Night delivery
6
Personalized service
7
Promotional offers
8
Pick up offline
9
Packaging and Message
10
Information push
540
Y. Xu and X. Tong
The forward question and the reverse question are set separately from the two aspects of whether they are satisfied. The structure refers to the Likert five-dimensional scale, from “I like this”, “It must be like this”, “I do not mind”, “I can stand it” to “I hate that”. Applying it to the “JD Daojia Customer Satisfaction Survey”, the Kano questionnaire design sample form shown in Fig. 3 can be obtained. Then, issue a questionnaire and conduct a survey.
Fig. 3. Sample questionnaire
This questionnaire survey adopts the method of online survey, sending electronic questionnaires online to collect the survey results. A total of 108 questionnaires were distributed and 90 were recovered. The invalid data were sorted through the questionnaires. There were 66 valid questionnaires, and the effective recovery rate was 73%. The relevant data collected by the questionnaire is mainly analyzed through Excel and SPSS 25. Collect data through fuzzy Kano questionnaires, and perform calculations and statistics, cross-analyze positive and negative questions, and get the result, for example, as shown in Table 3, thereby deriving the category of each service quality indicator. Table 3. Example diagram of service indicator type judgment Delivered on time
Positive
I like this
Negative I like this
It must be like this
I do not mind
I can stand it
I hate that
Count
0
0
4
19
19
Percentage (%)
0.00
0.00
9.52
45.24
45.24
0
0
5
2
11
O
28.79%
0.00
0.00
27.78
11.11
61.11
M
16.67%
It must Count be like Percentage this (%)
SUM
66
A
34.85%
(continued)
Research on the Service Quality of JD Daojia’s Logistics Distribution
541
Table 3. (continued) Delivered on time
Negative I like this
It must be like this
I do not mind
I can stand it
I hate that
SUM
66
Count
0
0
3
3
0
I
19.70%
Percentage (%)
0.00
0.00
50.00
50.00
0.00
R
0.00%
I can Count stand it Percentage (%)
0
0
0
0
0
Q
0.00%
0.00
0.00
0.00
0.00
0.00
I hate that
Count
0
0
0
0
0
Better/SI
0.63
Percentage (%)
0.00
0.00
0.00
0.00
0.00
Worse/DSI
−0.45
I do not mind
3.2 Data processing of Better-Worse coefficient According to the questionnaire survey made by the Kano model, the data is statistically sorted and divided to complete the demand classification for quality characteristics, and the next step is the Kano model analysis. The Kano model analysis is to determine the sensitivity of consumers to changes in the level of these service quality factors through the analysis of the satisfaction and dissatisfaction influence of various service quality factors, then determine the key factors for improving those service quality characteristics that are highly sensitive and more conducive to improving consumer satisfaction. According to the quality improvement coefficient theory proposed by Matzler (1998), the increase in satisfaction coefficient SI (Better coefficient) and the decrease in dissatisfaction coefficient DSI (Worse coefficient)—Better-Worse coefficient are calculated to show that the attribute of achieving this factor increases Satisfaction or elimination of the impact of dissatisfaction, Better’s value is usually positive, indicating that if the product provides a certain function or service, user satisfaction will increase [10]. The formula is as follows: Increase the coefficient of satisfaction SI SI(better coefficient) = (A + 0)/(A + 0 + M + I), Reduce the coefficient of dissatisfaction DSI DSI(worse coefficient) = −1 ∗ (0 + M)/(A + 0 + M + I) The Kano model method is used to summarize and summarize the questionnaire. According to the relationship between different types of quality characteristics and customer satisfaction, the quality characteristics of products and services are divided into
542
Y. Xu and X. Tong
five categories: basic demand, expected demand, exciting demand, and no Differential demand, reverse demand. The letters appearing in the Kano model satisfaction table represent the satisfaction between JD Daojia logistics services and customers, and the specific types are judged based on the data obtained, as shown in Table 4. Table 4. Kano model attribute type judgment number
Service quality factors
A(%)
O(%)
M(%)
I(%)
R(%)
Q(%)
Type
1
Delivered on time
34.85
28.79
16.67
19.70
0.00
0.00
A
2
Goods in good condition
13.64
39.39
24.24
22.73
0.00
0.00
O
3
Convenient query
25.76
34.85
15.15
24.24
0.00
0.00
O
4
Wide distribution range
43.94
30.30
1.52
19.70
4.55
0.00
A
5
Night delivery
45.45
3.03
1.52
42.42
6.06
1.52
A
6
Personalized service
57.58
7.58
4.55
27.27
3.03
0.00
A
7
Promotional offers
53.03
18.18
4.55
21.21
1.52
1.52
A
8
Pick up offline
31.82
10.61
7.58
43.94
6.06
0.00
I
9
Packaging and Message
51.52
6.06
3.03
39.39
0.00
0.00
A
10
Information push
21.21
3.03
1.52
59.09
15.15
0.00
I
Calculate the Better-Worse coefficient to show the degree of influence of achieving this factor attribute on increasing satisfaction or eliminating dissatisfaction. The value of Better is usually positive, which means that if JD Daojia provides a certain service, consumer satisfaction will increase [11]. The greater the positive value, the stronger the effect of improving consumer satisfaction and the faster the increase in satisfaction; The value of worse is usually negative, which means that if JD Daojia does not provide a certain service, consumer satisfaction will decrease. The larger the negative value, the stronger the effect of reducing consumer satisfaction, and the faster the decrease in satisfaction. According to the classification results of service factor attributes in Table 3, based on the Kano model, JD Daojia’s logistics distribution service types mainly include attractive attributes (A), expected attributes (O), and indifference attributes (I). Therefore, according to the Better-Worse coefficient, projects with higher absolute scores should be implemented first. As shown in Table 5. 3.3 Data Analysis and Conclusion In Kano model, the Better-Worse matrix mainly refers to the four-quadrant matrix graph based on the Better coefficient and the Worse coefficient. Through Kano research and analysis, according to the Better-Worse coefficient, a Better-Worse matrix is constructed as shown in Fig. 4. The Better-Worse matrix diagram more intuitively shows the classification and importance of JD Daojia’s logistics and distribution services. It can be seen that this data post-processing method is different from the two-dimensional attribute scale.
Research on the Service Quality of JD Daojia’s Logistics Distribution
543
Table 5. Better-Worse coefficient analysis Number
Service quality factors
1
Delivered on time
63.64
45.45
2
Goods in good condition
53.03
63.64
3
Convenient query
60.61
50.00
4
Wide distribution range
77.78
33.33
5
Night delivery
52.46
4.92
6
Personalized service
67.19
12.50
7
Promotional offers
73.44
23.44
8
Pick up offline
45.16
19.35
9
Packaging and Message
57.58
9.09
10
Information push
28.57
5.36
100.00%
Better(SI) coefficient(%)
Worse(DSI) coefficient(%)
worse
90.00% Wide distribution range Promotioal offers
80.00% 70.00% 60.00% better 50.00%
Personalized service Packaging and Message Night delivery
Delivered on time Convenient query Goods in good condition
better
Pick up offline 40.00% 30.00%
Information push
20.00% 10.00% 0.00% 0.00%
worse 10.00% 20.00% 30.00% 40.00% 50.00% 60.00% 70.00% 80.00% 90.00% 100.00%
Fig. 4. Better-Worse matrix
The first quadrant represents the better coefficient value is high, and the absolute value of the worst coefficient is also high. The factors that fall into this quadrant are called expectation factors (one-dimensional factors). The wide range of delivery, ontime delivery, and convenient inquiries fall into this quadrant, which means that JD Daojia provides this service and consumer satisfaction will increase. When this function is not provided, customer satisfaction will decrease; The second quadrant represents the situation where the better coefficient is high and the absolute value of the worst coefficient is low. The factors that fall into this quadrant are called glamour factors. Packaging and message services, personalized services and
544
Y. Xu and X. Tong
promotional services fall into this quadrant, which means that this service is not provided, and consumer satisfaction will not decrease. When JD Daojia When this service is provided, consumer satisfaction will be greatly improved; The third quadrant represents the better coefficient value is low, and the absolute value of the worst coefficient is also low. The factors that fall into this quadrant are called indifferent factors. Night distribution, offline pick-up and information push services fall into this quadrant, that is, whether these services are provided or not, consumer satisfaction will not change, these services are services that consumers don’t care about; The fourth quadrant represents the situation where the better coefficient is low and the absolute value of the worst coefficient is high. The factors that the good product service elements fall into this quadrant are called essential factors, which means that when JD Daojia provides this service, user satisfaction will not increase, and when this function is not provided, user satisfaction will be greatly reduced; Services that fall into this quadrant are the most basic delivery services.
4 Simulation Case JD Daojia’s Service Quality Improvement Strategy According to the data analysis structure, it can be concluded that in the actual service process, JD Daojia should first try its best to meet the most basic needs of consumers, that is, the intact factors of goods. These needs are what consumers believe JD Daojia has the obligation to do. This type of delivery service quality factor is also the main factor that causes customers to be dissatisfied with the service. JD Daojia should rationally allocate resources, make more perfect product classification and packaging, and use green environmentally friendly materials to package products to increase consumer awareness of environmental protection and achieve regeneration [12]. Utilize resources and promote the development of green logistics. Meet the basic needs of consumers, reduce the rate of customer complaints, and improve customer satisfaction. After realizing the most basic needs, JD Daojia should try its best to meet the expectations of customers, that is, the expectation factor expressed in the first quadrant, which is a competitive factor of service quality. Provide consumers with additional service functions that make their services superior to and different from competitors, and guide consumers to strengthen their good impression of JD Daojia. Finally, we strive to realize the glamorous needs of consumers, increase customer stickiness, which is the glamour factor represented by the second quadrant, and enhance consumer loyalty [13]. The consumption level is upgraded, the pace of life is accelerated, and convenient services can improve customer satisfaction. Therefore, the Better-Worse coefficient value calculated according to the Kano model shows that JD Daojia needs to optimize the wide range of delivery, on-time delivery service, and convenient query services, and then satisfy consumers for packaging and message services, personalized services and promotional offers Service demand, providing such new retail real-time logistics services is the key to improving customer satisfaction. JD Daojia can improve the corresponding distribution area division according to the actual situation in order to tap potential customers and expand the market. Carry out merchandise promotion and full discount activities, rationally use the membership system
Research on the Service Quality of JD Daojia’s Logistics Distribution
545
and coupons to attract users, and improve the market competitiveness of JD Daojia [14]. With the help of big data, artificial intelligence, 5G and other information technologies, we will improve the digital level of distribution services, rationalize the delivery time, improve the convenience of order inquiry, improve the online service platform, expand personalized business, and create high-quality logistics and distribution services. However, for consumers, night service, offline self-service and information push service, there is no difference in demand. Offline self-collection has less competitive advantages when facing retail stores in residential areas, and information push services are dispensable for online shopping, which is a highly purposeful shopping method. Therefore, when resources are limited under circumstances, there is no need to make great efforts to achieve it. Through the establishment of logistics distribution service quality improvement model, JD Daojia can set different weights for different service attributes in the process of improving logistics service quality, concentrate resources to develop customer-focused service attributes, and maximize customer satisfaction.
5 Conclusion Further improve the quality of JD Daojia home logistics and distribution services, and give priority to improving the important quality factors that affect consumer satisfaction. This paper uses the Kano model method and based on the data obtained from the questionnaire survey to establish a JD Daojia logistics distribution service quality improvement model to obtain the priority of the improvement of the delivery service quality factors, and obtains the distribution service quality elements of JD Daojia, which provides certain practical guidance for improving the service quality of JD Daojia logistics. Nowadays, in real life, there are more complicated and uncertain factors that affect the quality of JD Daojia’s logistics and distribution services. This article is just a few attempts to improve the quality of JD Daojia’s logistics and distribution services, and some preliminary results have been obtained. Therefore, Further research on the influence of JD Daojia’s delivery service quality is of practical significance to the study of factors that new retail enterprises improve the quality of logistics services. Under the conditions of continuous advancement of policies, continuous advancement of technology, and continuous growth of consumer demand, the integration of online and offline, new retail online platforms such as JD Daojia will usher in a golden period of development, and will better satisfy people’s consumption diversified demands on the Internet and bring better user experience.
References 1. Chen, K., Jin, J., Luo, J.: Big consumer opinion data understanding for Kano categorization in new product development. J. Ambient. Intell. Humaniz. Comput. 13(4), 2269–2288 (2021). https://doi.org/10.1007/s12652-021-02985-5 2. 2019 China Real-time Logistics Industry Research Report [DB/OL]. http://report.iresearch. cn/report_pdf.aspx?id=3415 3. kano model[DB/OL]. https://en.wikipedia.org/wiki/Kano_model 4. Cho, J.H., Kim, B.S.: Determining via the Kano Model the Importance of Quality Characteristics in Construction Management. KSCE J. Civ. Eng. 2022(26), 2555–2566 (2022)
546
Y. Xu and X. Tong
5. Chen, W.-K., Chang, J.-R., Chen, L.-S., Hsu, R.-Y.: Using refined kano model and decision trees to discover learners’ needs for teaching videos. Multimedia Tools and Applications 81(6), 8317–8347 (2022). https://doi.org/10.1007/s11042-021-11744-9 6. Ponnam, A., Sahoo, D., Balaji, M.: Satisfaction-based segmentation: Application of Kano model in Indian fast food industry. J. Target Meas Anal. Mark 2011(19), 195–205 (2011) 7. Florez-Lopez, R., Ramon-Jeronimo, J.M.: Managing logistics customer service under uncertainty: An integrative fuzzy Kano framework. Inf. Sci. 2012(202), 41–57 (2012) 8. Shi, Y., Peng, Q.: Enhanced customer requirement classification for product design using big data and improved Kano model. Adv. Eng. Inf. 2021(49), 101340 (2021) 9. Choudhury, D.K., Gulati, U.: Product attributes based on customer’s perception and their effect on customer satisfaction: the Kano analysis of mobile brands. Decision 47(1), 49–60 (2020). https://doi.org/10.1007/s40622-020-00233-x 10. Chen, M.-C., Hsu, C.-L., Lee, L.-H.: Investigating pharmaceutical logistics service quality with refined Kano’s model. J. Retailing Consumer Serv. 2020(57), 102231 (2020) 11. Suh, Y., Woo, C., Koh, J., Jeon, J.: Analysing the satisfaction of university–industry cooperation efforts based on the Kano model: a Korean case. Technol. Forecasting Soc. Change 2019(148), 119740 (2019) 12. Gulc, A.: Models and Methods of Measuring the Quality of Logistic Service. Procedia Eng. 2017(182), 255–264 (2017) 13. Zheng, B., Wang, H., Golmohammadi, A.-M., Goli, A.: Impacts of logistics service quality and energy service of Business to Consumer (B2C) online retailing on customer loyalty in a circular economy. Sustainable Energy Technol. Assessments 2022(52), 102333 (2022) 14. Do, Q.H., Kim, T.Y., Wang, X.: Effects of logistics service quality and price fairness on customer repurchase intention: the moderating role of cross-border e-commerce experiences. J. Retailing Consumer Serv. 2023(70), 103165 (2023)
CiteSpace Based Analysis of the Development Status, Research Hotspots and Trends of Rural E-Commerce in China Yunyue Wu(B) School of Transportation, Nanning University, Nanning 530200, China [email protected]
Abstract. Rural e-commerce is an important way to promote the circulation of rural economy and increase the income of rural population, and it is also a new important economic growth point in China. This paper uses Citespace 6.1 software and CNKI database as the literature source to retrieve 471 core journals and CSSCI journal papers related to rural e-commerce research in China. It analyzes the time distribution, authors, spatial distribution, keywords, etc. and forms a knowledge map, so as to obtain the research status and research trends in rural ecommerce in China. The following results are obtained through analysis: China’s rural e-commerce research is on the rise, and the current research focuses are rural revitalization, e-commerce, rural economy, targeted poverty alleviation and e-commerce poverty alleviation. In the future, the research focus will tend to farmers’ income, coordinated development, rural development, trade circulation, informatization and some other aspects. Keywords: Rural e-commerce · Rural vitalization · CiteSpace · Rural commercial circulation
1 Introduction The vast countryside provides a large market for commercial circulation, and the development of e-commerce is no exception [1, 2]. Rural modernization cannot exist without digital construction, and the development of the digital economy requires digital rural development as well. Rural e-commerce development can drive agricultural modernization through informatization, transform the mode of rural economic growth, raise rural residents’ income levels, and encourage migrant workers to return home for employment and entrepreneurship [3]. It is critical to improve the quality and efficiency of agricultural economic operations and to better understand the market’s decisive role in resource allocation [4, 5]. Previous studies have fully recognized the significance of rural e-commerce development for China’s economic and social development. However, the foundation for China’s rural e-commerce development is very uneven, with the “digital divide” between urban and rural areas, the “last mile” of logistics, rural e-commerce talents, and other issues remaining unresolved [6, 7]. As a result, using the Citespace visualization tool and content analysis research methods, this paper combs the research © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 547–557, 2023. https://doi.org/10.1007/978-3-031-36115-9_50
548
Y. Wu
literature of rural e-commerce in China in recent years and creates a visual chart, with the goal of analyzing the current research status and research hotspots of rural e-commerce and providing a reference for rural e-commerce research.
2 Research Design 2.1 Research Methods CiteSpace software is used to conduct quantitative analysis on the reference data of rural e-commerce research in China. Starting from the number of annual papers published, the number of source journals and colleges, the number of documents cited, and the cooperative research of multiple authors, the research status of a selected topic is explored. Then, through keyword co-occurrence, clustering, highlighting and other analysis functions, the hot spots and development trends of rural e-commerce research in China are found. 2.2 Data Source Aiming at the theme of rural e-commerce development in China, this research uses the CNKI database as a literature source and conducts quantitative analysis on the literature using literature retrieval. The following are the specific operations: In the CNKI database, choose the “Advanced Search” type and the “Subject” search. The search term is “Rural E-commerce,” and the source of literature is academic journals. There is no time constraint. A total of 5239 journal papers were searched. Based on the quality of the literature, this study only selected the core journals of Peking University and CSSCI literature, removing irrelevant literature such as book reviews, news, enrollment brochures, and so on, to obtain 471 research papers.
3 Analysis of Research Status 3.1 Time Distribution Analysis The search revealed that Song Jian, the first core journal paper published on rural ecommerce development, was Research on the Impact of Broadband. Potential Development of China’s Rural E-commerce Logistics Market. According to statistics (Fig. 1), there were only one paper issued in 2014, and 62 in 2017. According to the trend, the number of documents issued in recent years has increased, which is consistent with the current domestic and international situation. The development of rural e-commerce has sparked heated debate in the academic community, particularly with the national development of the digital economy and the implementation of digital rural policies.
CiteSpace Based Analysis of the Development Status
549
Fig. 1. Time distribution analysis of rural e-commerce in China
3.2 Author Statistics and Cooperative Network Analysis CiteSpace is used to generate the author statistics and cooperation network diagram, select the authors whose threshold is greater than 2, and finally get the information of 29 authors, as shown in Fig. 2. It can be seen from the figure that Cao Lingling, Wang Yihuan and Nie Zhaoying have the largest number of articles, all of which are 4. At the same time, from the connection between the authors, Wang Yihuan and Nie Zhaoying were the two scholars who cooperated most closely.
Fig. 2. Author statistics and cooperative network analysis of rural e-commerce in China
550
Y. Wu
3.3 Analysis of Paper Publishing Journals and Cooperation Networks CiteSpace was used to generate the distribution map of source journals, and the paper publishing institutions with a threshold greater than 2 were selected, and the published papers of 27 organizations were finally obtained, as shown in Fig. 3. It can be seen from the figure that the largest number of papers published is China Agricultural University, with 9 papers published, followed by South China Agricultural University, with 5 papers published, followed by Yiwu Industrial and Commercial Vocational and Technical College, Shanghai Ocean University, and Shanghai Jianqiao University, with 3 papers published. From the connection point of view, the cooperation between publishing institutions is mostly based on regions, and universities in the same province and city cooperate closely.
Fig. 3. Cooperative network of publishing institutions of rural e-commerce papers in China
4 Research Hotspot and Trend Analysis 4.1 Keyword Co-occurrence Analysis The keyword co-occurrence analysis method is a direct statistics of the currently published literature. It seeks the topics focused on by the current papers and reflects the focus and hot spots after the trend is formed [8]. Through the co-occurrence analysis of keywords in rural e-commerce research literature, Fig. 4 is obtained. “Rural e-commerce” has the highest frequency and the largest nodes, covering the whole process of literature research. However, the source of this paper is “rural e-commerce” as the search subject, so it will not be discussed in this paper. After excluding “rural e-commerce”, the top five keywords are selected for ranking analysis according to the two indicators of keyword frequency and centrality, as shown in Table 1 and Table 2.
CiteSpace Based Analysis of the Development Status
551
Fig. 4. Key words of rural e-commerce research in China Table 1. High frequency keywords of rural e-commerce research in China Serial number
Hot words
Frequency of occurrence
Middle coefficient
1
rural revitalization
57
0.26
2
e-commerce
37
0.19
3
rural economy
21
0.16
4
targeted poverty alleviation
17
0.41
5
e-commerce poverty alleviation
16
0.06
Table 2. Key words in the research of rural e-commerce in China Serial number
Hot words
Frequency of occurrence
Middle coefficient
1
intenet+
8
0.88
2
rural development
2
0.65
3
taobao village
14
0.64
4
pattern
3
0.62
5
development
4
0.42
552
Y. Wu
It can be achieved from Table 1 that the high-frequency keywords are “rural revitalization”, “e-commerce”, “rural economy”, “targeted poverty alleviation” and “e-commerce poverty alleviation”. However, not all high-frequency keywords have high center coefficients, so we cannot judge research hotspots simply by keyword frequency [9]. From the perspective of center coefficient, “Internet plus”, “Rural Development” and “Taobao Village” are the network centers of rural e-commerce. Rural e-commerce development needs to rely on the development and upgrading of Internet technology. Rural e-commerce development will drive rural development. One of the important forms of rural e-commerce is the construction of Taobao Village. The central coefficient of “Internet plus” is 0.88, with the highest degree of support for the network, followed by “rural development” with the central coefficient of 0.65, and the third “Taobao Village” with the central coefficient of 0.64. Therefore, “Internet plus”, “Rural Development” and “Taobao Village” are hot spots in rural e-commerce research. 4.2 Keyword Emergence Analysis Emergent words refer to keywords that increase suddenly or appear more frequently in a certain period. The research frontier of the discipline can be determined by analyzing the emergence of keywords [8]. Through keyword emergence analysis of rural e-commerce research field (Figs. 5 and 6), 26 emerging words were obtained: e-commerce platform, mode, rural shop, rural Internet users, rural Taobao, development, development status, development mode, targeted poverty alleviation, Taobao village, industrial evolution, development path, dynamic mechanism, joint distribution, e-commerce poverty alleviation, shared logistics, rural poverty alleviation, innovative development, influencing factors, three rural issues, countermeasures Farmers’ income, coordinated development, rural development, trade circulation, and informatization. The research on rural e-commerce in China can be divided into three stages according to the emergence of key words. (1) Stage 1 The first stage is from 2014 to 2016. The research on rural e-commerce in China is still at the initial stage, with a weak foundation. In the process of development, it faces many difficulties, such as farmers’ low digital technology, inconsistent ideas with the development of rural e-commerce, and inadequate circulation channels, which restrict the development of rural e-commerce [10]. However, a lot of explorations have been carried out in various places. Most of these explorations are based on e-commerce platforms. For example, Alibaba, JD, Suning and other e-commerce platforms have their own rural e-commerce plans [11]. Rural Taobao has gradually developed into the most important form of rural e-commerce [12]. Taobao Village has become a typical case of rural ecommerce development mode, and the industry of “Taobao Village”, It takes only a few years to complete the industrial process that traditionally takes hundreds of years [13]. (2) Stage 2 The second stage is from 2017 to 2019. With the deepening of targeted poverty alleviation policies, scholars have put forward a lot of suggestions on how rural e-commerce can
CiteSpace Based Analysis of the Development Status
553
promote rural revitalization and achieve targeted poverty alleviation, such as the government should specifically support featured e-commerce projects, rural e-commerce talent introduction strategies, improve the talent security mechanism, and create new ecommerce brands [14]. However, to achieve sustainable development, rural e-commerce needs to be driven by market demand, national policy guarantee, farmers’ willingness, resource supply regeneration and other forces [15]. However, rural e-commerce also faces the problem of distribution, especially in the context of sharing economy, how to optimize distribution will be the focus of future research [16]. (3) Stage 3 The third stage is from 2020 till now. In 2020, China will build a moderately prosperous society in an all-round way, and rural development will focus on rural revitalization. At this stage, rural e-commerce focuses on issues such as farmers’ income, coordinated development, rural development, trade circulation, and informatization. Rural e-commerce has had a positive impact on improving farmers’ income, driving rural employment, attracting e-commerce entrepreneurship, reducing local unsalable agricultural products, targeted poverty alleviation and other aspects [17], maintaining the stable growth of farmers’ income, which is conducive to promoting the development of rural e-commerce [18]. The development of rural e-commerce can not be separated from the development of rural industries, and more importantly, it needs to be coordinated with the development of rural commercial infrastructure and rural consumption structure [1+]. In the context of rapid development of rural e-commerce, the development of rural e-commerce directly promotes the development of trade circulation industry. However, rural e-commerce can form a virtuous circle only when rural e-commerce enterprises,
Fig. 5. Key words of rural e-commerce research in China
554
Y. Wu
trade circulation industry and farmers are in a “win-win” situation [20, 21]. Especially in the context of the digital revolution, how to achieve industrial upgrading of rural e-commerce will be the focus of future research.
Fig. 6. Time zone chart of research keywords of rural e-commerce in China
4.3 Analysis Results and Development Strategies of Rural E-Commerce 4.3.1 Analysis Results and Limitations of Rural E-Commerce From the above analysis, it can also be seen that the development of rural e-commerce in China cannot be separated from the support of policies and the guarantee of systems. However, the development of rural e-commerce can not only rely on policies and systems, but the key is the ability of rural self-development. So, how to improve farmers’ awareness of integrating into e-commerce and improve farmers’ ability to develop e-commerce should be further studied. This paper uses CNKI database as the core of Peking University and CSSCI journal paper data source. The limitations of the research are: first, the literature source does not include other databases and general papers, and the domestic literature has not been comprehensively screened and combed. Second, the literature only analyzed the domestic research situation, and failed to analyze the foreign literature, and failed to compare the domestic and foreign literature.
CiteSpace Based Analysis of the Development Status
555
4.3.2 Important Problems to be Solved in Rural E-Commerce Development There are two important problems that need to be solved in the process of rural ecommerce development. (1) How to integrate small farmers into e-commerce The integration of small farmers into e-commerce can not only timely obtain market demand, consumer feedback and other information, but also play a positive role in improving the economic income of small farmers and promoting the promotion of agricultural products. However, small farmers are limited by their skills and knowledge, so it is worth further studying how to integrate into e-commerce. (2) Public service system of rural e-commerce The development of rural e-commerce requires the construction of infrastructure and the provision of corresponding public services by the government, such as personnel training, comprehensive agricultural services, and rural e-commerce planning. In the process of vigorously developing rural e-commerce, the country should improve the public service system of rural e-commerce and create a good entrepreneurial environment.
5 Research Conclusions and Prospects 5.1 Research Conclusion In this paper, Cite Space software is used to analyze and visualize the knowledge atlas and related data of literature generation on rural e-commerce of core journals and CSSCI journals in the CNKI database at different levels. The research draws the following conclusions: (1) From the perspective of time distribution, the research on rural e-commerce in China started in 2014 and started in 2017, showing a continuous upward trend in recent years. (2) From the perspective of spatial distribution, most of the researchers focus on rural e-commerce research in colleges and universities, but there is still a lack of research institutions and leaders with sufficient influence. Cooperation among research institutions is not close enough, especially cross regional cooperation is rare. (3) From the perspective of keyword co-occurrence, “Internet plus”, “rural development” and “Taobao Village” are the current research hotspots. (4) From the perspective of keyword emergence, rural e-commerce research has shifted from e-commerce platforms, models, rural stores, rural Internet users, and rural Taobao to farmers’ income, coordinated development, rural development, trade circulation, and informatization, which is consistent with the current social development trend. 5.2 Prospect of Future Research on Rural E-Commerce With the rapid development of We Media, farmers can also become the protagonists of rural e-commerce. The development of rural e-commerce requires more talents who
556
Y. Wu
understand technology and business. Future research can be carried out in the following directions. First, the cultivation of rural e-commerce talents, including farmers’ digital skills, rural e-commerce service talents, etc. Only by improving people’s ability can rural e-commerce be better developed. Second, the development path of rural e-commerce, especially in the era of popular short videos, how to give full play to the main role of rural Internet users and farmer anchors will be the focus of relevant research. Acknowledgment. This project is supported by 2022 Guangxi Vocational Education Teaching Reform Research Project “Effective Connection of Transportation and Logistics Major Group Undergraduate Program and Research and Practice of Curriculum System Construction” (GXGZZJG2022B181).
References 1. Koniew, M.: Classification of the user’s intent detection in e-commerce systems–survey and recommendations. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 12(6), 1–12 (2020) 2. Haji, K.: E-commerce development in rural and remote areas of BRICS countries. J. Integrative Agric. 20(4), 979–997 (2021) 3. Nurhuda, A., Khoirunnita, A., Rusli, A., et al.: Development E-commerce information system of agriculture in Samarinda. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 14(6), 46–54 (2022) 4. Kshetri, N.: Rural e-commerce in developing countries. IT Professional 20(2), 91–95 (2018) 5. Qin, N.: Quantitative research on China’s rural e-commerce policy text – content analysis based on policy tools and business ecosystem. Econ. Syst. Reform 04, 25–31 (2016). (in Chinese) 6. Lei, Z., Lei, H.: Does the development of e-commerce economy expand the income gap between urban and rural residents? Econ. Manage. Res. 38(05), 3–13 (2017). (in Chinese) 7. Baako, I., Umar, S.: An integrated vulnerability assessment of electronic commerce websites. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 12(5), 24–32 (2020) 8. Azam, A., Ahmed, A., Wang, H., et al.: Knowledge structure and research progress in wind power generation (WPG) from 2005 to 2020 using CiteSpace based scientometric analysis. J. Clean. Prod. 295, 126496 (2021) 9. Yue, C., Liu, Z.: The rising map of scientific knowledge. Sci. Sci. Res. 02, 149–154 (2005). (in Chinese) 10. Zhu, S.: The impact of rural E-commerce development on the logistics industry and the construction of rural logistics system. Price Monthly 03, 75–78 (2016). (in Chinese) 11. Dong, K., Hou, W., Ding, H., Wang, P.: Research on innovation oriented rural E-commerce cluster development - analysis based on Suichang Model And Shaji Model. Agricultural Econ. Issues 37(10), 60–69+111 (2016). (in Chinese) 12. Guo, C.: Analysis of rural E-commerce mode - based on the research of taobao village. Econ. Syst. Reform. 05, 110–115 (2015). (in Chinese) 13. Liu, Y., Chu, X.: Research on the industrial evolution of China’s “Taobao Village.” China Soft Sci. 02, 29–36 (2017). (in Chinese) 14. Lu, B.: Research on the dynamic mechanism and development path of rural e-commerce promoting rural revitalization. Agric. Econ. 2019(12), 129–130 (2019). (in Chinese) 15. Li, X.: Discussion on the dynamic mechanism of the sustainable development of rural ECommerce in the view of industrial Chain. Bus. Econ. Res. 2019(02), 73–75 (2019). (in Chinese)
CiteSpace Based Analysis of the Development Status
557
16. Song, L.: Research on rural E-Commerce joint distribution operation mode from the perspective of shared logistics. Bus. Econ. Res. 08, 132–135 (2019). (in Chinese) 17. Yi, F., Sun, Y., Cai, Y.: Evaluation of the policy effect of the government on promoting rural e-commerce development - empirical research from the “comprehensive demonstration of E-commerce in rural areas.” Nankai Econ. Res. 03, 177–192 (2021). (in Chinese) 18. Tang, H., Li, S.: E-commerce, poverty alleviation and rural revitalization: role and path. J. Guangdong Univ. Finance Econ. 35(06), 65–77 (2020). (in Chinese) 19. Liu, Y.: Research on the difference of the impact of rural E-commerce development on rural residents’ consumption structure from the perspective of double cycle. Bus. Econ. Res. 09, 64–68 (2021). (in Chinese) 20. Liu, Y.: Research on the price threshold effect of rural E-commerce development on the commercial circulation industry. Commercial Econ. Res. 01, 25–29 (2022). (in Chinese) 21. Aparco, R.H., Del Carmen Delgado Lim, M., Tadeo1, F.T., et al.: Sustainability of rural agribusiness through e-commerce information systems. IOP Conf. Ser. Earth Environ. Sci. (968), 012002 (2022)
The Optimization and Selection of Deppon Logistics Transportation Scheme Based on AHP Long Zhang1 and Zhengxie Li2(B) 1 School Logistics, Wuhan Technology and Business University, Wuhan 430065, China 2 Philippine Christian University Center for International Education, 1004 Manila, Philippines
[email protected]
Abstract. The selection and optimization of logistics transportation scheme is crucial to the operation of enterprises. Deppon Logistics Company is a leading enterprise in the logistics industry and occupies a pivotal position in the logistics industry. In recent years, Deppon Logistics performance is outstanding, but there are still many logistics transportation problems in the end of the distribution link, such as Poor timeliness of transportation, the express damaged, and cargo lost, restricting development of the company. In order to maintain the long-term stable development of enterprises, enterprises should start from the details, combing the existing transportation routes, planning optimize the logistics and transportation scheme, reduce logistics costs, improve logistics and transportation benefits. In this paper, the analytic hierarchy process is used to analyze the logistics transportation scheme of Deppon Logistics. Starting from the influential factors such as timeliness, economy and reliability of logistics transportation, the logistics transportation model is built to help Deppon Logistics find the most appropriate transportation scheme for the enterprise. Keywords: Logistics transportation · AHP · Transportation scheme
1 Introduction In the face of human society in the 21st century, e-commerce is developing rapidly and vigorously with the support of the Internet. The logistics industry closely related to e-commerce has deeply integrated into people’s daily life and become an important pillar industry of the national economy. The definition of the term “logistics” in China is “The physical flow of goods from the place of supply to the place of receipt. According to the actual needs, the basic functions of transportation, storage, handling, packaging, circulation processing, distribution and information processing are organically combined [1]. “What people refer to as “logistics” in daily life is just the flow of goods, while the real “logistics” is a huge industry, including complicated links and specific processes, which profoundly affects and changes the production and life of modern people and occupies an important share in the gross product of the national economy.
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 558–572, 2023. https://doi.org/10.1007/978-3-031-36115-9_51
The Optimization and Selection of Deppon Logistics Transportation
559
1.1 Status Analysis of Deppon Logistics Transportation Scheme Types of Deppon logistics transportation schemes, Deppon has logistics transportation schemes such as land transportation and air transportation, and its main services include the following main forms: precise air transportation, Precision Kahang, precise motor transport. Problems of Deppon logistics transportation, Unreasonable logistics and transportation handover links harm the interests of customers, In order to protect their own interests and avoid the responsibility for goods damage, some employees of Deppon repackaged the damaged goods and asked the consignee to sign for acceptance [2]. If the goods are found damaged after the customer signs, Deppon Logistics staff will shift the responsibility to the sender, which will disrupt the delivery link and damage the interests of the customer [3].
2 Deppon Logistics Transportation Plan Optimization and Selection Analysis 2.1 Introduction to the Analytic Hierarchy Process T.L.saaty first proposed analytic hierarchy process (AHP) in the 1970s, which is a qualitative and quantitative analysis method. It divides key factors in a problem into hierarchical structures such as target layer, criterion layer and scheme layer for qualitative and quantitative comprehensive analysis [4]. Use the analytic hierarchy process to conduct in-depth analysis and research on things, draw the hierarchy chart of things, use the judgment matrix containing the key factors to calculate the hierarchy weight, repeatedly carry out hierarchical iteration, according to the evaluation results, and finally get the best decision plan or the highest target ranking, shown in Fig. 1.
Fig. 1. Hierarchical structure model of analytic hierarchy process
The advantages of AHP are as follows: (1) Systematic. When the analytic hierarchy process is used to solve the problem, the problem is often regarded as a whole and analyzed from the point of view of the system. (2) Less information. According to the operation steps of analytic hierarchy process, the evaluation of a problem is not based on specific numbers, but on the importance standard identified by people, and converted into the calculation results of weight,
560
L. Zhang and Z. Li
through this process to solve some problems that are difficult to express data and statistics. (3) Simple and practical. Analytic hierarchy process has great advantages in processing uncertain and subjective information [5]. At the same time, AHP also has many disadvantages: 1) Unable to provide new solutions. Analytic hierarchy process can only find the optimal solution in the existing scheme. 2) Qualitative data contains subjective factors. Analytic hierarchy process (AHP) combines qualitative and quantitative methods to analyze data, which inevitably brings in subjective factors. 3) Over-emphasis on consistency testing. In the process of using the analytic hierarchy process, it is necessary to check the consistency of the matrix [6]. 2.2 The Mathematical Model of Analytic Hierarchy Process In the analytic hierarchy process, Table 1 is usually used to represent the weight comparison of two factors in the judgment matrix scaling table [7]. The nine-level scale table is shown in Table 1. Table 1. 9-level scale table of judgment matrix. Relative ratios Weight ratio meaning 1
Two evaluation indicators are equally important
3
The former is slightly more important than the latter
5
The former is significantly more important than the latter
7
The former is more important than the latter
9
The former is absolutely more important than the latter
2, 4, 6, 8
The median of two adjacent evaluations
Reciprocal
The evaluation index ai and aj is equal to aij when compared with aj , then aj is equal to 1 / aij when compared with ai
Through analytic hierarchy process, the problem can be transformed into a matrix containing the listed factors: ⎤ ⎡ a11 · · · a1n ⎥ ⎢ (1) A = ⎣ ... . . . ... ⎦ an1 · · · ann
where n is the number of factors, and aij is the weight ratio of ai and aj corresponding to this position. Then the sum product method is used to normalize the matrix A by column to get the matrix Z. According to the matrix Z, the relative weight ω of each factor can
The Optimization and Selection of Deppon Logistics Transportation
561
be obtained. Then, the consistency test of the matrix is carried out, and the formula is used to calculate the maximum eigenvalue λ: λmax =
n [Aω ]i i=1
(2)
nωi
Consistency index calculation formula: Ci =
λ−n n−1
(3)
Ci Ri
(4)
Consistency index calculation formula: CR =
where n is the order of matrix Z, Ri is the random consistency index, which can be obtained by referring to the table, as shown in Table 2. Table 2. Reference table of average random consistency index Ri n
1
2
3
4
5
6
7
8
9
Ri
0
0
0.52
0.9
1.12
1.26
1.36
1.41
1.4
When the calculated C R is less than 0.1, it is considered that the judgment matrix has consistency, and the normalized feature vector can be used as the weight vector; otherwise, the matrix needs to be rebuilt. A matrix is constructed for the weight of each factor compared with each scheme to form the matrix A1 , A2 , A3 … The An. Using the above method to calculate the weight of each matrix ωi , check the consistency of the matrix, this process is called hierarchical single sorting. Finally, the consistency test of total hierarchy ranking is carried out to complete the analysis of the problem by AHP [8]. 2.3 Analysis of Influencing Factors of Deppon Logistics Transportation Scheme 2.3.1 Transport Route Selection Principle The selection of transport path is to plan out one or more reasonable vehicle routes for all the places through [9]. Logistics transportation route selection is directly related to logistics cost and transportation convenience, which affects the transportation efficiency of logistics enterprises and customer service experience. According to the service characteristics of Deppon, it is concluded that the selection of logistics transportation routes should follow the four principles of timeliness, economy, stability and convenience [10]. (1) Timeliness. The timeliness of goods transportation is crucial for logistics enterprises. People have higher and higher requirements for service quality. The goods can not only be delivered, but also need to be delivered as soon as possible. Timeliness has become the focus of competition of logistics enterprises, speed is the advantage [11].
562
L. Zhang and Z. Li
(2) Economy. Transportation cost has always been an important expenditure of logistics enterprises, accounting for a large proportion of the total cost of logistics enterprises. In the guarantee of service quality at the same time, as far as possible to reduce logistics transportation costs, to achieve the purpose of economic is always the goal of logistics enterprises [12]. (3) Stability. Stability refers to the probability of accidents in the process of logistics transportation. For example, on the conventional highway transport route, whether there is a high incidence of accidents on this route, which will cause traffic jams and affect the passing of vehicles; Whether the operation is closed during special periods or festivals, and the traffic is prohibited; Whether it will be renovated and expanded in the short term [13]. (4) Convenience. Convenience refers to whether the route is convenient for logistics vehicles to transport goods [14]. Logistics freight vehicles are mostly medium and large heavy trucks, with long body, high roof, heavy cargo, wide road, height and weight limits, which lead to difficult vehicles, and will not be able to meet the principle of convenient transportation. Convenient transportation also requires the choice of a scientific and reasonable route, does not deviate from any point of the driving route, avoid secondary loading and unloading transportation, control time and economic costs [15]. 2.3.2 Transport Route Selection Principle According to the transport route selection principle, the influencing factors required by the analytic hierarchy process (AHP) are selected, namely the criterion layer of the analytic hierarchy process (AHP) [16]. Combined with practical investigation and theoretical analysis, the initial influencing factors selected are shown in Table 3. Table 3. Initial influencing factors of Deppon transport scheme selection Serial number
Influencing factors
1
The distance of the starting point and destination from the transport route
2
Vehicle operation and maintenance costs
3
Road smoothness
4
Distance of transport line
5
Cargo quantity
6
Type of goods and requirements
7
Consistency of transport
2.3.3 Analysis of Influencing Factors The selection of too many factors is not conducive to the analytic hierarchy process to solve the problem. Under normal circumstances, it is common to select 3–5 factors for
The Optimization and Selection of Deppon Logistics Transportation
563
evaluation. Therefore, the above initial influencing factors are summarized or screened, and the following four criteria are obtained [17]. (1) Transportation cost: refers to the total cost of logistics vehicles during the operation and maintenance period. (2) Time spent: it refers to the total time spent by logistics vehicles from the transfer field to the destination for unloading the goods. (3) Anti-risk ability: it refers to the ability of logistics enterprises to cope with emergencies and transport stability. (4) Cargo loading rate: refers to the cargo loading rate of the vehicle, that is, the utilization rate of the carriage space [18]. 2.4 Application of Analytic Hierarchy Process in Deppon Logistics Transportation Scheme Optimization In the analytic hierarchy process, we usually use Table 1 to represent the weight comparison of two factors in the judgment matrix scaling table. The nine-level scale table is shown in Table 1. (1) Use analytic hierarchy process (AHP) to establish hierarchy structure Logistics enterprises should first consider the problem of profit, the second is to serve the public. Therefore, logistics enterprises need to control logistics costs, but also to grasp the transportation time. In order to obtain the optimal transport scheme, logistics enterprises should consider transport path, transport time, transport stability and other factors to comprehensively screen out and optimize the logistics transport scheme [19]. This paper has selected four evaluation indexes of transportation cost, time spent, risk resistance ability and cargo loading rate as evaluation indexes of Deppon logistics transportation scheme optimization. Combining with the structural model and evaluation index system of analytic hierarchy process, the hierarchical structure model of Deppon logistics is obtained as shown in Fig. 1.
Fig. 2. Transport scheme selection hierarchy model
(2) Expert consultation method to determine the index weight The weight reflects the importance of each factor in the system. This paper uses the expert consultation method to determine the weight. Generally let the industry experts score the indicators, according to the results decide the weight [20]. In Deppon logistics
564
L. Zhang and Z. Li
optimization problem, after consulting 13 authoritative managers in the logistics industry, according to their opinions, the 9-level scale table is listed, and the final integration is as follows: Table 4. Table 4. Weight table of AHP criterion layer Index weight
Transportation cost Time spent Anti-risk ability Cargo loading rate
Transportation cost 1
1/3
2/3
4
Time spent
3
1
2
5
Anti-risk ability
2
1/2
1
3
Cargo loading rate
1/4
1/5
1/3
1
3 Solution and Application of Numerical Examples 3.1 General Situation of Logistics Transportation in a Certain Area of Wenzhou Through the understanding of analytic hierarchy process, we will select an example to analyze its application. For example, there are three Deppon business offices (hereinafter collectively referred to as Part A,B and C) in L Town,C County, Wenzhou City. Their location distribution is shown in Fig. 3. The logistics shuttles to the three business departments are independent of each other, that is, they all have their own logistics shuttles.
Fig. 3. Distribution map of Deppon business department in 1L town
Due to the small scale and small cargo volume of Part A, in order to avoid the excessive empty load rate, it has been sharing a vehicle with another small department in the neighboring town, tentatively called Part D, and its geographical location is shown in Fig. 4.
The Optimization and Selection of Deppon Logistics Transportation
565
Fig. 4. Location map of part D
Fig. 5. Bus route map
The starting point of the logistics shuttle bus is the transfer center of L Island, and its route is shown in Fig. 5. From the figure, it can be clearly seen that the shuttle bus starts from L Island, drives into the highway, and then you can choose to go down the highway Speed, drive west to Part A first, then east to Part D after unloading the goods of Part A, or get off the highway, directly to Part D, and then to Part A. In this process, the company does not have hard and fast rules about which department to go to first, the department manager can communicate with the driver which department to go to first, but it is also restricted by the loading order of the transfer field, because the goods of the two departments will be loaded first and then, distributed in the front and back sections of the bus carriage, there will be no mixed loading, and the first arrival is more conducive to unloading. 3.2 Examples of Three Logistics Transportation Schemes 3.2.1 Single Transport Scheme The scheme has a single shuttle bus for the transport of Part A. After the change of scale, the cargo volume of Part A is maintained at about 1 to 1.5 carloads. Without transporting the goods of Part D, one carload still cannot be fully loaded, but the remaining quantity is not much. The plan is to ship only one truckload per day, and the rest of the goods will be shipped the next day, but the goods are overstocked. One temporary vehicle will
566
L. Zhang and Z. Li
be added every 3–5 days, and two truckloads will be shipped to Part A on another day. The feasibility of this scheme lies in the priority level of the goods. 3.2.2 Multi-vehicle Combined Transport Scheme This plan is a temporary plan adopted by Deppon after the change of the department. It remains unchanged with the same car of Department D. If the car of the main department cannot be loaded, the shuttle bus of the other department will be loaded with additional part. As a result, there is an extra train and all the goods can be sent through the transfer yard to the terminal delivery department on the same day. 3.2.3 Multi-vehicle Single Transport Scheme In this scheme, two buses are directly assigned to Part A, so that the goods cannot be loaded, and there is no need to share the bus with other departments to cause some potential risks, such as wrong unloading and leakage of the goods. In the time arrangement more freedom, do not have to consider the shuttle bus arrival time of other departments, the disadvantage is high transportation costs. 3.3 Comparative Analysis of Three Logistics Transportation Schemes 3.3.1 Transportation Cost Analysis After the per-selection scheme is defined, the analytic hierarchy process is used to analyze the different schemes one by one. The first is the single transport scheme (hereinafter referred to as Scheme 1). From the perspective of transport cost, Scheme 1 uses A shuttle bus specifically for the transport of Part A, and then sends an extra car to clean up the surplus every once in a while. If the cost of a car is expressed as c, the transport cost of Scheme 1 is (1 + 0.25) c. Compared with scheme 1, multi-vehicle combined transport scheme (hereinafter referred to as Scheme 2) adopts the combined transport with Part D. The actual transport cost borne by Part A is 0.1c, excluding part of the cost of the shuttle bus going to Part D normally, and the transport cost is (1 + 0.1c) with the addition of one additional bus in Part A. Multi-vehicle single transport scheme (hereinafter referred to as Scheme 3) directly add two shuttle buses, the cost of which is 2c. 3.3.2 Take the Time to Analyze In scenario 1, if only one bus arrives, the usual shuttle bus will leave at 9 am, take 1.5 h and arrive before 11 am, plus it takes an average of 1 h for the Courier to unload a load of goods, so the total time is about 2.5 h. Scheme two due to the intervention of the car, it will inevitably produce time conflict, such as the first bus unloaded, the second bus has not arrived, the Courier must wait in the department, cannot start delivery in time, otherwise the shuttle bus arrived no delivery. Or two buses arrive at the same time, unloading busy. These situations do occur in real work. Therefore, in terms of time cost, even in the best case, the second bus will arrive
The Optimization and Selection of Deppon Logistics Transportation
567
immediately after the first bus is unloaded. In this way, apart from the 2.5 h required in scheme 1, it still needs 30 min for the second bus to unload (because the second bus is a combined bus, with only half or even less cargo), which takes about 3 h. Similarly, Plan 3 and Plan 2 take similar time, but because the bus is not combined, the shuttle bus can stay in the department until the evening before departure, so the time spent is not as urgent as Plan 2, relatively better than plan 2, but the absolute time spent is still higher than plan 1. 3.3.3 Risk Resistance Analysis Obviously, the weakest risk prevention is the first plan, which is prone to customer dissatisfaction due to not receiving goods on time. In the special period, when the goods increase abruptly, a large amount of goods will be piled up and cannot be transferred in time, resulting in loss of customer experience. Scheme 2 and Scheme 3 have strong anti-risk ability and can alleviate these problems to varying degrees. Scheme 3 has the strongest anti-risk ability because both vehicles belong to Part A with large freedom. 3.3.4 Cargo Loading Rate Analysis Scheme one has only one car, which can be filled with the whole car every time, and there is no insufficient utilization of space. In scheme 2, the shuttle bus shared with Part D can also guarantee the full load, but another shuttle bus separately transporting the goods of Part A cannot be fully loaded, so there is a waste of space. Scheme three wastes more space.
4 Introduction of Calculation 4.1 Model After sorting out the above analysis, the weight scoring table of each factor is obtained through pairwise comparison, as shown in Table 5, 6, 7 and 8. Table 5. Comparison of the relative importance of transport costs A1
Scheme 1
Scheme 2
Scheme 3
Scheme 1
1
1/2
3
Scheme 2
2
1
5
Scheme 3
1/3
1/5
1
568
L. Zhang and Z. Li Table 6. Comparison of the relative importance of time spent A2
Scheme 1
Scheme 2
Scheme 3
Scheme 1
1
4
3
Scheme 2
1/4
1
1/2
Scheme 3
1/3
2
1
Table 7. Comparison of the relative importance of risk resistance A3
Scheme 1
Scheme 2
Scheme 3
Scheme 1
1
1/3
1/7
Scheme 2
3
1
1/3
Scheme 3
7
3
1
Table 8. Comparison of the relative importance of cargo loading rate A4
Scheme 1
Scheme 2
Scheme 3
Scheme 1
1
1/2
5
Scheme 2
2
1
7
Scheme 3
1/5
1/7
1
Firstly, the information obtained by expert scoring method is used to establish a judgment matrix for the pairwise comparison of transportation cost, time spent, risk resistance ability and cargo loading rate. ⎡
⎤ 1 1/3 2/3 4 ⎢ 3 1 2 5⎥ ⎥ A=⎢ ⎣ 2 1/2 1 3 ⎦ 1/4 1/5 1/3 1
(5)
And the weight judgment matrix between each factor: ⎡
⎡ ⎤ ⎤ 1 1/2 3 1 4 3 A1 = ⎣ 2 1 5 ⎦A2 = ⎣ 1/4 1 1/2 ⎦ 1/3 1/5 1 1/3 2 1 ⎡ ⎡ ⎤ ⎤ 1 1/3 1/7 1 1/2 5 A3 = ⎣ 3 1 1/3 ⎦A4 = ⎣ 2 1 7 ⎦ 7 3 1 1/5 1/7 1
(6)
The Optimization and Selection of Deppon Logistics Transportation
569
4.2 Weight Calculation First, calculate the maximum average value of the factor matrix of the criterion layer and verify the consistency of the matrix. Step 1: Calculate the weight order corresponding to the pairwise comparison ⎡ ⎤ ⎤ ⎡ ⎤ 0.8273 1 1/3 2/3 4 0.1996 ⎢ 1.9594 ⎥ ⎢ 3 1 2 5⎥ ⎢ 0.4641 ⎥ ⎢ ⎥ ⎥ ⎢ ⎥ A=⎢ ⎣ 2 1/2 1 3 ⎦ω = ⎣ 0.2617 ⎦→ Aω = ⎣ 1.1168 ⎦ 0.3046 1/4 1/5 1/3 1 0.0747 ⎡
(7)
Step 2: Calculate the maximum eigenvalue λmax =
n [Aω ]i i=1
nωi
= 4.1789
(8)
Step 3: Calculate the consistency index C i and the consistency ratio C R λ−n = (4.1789 − 4)/(4 − 1) = 0.0596 n−1
Ci =
(9)
C R = 0,0596/0.9 = 0.0662 < 0.1, meet consistency check. Table 9. Transport cost index comparison matrix, weight and consistency test table Transportation Cost
Scheme 1
Scheme 2
Scheme 3
Weight
Scheme 1
1
1/2
3
0.3092
Scheme 2
2
1
5
0.5813
Scheme 3
1/3
1/5
1
0.1096
Consistency Check
λ = 3.0036 Ci = 0.0018 Ri = 0.52 CR = 0.0035 Test Passed
Table 10. Comparison matrix, weight and consistency test table of time spent index Time Spent
Scheme 1
Scheme 2
Scheme 3
Weight
Scheme 1
1
4
3
0.6232
Scheme 2
1/4
1
1/2
0.1373
Scheme 3
1/3
2
1
0.2395
Consistency Check
λ = 3.0183 Ci = 0.0091 Ri = 0.52 CR = 0.0176 Test Passed
In the same way, the other items are checked for consistency, as shown in Table 9, 10, 11 and 12.
570
L. Zhang and Z. Li
Table 11. Comparison matrix, weight and consistency test table of risk resistance indicators Anti-risk Ability
Scheme 1
Scheme 2
Scheme 3
Weight
Scheme 1
1
1/3
1/7
0.0882
Scheme 2
3
1
1/3
0.2431
Scheme 3
7
3
1
0.6687
Consistency Check
λ = 3.0070 Ci = 0.0035 Ri = 0.52 CR = 0.0067 Test Passed
Table 12. Comparison matrix, weight and consistency check table of cargo loading rate indicators Cargo Loading Rate
Scheme 1
Scheme 2
Scheme 3
Weight
Scheme 1
1
1/2
5
0.3338
Scheme 2
2
1
7
0.5907
Scheme 3
1/5
1/7
1
0.0755
Consistency Check
λ = 3.0141 Ci = 0.0070 Ri = 0.52 CR = 0.0136 Test Passed
Table 13. Calculation table of total hierarchical sorting Z
ω
Scheme 1
Scheme 2
Scheme 3
Transportation Cost
0.1996
0.3092
0.5813
0.1096
Time Spent
0.4641
0.6232
0.1373
0.2395
Anti-risk Ability
0.2617
0.0882
0.2431
0.6687
Cargo Loading Rate
0.0747
0.3338
0.5907
0.0755
0.3989405
0.287432
0.31362756
Final Weight
Table 14. Consistency check of the total hierarchical order Ci
ω
A1
0.0018
0.1996
A2
0.0092
0.4641
A3
0.0035
0.2617
A4
0.0071
0.0747
Total Hierarchical Sort
Ci = 0.0061 Ri = 0.52 CR = 0.0116 < 0.1
The Optimization and Selection of Deppon Logistics Transportation
571
4.3 Hierarchical Sorting and Order The overall hierarchical ranking of the comparison of all influencing factors can be obtained as follows, shown in Table 13 and Table 14. As can be seen from Table 13 above, Scheme 1 has the greatest weight, and through calculation in Table 14, its consistency test passes and meets the requirements. In summary, this paper presents a solution to the selection of Deppon logistics terminal transportation scheme in Wenzhou by using analytic hierarchy process (AHP), and makes a specific analysis according to the special situation of the region. Finally, it is calculated that scheme 1 has the highest weight and is the preferred scheme among the current three schemes.
5 Conclusion Logistics transportation scheme is of great importance to the development of logistics enterprises. Choosing the appropriate logistics transportation scheme is conducive to the logistics enterprises to reduce fixed asset investment, avoid operating risks, improve profits, reduce operating costs, and improve the competitiveness of enterprises. This paper takes the AHP as the main tool, Deppon terminal transportation scheme as the target body, uses the AHP to determine the index weight, Combination of qualitative and quantitative analysis, builds a set of evaluation index system, combines the case analysis, provides a scientific basis for logistics enterprises to make decisions. However, due to the expert scoring method used in the selection of index weights, it is inevitable that there are subjective factors, which lead to inaccurate data analysis and affect the analysis results. The author will focus on how to improve and avoid the influence of human factors on the analysis results, so as to make the analysis results more accurate and convincing, provide more scientific basis for logistics enterprises to make decisions, and really help enterprises to reduce costs and increase efficiency.
References 1. Li, W.: Logistics and Transportation Management, p. 218. Science Press, Beijing (2021) 2. Ke, W.: Research on the location of Ya Long Logistics Park based on AHP. Kunming University of Science and Technology, Kunming (2017) 3. Cheung, Y., et al.: Adaptive decision evaluation method based on analytic hierarchy process. J. Beijing Univ. Aeronautics Astronautics 9, 79–80 (2016) 4. Bi Xin Hua et al.: Modern Logistics Management, p. 148. Science Press, Beijing (2021) 5. Lu Hua Wei: Study on transportation route optimization of international engineering bulk logistics. Dalian Maritime University, Dalian (2017) 6. Jin, Y.: Study on the production and countermeasures of logistics cost “Benefit reversal.” Hebei Enterprise 4, 14–15 (2016) 7. Gu, Q.: Application of Analytic Hierarchy Process in Supplier Selection of a Company. Shenzhen University, Shenzhen (2017) 8. Li, B.: Construction and implementation of logistics Business performance management system. Discuss Study 4, 138–139 (2021) 9. Yaling, X.: Ahp-based Evaluation of Zhoushan River-sea Combined Transport Public Information Platform. Zhejiang Ocean University, Zhoushan (2017)
572
L. Zhang and Z. Li
10. Le, W.: Research on Subject Service Evaluation of University Library Based on Analytic Hierarchy Process. Anhui University, Hefei (2017) 11. Aguezzoul, A.: Third-party logistics selection problem: a literature review on criteria and methods. Omega 16(5), 24–34 (2014) 12. Huang, M., Cui, Y., Yang, S., et al.: Fourth party logistics routing problem with fuzzy duration time. Int. J. Prod. Econ. 145(1), 2–3 (2013) 13. Liu, Y., Zhou, P., Li, L., Feng, Z.: Interactive decision-making method for third-party logistics provider selection under hybrid multi-criteria. Symmetry 11(5), 11–13 (2020) 14. Yong, W., Jie, Z., Guan, X., et al.: Collaborative multiple centers fresh logistics distribution network optimization with resource sharing and temperature control constraints. Expert Syst. Appl. 17(3), 165–166 (2021) 15. Fan, Z., Wang, L.: Evaluation of university scientific research ability based on the output of sci-tech papers: A D-AHP approach. PLoS ONE 17(2), 3–4 (2017) 16. Fan, G., Zhong, D., Yan, F., Yue, P.: A hybrid fuzzy evaluation method for curtain grouting efficiency assessment based on an AHP method extended by D numbers. Expert Syst. Appl. 11(3), 14–15 (2016) 17. Deng, X., Yong, H., Deng, Y., Mahadevan, S.: Supplier selection using AHP methodology extended by D numbers. Expert Syst. Appl. 17(1), 8–9 (2014) 18. Rabinovich, E., Bailey, J.P.: Physical distribution service quality in Internet retailing: service pricing, transaction attributes, and firm attributes. J. Oper. Manage. 11(6), 13–14 (2003) 19. Saaty, T.L.: Decision-making with the AHP: why is the principal eigenvector necessary. Eur. J. Oper. Res. 5(1), 6–7 (2002) 20. Zhong, X., Wen, Z., Wei, L., Wenjie, X.: Analysis of influencing factors of cold chain logistics cost of dairy products. Ind. Eng. Innov. Manage. 2(5), 3–4 (2022)
Pharmacological and Non-pharmacological Intervention in Epidemic Prevention and Control: A Medical Perspective Yanbing Xiong1 , Lijing Du1,2,3(B) , Jing Wang1 , Ying Wang4 , Qi Cai1 , and Kevin Xiong5 1 School of Safety Science and Emergency Management, Wuhan University of Technology,
Wuhan 430070, China [email protected] 2 School of Management, Wuhan University of Technology, Wuhan 430072, China 3 Research Institute of Digital Governance and Management Decision Innovation, Wuhan University of Technology, Wuhan 430072, China 4 School of Business, Wuchang University of Technology, Wuhan 430223, China 5 Information Technology Consulting Services, Ontario Limited, Markham, ON 1750351, Canada
Abstract. Since COVID-19 broke out in Wuhan, Hubei Province in 2020, COVID-19 has spread rapidly worldwide. Until today, the epidemic situation in some countries has not been effectively controlled. Therefore, the research on epidemic prevention and control has become a key research direction of global concern. At present, according to the classification and summary of this article, medical research on epidemic prevention and control measures can be divided into pharmacological intervention research and non-pharmacological intervention research. Researchers mainly study epidemic prevention and control from these two research directions. This paper briefly introduces the relevant research on epidemic prevention and control measures in medicine in the past five years, summarizes the current research status of epidemic prevention and control in medicine, classifies all papers, points out the shortcomings of existing research in the field of medicine, and provides future research ideas for researchers in the field of medicine. At the same time, draw some useful experience in operation management to help the government and enterprises improve production and supply chain management, so as to better promote economic recovery under the condition of ensuring epidemic prevention and control. Keywords: COVID-19 · Epidemic prevention and control · Vaccine Intervention · Social restrictions · Isolation
1 Introduction Since the outbreak of novel coronavirus in 2020, the death toll has been increasing rapidly. This public health emergency has caused varying degrees of damage to the economy, society and politics of various regions. Therefore, in order to reduce the negative impact of the epidemic on social and economic losses, it is necessary to intervene © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 573–582, 2023. https://doi.org/10.1007/978-3-031-36115-9_52
574
Y. Xiong et al.
in the development and spread of the epidemic and prevent the further spread of the epidemic. A large number of researchers began to study the epidemic prevention and control measures in the medical field, from the aspects of pharmacological intervention and non-pharmacological intervention, to study the role of various vaccines and drugs, as well as other non-pharmacological measures for epidemic prevention and control. This paper collects and collates the research on epidemic prevention and control in the medical field in the past five years, summarizes the current research status of epidemic prevention and control, and provides reference for future research directions of epidemic prevention and control. 1.1 Selection of Journals The time span of this review is nearly 5 years. The journals included in this part of the study are listed below in alphabetical order: Cell, Nature, Science, MECS, the Journal of American Medical Association, The Lancet, The New England journal of medicine. In this study, for each journal and paper, there is at least one of the following keywords in the research text, which is considered as the potential research object of this review: “pandemic” “infection” “epidemic” “influenza” “flu” “epidemiological” “infectious” “contagious” “SARS” “H1N1” “MERS” “Ebola” “COVID-19” “cholera” “plague” “intervention” “control” “prevention” “mitigate” “mitigation” “mitigating” “evolve” “evolution”. In the process of preliminary search and screening, 551 research papers with the above keywords were published in six journals. Each paper was reviewed by three authors, and after repeated screening, 94 related papers were finally obtained. Through specific reading of the papers, 72 research papers highly related to epidemic prevention and control were obtained, including 54 in the past five years. 1.2
Chronology of Growth in Epidemic Prevention and Control Research
Figure 1 shows the number of papers by journal and year. From this table, we can see that the journal with the largest number of papers is The Lancet (17), followed by The Journal of American Medical Association (10), Science (9), The New England journal of medicine (7), Nature (7) MECS (3) and Cell (1). The numbers in brackets indicate the number of papers. In order to study the growth rate of research literature on epidemic prevention and control in the medical field, we also used a three-year moving average. Table 1 and Fig. 2 describe the three-year moving average of research literature on epidemic prevention and control. From the figure, we can see that the three-year moving average in 2021 is 8.00 (higher than any previous year), which is due to the publication of 21 papers in 2020. By 2022, the three-year moving average will continue to rise to 14.00, and this also takes into account that we only collected papers in the first half of 2022. These figures clearly show that researchers in the medical field are increasingly interested in epidemic prevention and control.
Pharmacological and Non-pharmacological Intervention
575
Fig. 1. Count of journals and papers
Table 1. Year and count of papers Year 2018
Count
Three-year moving average
3
0.33
2019
0
1.00
2020
21
1.00
2021
23
8.00
2022
7
14.67
Total
54
2 Pharmacological Intervention Table 2 lists all pharmacological intervention papers by year of publication, disease type and control measures. From the perspective of disease types, since the outbreak of COVID-19 in 2020, almost all the papers related to epidemic prevention and control are about COVID-19. Only before 2020 have scholars studied other types of diseases, but few [1–3]. Specifically, pharmacological intervention can be divided into vaccines and other drugs. The research of vaccine intervention mainly focuses on the effectiveness and safety of vaccine treatment for various diseases, which is verified by randomized controlled trials [4, 5], or based on data analysis [6–15]. Other pharmacological intervention studies are aimed at other drugs other than vaccines, and other drug studies are basically verified by means of controlled trials [16–30]. In addition, there are studies on the use of ECMO in the treatment of COVID-19 [31]. Through these studies, the effectiveness and safety of various vaccines and drugs against different diseases have been verified. Among them, the RECOVERY Collaborative Group represented by Peter W Horby published a
576
Y. Xiong et al.
Fig. 2. Three-Year moving average Table 2. Research related to pharmacological intervention Year
Disease Type
Control Measures
Authors
2022
COVID-19
vaccination
Analía Rearte et al. (2022) Makrufa Sh. Hajirahimova et al. (2022)
medicine
Peter W Horby et al. (2022) Mark N Polizzotto et al. (2022) Peter W Horby et al. (2022) Eduardo Ramacciotti et al. (2022) M.J. Levin et al. (2022)
2021
COVID-19
vaccination
Caroline E. Wagner et al. (2021) Raphael Sonabend et al. (2021) Nathan D. Grubaugh et al. (2021) Alasdair P S Munro et al. (2021) Alice Cho et al. (2021) M.G. Thompson et al. (2021) Ioannis Katsoularis et al. (2021) Ewen Callaway(2021)
2021
COVID-19
medicine
Peter W Horby et al. (2021) Ly-Mee Yu et al. (2021) Christopher C Butler et al. (2021) Peter W Horby et al. (2021) Peter W Horby et al. (2021) Daniel R. Kuritzkes (2021) O. Mitjàet al. (2021) Courtney Temple et al. (2021)
ECMO
Ryan P Barbaro et al. (2021)
(continued)
Pharmacological and Non-pharmacological Intervention
577
Table 2. (continued) Year
Disease Type
Control Measures
Authors
2020
COVID-19
vaccination
Saad B. Omer et al. (2020) Eric J. Rubin et al. (2020)
medicine
Peter W Horby et al. (2020) Mary Marovich et al. (2020)
Tuberculosis Infection 2018
Mosquito-borne diseases
D. Ganmaa et al. (2020) vaccination
M. tuberculosis Infection
Neil M. Ferguson (2018) E. Nemes et al. (2018)
total of six papers related to it from 2020 to 2022, respectively verifying the effectiveness of aspirin, Tocilizumab, Azithromycin, Lopinavir ritonavir, convalescent plasma of COVID-19 patients, and Casivivimab and imdevimab in the treatment of COVID-19 [15, 17, 20, 23, 24, 28]. Moreover, the trials in Casivivimab and imdevimab proved for the first time that antiviral treatment can reduce the mortality of patients hospitalized for COVID-19. Therefore, the current research of pharmacological intervention is mainly to evaluate the effectiveness of vaccines and drugs against the epidemic situation through control experiments or actual data, which is also the focus of this research. Through this research, we can find out the targeted vaccines and drugs as soon as possible to control the epidemic.
3 Non-pharmacological Intervention Table 3 lists all non-pharmacological intervention papers by year of publication, disease type, control type and control measures. Like pharmacological intervention research, non-pharmacological intervention related research is also mainly focused on COVID19. It is mainly concentrated in 2020, the year when COVID-19 just broke out. Before 2020, the research on this aspect was seriously insufficient. After 2020, due to the effective control of the epidemic, it gradually lost interest. Of the 22 non-pharmacological intervention related studies we collected, 20 were about COVID-19, and only 2 were about other infectious diseases [32, 33]. The research on COVID-19 basically adopts the method of establishing mathematical models or analyzing data to evaluate the effect of various control measures [8, 34–52]. Only two studies have different directions, and study the detection and tracking of epidemic situation [53, 54]. In addition, according to the different functions of the control measures, we divide the control measures into the following four categories: isolation (case isolation and contact isolation), social restrictions (closing schools, entertainment places, prohibiting large gatherings, etc.), traffic restrictions (travel restrictions and traffic restrictions, etc.) and personal protection (wearing masks, washing hands, etc.). The current research of non-pharmacological intervention is mainly to evaluate the impact of different control measures on the evolution of the epidemic by analyzing the measures taken in the epidemic area and the epidemic related data. This research needs
578
Y. Xiong et al.
to analyze the data generated by different control measures, but due to the particularity of the epidemic, researchers are difficult to control the epidemic area and implement a single measure, so it is impossible to evaluate the effect of a specific measure. Through this research, we can evaluate the effect of different prevention and control measures, find out the effective epidemic prevention and control measures, and inhibit the evolution of the epidemic. Table 3. Research related to non-pharmacological intervention.
4 Limitations and Directions for Future Research The above research also has some limitations. First of all, because the data used by researchers are mainly provided by the government or relevant departments, some data may inevitably be lost or distorted due to the severity of the epidemic or the effectiveness of statistical capacity, leading to deviation of research conclusions. Secondly, the research of prevention and control measures needs to be based on the measures implemented by the government. Most governments take multiple measures to jointly implement the epidemic situation. The data obtained by researchers are generated by the joint implementation of multiple measures. It is difficult to obtain the data of the implementation of a single measure.
Pharmacological and Non-pharmacological Intervention
579
At present, the research on epidemic prevention and control mainly focuses on the evolution of the epidemic. In the future, we can try to study how the government should support and encourage enterprises to produce drugs and vaccines, and how to reasonably allocate limited medical resources in the face of the shortage of medical supplies at the beginning of the epidemic. In the case of insufficient medical supplies, how to promote enterprises to expand production and how to deal with the oversupply market situation after the epidemic. At the same time, how to reduce the impact of the epidemic on the economy and minimize the overall risk of the evolution of the epidemic and economic fluctuations of the impact of non-pharmacological interventions (such as home orders, traffic restrictions, etc.) on the social economy.
5 Conclusions In this study, we reviewed the epidemic prevention and control research published in six major medical journals in the past five years. The purpose of this study is to evaluate and present the current research situation in this field from a macro perspective, rather than simply analyze a single paper in depth. We hope to find out the limitations of current research and future research directions from these studies, and get some reference experience in operation and management, which can be used to improve production and supply chain management, so as to promote epidemic prevention and control and economic recovery. First of all, the research on epidemic prevention and control measures in the medical field analyzed in this paper can be divided into two categories: pharmacological intervention measures and non-pharmacological intervention measures. The research of pharmacological intervention measures can be divided into vaccines and other drugs. The research on vaccines is mainly to evaluate the effectiveness of different vaccines for COVID-19 treatment by analyzing the test data generated by the government’s vaccination activities. Different from the research of vaccines, the research of other drugs is mainly to judge the effects of different drugs by means of control experiments. The research of non-pharmacological intervention measures is mainly to evaluate the impact of different control measures on the evolution of the epidemic situation by analyzing the measures taken in the epidemic area and the epidemic related data, so as to provide reference for the epidemic prevention and control in all regions. Secondly, because various prevention and control measures (such as traffic restrictions, social restrictions, etc.) during the epidemic have had a huge impact on the global supply chain, leading to enterprises facing supply chain interruption, labor shortage and other situations, enterprises have to close down or even close down (especially manufacturing industry), resulting in market economic depression. Therefore, we give some suggestions for future research directions: (1) Combine epidemic prevention and control with operation management, and study how to improve production and supply chain management during epidemic prevention and control to reduce the impact of prevention and control measures on enterprises;
580
Y. Xiong et al.
(2) With the end of the epidemic, various prevention and control measures have begun to be lifted. How can the government effectively help enterprises speed up the pace of resumption of work and production, thus promoting economic recovery. (3) In the early stage of the epidemic, how to promote enterprises to expand the production of medical materials in the case of insufficient medical materials, and how to deal with the market situation of oversupply of medical materials in the case of overcapacity. Acknowledgment. This research was supported by the National Natural Science Foundation of China (No. 72104190) and the Humanities and Social Sciences Program of the Ministry of Education (No. 20YJC630018).
References 1. Davaasambuu, G., Buyanjargal, U., Xin, Z., et al.: Vitamin D supplements for prevention of tuberculosis infection and disease. N. Engl. J. Med. 383(4), 359–368 (2020) 2. Ferguson, N.M.: Challenges and opportunities in controlling mosquito-borne infections. Nature 559(7715), 490–497 (2018) 3. Nemes, E., Geldenhuys, H., Rozot, V., et al.: Prevention of M. Tuberculosis infection with H4:IC31 vaccine or BCG revaccination (Article). New England J. Med. 379(2), 138–149 (2018) 4. Rearte, A., et al.: Effectiveness of rAd26-rAd5, ChAdOx1 nCoV-19, and BBIBP-CorV vaccines for risk of infection with SARS-CoV-2 and death due to COVID-19 in people older than 60 years in Argentina: a test-negative, case-control, and retrospective longitudinal study. Lancet 399(10331), 1254–1264 (2022) 5. Munro Alasdair, P.S., Janani, L., et al.: Safety and immunogenicity of seven COVID-19 vaccines as a third dose (booster) following two doses of ChAdOx1 nCov-19 or BNT162b2 in the UK (COV-BOOST): a blinded, multicentre, randomised, controlled, phase 2 trial. The Lancet 398(10318), 2258–2276 (2021) 6. Wagner, C.E., Saad-Roy, C.M., Morris, S.E., et al.: Vaccine nationalism and the dynamics and control of SARS-CoV-2. Science 373(6562), 1488 (2021) 7. Sonabend, R., Whittles, L.K., et al.: Non-pharmaceutical interventions, vaccination, and the SARS-CoV-2 delta variant in England: a mathematical modelling study. Lancet 398(10313), 1825–1835 (2021) 8. Grubaugh, N.D., Hodcroft, E.B., et al.: Public health actions to control new SARS-CoV-2 variants. Cell 184(5), 1127–1132 (2021) 9. Cho, A., Muecksch, F., et al.: Anti-SARS-CoV-2 receptor-binding domain antibody evolution after mRNA vaccination. Nature 600(7889), 517–522 (2021) 10. Thompson Mark, G., et al.: Prevention and attenuation of COVID-19 with the BNT162b2 and mRNA-1273 vaccines. New England J. Med. 385(4), 320–329 (2021) 11. Katsoularis, I., Fonseca-Rodríguez, O., et al.: Risk of acute myocardial infarction and ischaemic stroke following COVID-19 in Sweden: a self-controlled case series and matched cohort study. Lancet 398(10300), 599–607 (2021) 12. Callaway, E.: Beyond Omicron: what’s next for COVID’s viral evolution. Nature 600(7888), 204–207 (2021) 13. Omer, S.B., Yildirim, I., Forman, H.P.: Herd immunity and implications for SARS-CoV-2 control. JAMA 324(20), 2095–2096 (2020)
Pharmacological and Non-pharmacological Intervention
581
14. Rubin Eric, J., Longo Dan, L.: SARS-CoV-2 vaccination - an ounce (actually, much less) of prevention. New England J. Med. 383(27), 2677–2678 (2020) 15. Hajirahimova, M.Sh., Aliyeva, A.S.: Analyzing the impact of vaccination on COVID-19 confirmed cases and deaths in azerbaijan using machine learning algorithm. Int. J. Educ. Manage. Eng. (IJEME) 12(1), 1–10 (2022) 16. RECOVERY Collaborative Group: Aspirin in patients admitted to hospital with COVID19 (RECOVERY): a randomised, controlled, open-label, platform trial. Lancet 399(10320), 143–151 (2022) 17. ITAC (INSIGHT 013) Study Group. Hyperimmune immunoglobulin for hospitalised patients with COVID-19 (ITAC): a double-blind, placebo-controlled, phase 3, randomised trial. Lancet (London, England) 399(10324), 530–540 (2022) 18. RECOVERY Collaborative Group: Casirivimab and imdevimab in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial. Lancet 399(10325), 665–676 (2022) 19. Eduardo, R., et al.: Rivaroxaban versus no anticoagulation for post-discharge thromboprophylaxis after hospitalisation for COVID-19 (MICHELLE): an open-label, multicentre, randomised, controlled trial. Lancet 399(10319), 50–59 (2022) 20. Levin, M.J., Ustianowski, A., De Wit, S., et al.: Intramuscular AZD7442 (tixagevimabcilgavimab) for prevention of COVID-19. N. Engl. J. Med. 386(23), 2188–2200 (2022) 21. RECOVERY Collaborative Group: Tocilizumab in patients admitted to hospital with COVID19 (RECOVERY): a randomised, controlled, open-label, platform trial. Lancet 397(10285), 1637–1645 (2021) 22. Ly-Mee, Yu., Bafadhel, M., et al.: Inhaled budesonide for COVID-19 in people at high risk of complications in the community in the UK (PRINCIPLE): a randomised, controlled, openlabel, adaptive platform trial. Lancet 398(10303), 843–855 (2021) 23. Butler, C.C., Dorward, J., Ly-Mee, Yu., et al.: Azithromycin for community treatment of suspected COVID-19 in people at increased risk of an adverse clinical course in the UK (PRINCIPLE): a randomised, controlled, open-label, adaptive platform trial. Lancet 397(10279), 1063–1074 (2021) 24. RECOVERY Collaborative Group: Azithromycin in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial. Lancet 397(10274), 605–612 (2021) 25. RECOVERY Collaborative Group: Convalescent plasma in patients admitted to hospital with COVID-19 (RECOVERY): a randomised controlled, open-label, platform trial. Lancet 397(10289), 2049–2059 (2021) 26. Kuritzkes, D.R.: Bamlanivimab for prevention of COVID-19. JAMA 326(1), 31–32 (2021) 27. Oriol, M., Marc, C.-M., et al.: A cluster-randomized trial of hydroxychloroquine for prevention of COVID-19. N. Engl. J. Med. 384(5), 417–427 (2021) 28. Temple, C., Hoang, R., Hendrickson, R.G.: Toxic effects from ivermectin use associated with prevention and treatment of COVID-19. N. Engl. J. Med. 385(23), 2197–2198 (2021) 29. RECOVERY Collaborative Group: Lopinavir-ritonavir in patients admitted to hospital with COVID-19 (RECOVERY): a randomised, controlled, open-label, platform trial. Lancet 396(10259), 1345–1352 (2020) 30. Marovich, M., Mascola, J.R., Cohen, M.S.: Monoclonal antibodies for prevention and treatment of COVID-19. JAMA 324(2), 131–132 (2020) 31. Barbaro, R.P., MacLaren, G., Boonstra, P.S., et al.: Extracorporeal membrane oxygenation for COVID-19: evolving outcomes from the international Extracorporeal Life Support Organization Registry. Lancet 398(10307), 1230–1238 (2021) 32. Campbell, K.: The art of infection prevention. Nature 586(7830), S53–S54 (2020)
582
Y. Xiong et al.
33. Lessler, J., Moore, S.M., Luquero, F.J., et al.: Mapping the burden of cholera in sub-Saharan Africa and implications for control: an analysis of data across geographical scales. Lancet 391(10133), 1908–1915 (2018) 34. Brauner, J.M., Mindermann, S., Sharma, M., et al.: Inferring the effectiveness of government interventions against COVID-19. Science 371(6531), 802 (2021) 35. Tenforde, M.W., Fisher, K.A., Patel, M.M.: Identifying COVID-19 risk through observational studies to inform control measures. JAMA 325(14), 1464–1465 (2021) 36. Brooks, J.T., Butler, J,C.: Effectiveness of mask wearing to control community spread of SARS-CoV-2. JAMA 325(10), 998–999 (2021) 37. Crane, M.A., Shermock, K.M., Omer, S.B., Romley, J.A.: Change in reported adherence to nonpharmaceutical interventions during the COVID-19 pandemic, April-November 2020. JAMA 325(9), 883–885 (2021) 38. Walensky Rochelle, P., del Rio, C.: From mitigation to containment of the COVID-19 pandemic: putting the SARS-CoV-2 genie back in the bottle. JAMA 323(19), 1889–1890 (2020) 39. Dehning, J., Zierenberg, J., Spitzner, F P., et al.: Inferring change points in the spread of COVID-19 reveals the effectiveness of interventions. Science 369(6500), 160 (2020) 40. Walker Patrick, G.T., et al.: The impact of COVID-19 and strategies for mitigation and suppression in low- and middle-income countries. Science 369(6502), 413–422 (2020) 41. Kraemer Moritz, U.G., Yang, C.-H., Bernardo, G., et al.: The effect of human mobility and control measures on the COVID-19 epidemic in China. Science 368(6490), 493–497 (2020) 42. Lerner, A.M., Folkers, G.K., Fauci, A.S.: Preventing the spread of SARS-CoV-2 with masks and other “low-tech” interventions. JAMA 324(19), 1935–1936 (2020) 43. Candido Darlan, S., et al.: Evolution and epidemic spread of SARS-CoV-2 in Brazil. Science 369(6508), 1255–1260 (2020) 44. Huaiyu, T., Yonghong, L., Yidan, L., et al.: An investigation of transmission control measures during the first 50 days of the COVID-19 epidemic in China. Science 368(6491), 638–642 (2020) 45. Parodi, S.M., Liu, V.X.: From containment to mitigation of COVID-19 in the US. J. Am. Med. Assoc. 323(15), 1441–1442 (2020) 46. Leung, K., Wu, J.T., Liu, D., Leung, G.M.: First-wave COVID-19 transmissibility and severity in China outside Hubei after control measures, and second-wave scenario planning: a modelling impact assessment. Lancet 395(10233), 1382–1393 (2020) 47. Lai, S., Ruktanonchai, N.W., Zhou, L., et al.: Effect of non-pharmaceutical interventions for containing the COVID-19 outbreak in China. Nature 585, 410–413 (2020) 48. An, P., Li, L., Chaolong, W., et al.: Association of public health interventions with the epidemiology of the COVID-19 outbreak in Wuhan, China. JAMA 323(19), 1915–1923 (2020) 49. Flaxman, S., Gandy, A., Zhu, H., et al.: Estimating the effects of non-pharmaceutical interventions on COVID-19 in Europe. Nature 584(7820), 257–261 (2020) 50. Ali, S.T., Wang, L., Lau, E.H.Y., et al.: Serial interval of SARS-CoV-2 was shortened over time by nonpharmaceutical interventions. Science 369(6507), 1106–1109 (2020) 51. Abriham, A., Dejene, D., Abera, T., et al.: Mathematical modeling for COVID-19 transmission dynamics and the impact of prevention strategies: a case of ethiopia. Int. J. Math. Sci. Comput. (IJMSC) 7(4), 43–59 (2021) 52. Tatkare, A., Patil, H., Salunke, T., et al.: COVID-19 patient health monitoring system. Int. J. Eng. Manuf. (IJEM) 11(5), 48–55 (2021) 53. Ferretti, L., Wymant, C., Kendall, M., et al.: Quantifying SARS-CoV-2 transmission suggests epidemic control with digital contact tracing. Science 368(6491), 619 (2020) 54. Pullano, G., Di Domenico, L., Sabbatini, C.E., et al.: Underdetection of cases of COVID-19 in France threatens epidemic control. Nature 590(7844), 134–139 (2021)
Passenger Flow Forecast of the Section of Shanghai-Kunming High-Speed Railway from Nanchang West Station to Changsha South Station Cheng Zhang, Puzhe Wei(B) , and Xin Qi School of Transportation Engineering, East China Jiaotong University, Nanchang 330000, China [email protected]
Abstract. Passenger flow is an important basis for the formulation of transportation organization schemes of a high-speed railway. A reasonable method of passenger flow forecasting can not only accurately predict the passenger flow of railway in the coming years, but also greatly reduce the difficulty and workload of forecasting. Thanks to its adaptive ability and self-learning ability, artificial neural network can be well applied to the passenger flow forecasting of high-speed railways. This study constructed an artificial neural network forecasting model and collected historical data of population, GDP, per capita disposable income, tourism and passenger flow of cities along the section of Shanghai-Kunming Highspeed Railway from Nanchang West Station to Changsha South Station for the training of the forecasting model. After comparative analysis with the exponential smoothing method and the time series method, this study used the regression analysis method to forecast the population, GDP, per capita disposable income, tourism and other data of cities along this route, and then forecast the middle and long-term passenger flow between Nanchang West Station and Changsha South Station, hoping to provide reference for the transportation organization of this section. Keywords: Passenger flow forecasting · Artificial neural network · Exponential smoothing · Shanghai-Kunming high-speed railway
1 Introduction In 2008, China opened its era of high-speed railways with the opening of the BeijingTianjin Intercity High-speed Railway. In the second decade of this century, China has built the Beijing-Shanghai, Shanghai-Kunming, Xuzhou - Lanzhou, Harbin - Dalian, Beijing - Zhangjiakou high-speed railways in succession. Passenger flow forecasting is the main basis for railway operation departments to coordinate their work, and is the basis for supporting the fine development of product design of railway passenger transportation. As more and more high-speed railways are opened, feasible methods for passenger flow forecasting and accurate passenger flow forecasting results are paid more © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 583–597, 2023. https://doi.org/10.1007/978-3-031-36115-9_53
584
C. Zhang et al.
and more attention by the organizers of high-speed railway operation [1]. By analyzing the passenger flow of Shanghai-Kunming High-speed Railway in previous years and other related indicators, this paper intends to forecast the overall trend of passenger flow in the medium and long term, so as to enable transportation organization personnel to formulate targeted transportation organization schemes for the high-speed railway, and tap the potential of the line and meet the travel needs of residents along the line [2].
2 Influencing Factors and Forecasting Methods of Passenger Flow 2.1 Analysis of Influencing Factors of Passenger Flow Forecast Passenger flow forecast must be related to the overall development of the cities. This paper will be based on the local population, regional gross national product, per capita disposable income of urban residents, tourism and other factors for an overall analysis [3]. 1) Local population 2) Level of economic development 3) Tourism 2.2 Commonly Used Methods of Passenger Flow Forecasting and Their Applicability Analysis Passenger flow refers to the number of passengers passing through a section of a line. It plays an important role in the planning of line networks, the setting of passenger station scale, the selection of passenger transport means and the formulation of operation schemes [4]. It depends on such factors as the industrial and agricultural conditions, the distribution of tourism resources, the living and income levels of urban and rural residents, as well as the developed degree of transportation network in the regions along the route of various means of transportation [5]. (1) The time series method The time series method refers to the application of the data analysis methods to forecast future trends through data series of a specific period. This method is particularly suitable for forecasting events that are in a continuous state. It requires a large amount of historical data, which is sequenced in chronological order. In other words, it uses the past to forecast the future [6]. Because many of the data are subject to certain uncertainties, the trends are further muddled and the accuracy of the forecasts is compromised. (2) The regression analysis method After simple processing of a large number of data, a regression equation of dependent variable and independent variable is obtained by the regression analysis method, and then known relationships are used to forecast the result [7]. The advantage of regression analysis is that it can show the significant relationships between independent variables and dependent variables. The relationship between independent and dependent variables can be many-to-one. However, for passenger flow forecast, the overall process of this
Passenger Flow Forecast of the Section of Shanghai-Kunming
585
algorithm is relatively simple, and the forecasting process is difficult to control. Most of the time, it is used to forecast with simply processed data, so as to approximately reflect the relationships between variables. (3) The exponential smoothing method The exponential smoothing method was proposed by Brown. He believed that the tendency of time series has stability or regularity, so time series can be reasonably extended along the tendency. He believed that the tendency in the recent past would to some extent affect the future [8]. The exponential smoothing method is a sort of moving average method. Its feature is to give different weights to the past observations. The weights of the recent observations are higher than those of the observations obtained earlier. According to the different smoothing times, exponential smoothing method can be divided into first exponential smoothing method, second exponential smoothing method and third exponential smoothing method. But they share the same idea: the forecast value is the weighted sum of previously observed values, and different weights are given to different data, with more weight given to new data and less weight given to old data [9]. (4) Artificial neural network. Artificial neural network is a mathematical model of distributed and parallel information processing based on the behavior characteristics of biological neural network. It is used to estimate or approximate a function [10]. Depending on the complexity of the system, the artificial neural network algorithm achieves the purpose of information processing by adjusting the interconnections between a large number of internal nodes. It is especially suitable for complex nonlinear structures. It is characterized by distributed storage and parallel cooperative management of information. While individual neurons are extremely simple in structure and limited in function, a large number of neurons can achieve a wide variety of behaviors. The unique ability of nonlinear adaptive processing of artificial neural network overcomes the defects of traditional artificial intelligence methods in intuitions, such as patterns, speech recognition, and unstructured information. As a result, it has been successfully applied in such fields as neural expert systems, pattern recognition, intelligent control, combination optimization, and forecasting [11]. Its adaptability depends on the results of a lot of training, and according to different external characteristics, the internal structure of the neural network will also change accordingly, which is a self-adapting ability [12]. The generation of railway passenger flow is mostly based on the data set of stations. Factors affecting passenger flow are changing with the development of society, and passengers’ travel behavior will also change accordingly. In addition, there are many internal and external factors affecting railway passenger flow, and the relationship between these factors is difficult to grasp accurately [13]. The entire transportation system is in a changeable and complicated state. In order to accurately combine these factors which are not closely related with the passenger flow forecasting, this paper selects the neural network method to forecast the passenger flow, and uses the mapping relationship between input and output data to forecast the future passenger flow.
586
C. Zhang et al.
3 Passenger Flow Forecasting for the Section of Shanghai-Kunming High-Speed Railway from Nanchang West Station to Changsha South Station 3.1 An Overview of the Shanghai-Kunming High-speed Railway Shanghai-Kunming High-speed Railway is referred to as Hu-Kun High-speed Railway for short. Making it the longest east-west high-speed railway in China and the one that passes through the most provinces among east-west high-speed railways. The Shanghai-Kunming High-speed Railway runs through three economic circles on the upper, middle and lower reaches of the Yangtze River. The whole line consists of the Shanghai-Hangzhou section, the Hangzhou-Changsha Section and the ChangshaKunming section. This paper has selected the section from Nanchang West Station to Changsha South Station for a case analysis (Table 1). Table 1. Information of stations in the section from Nanchang West Station to Changsha South Station Station
Grade
Scale of the Station
Nanchang West
Special grade
12 platforms and 26 lines
Gaoan
Third grade
2 platforms and 4 lines
Xinyu North
Second grade
2 platforms and 5 lines
Yichun
Second grade
Pingxiang North
Population
644
GDP
Per Capita Disposable Income
6651
5.04
530
4.04
120
1155
4.57
4 platforms and 10 lines
497
3191
3.99
Second grade
3 platforms and 7 lines
181
1108
4.34
Liling East
Third grade
2 platforms and 4 lines
825
4.84
Changsha South
Special grade
13 platforms and 28 lines
13300
6.21
74.5
88.6 1024
Passenger Flow Forecast of the Section of Shanghai-Kunming
587
In this section, there are 7 stations in total, with 21 OD. According to the comprehensive information in the table and the administrative level of the cities, the seven stations in the table have been divided into three types: large, medium and small, and six sections have been selected as the training sample: Nanchang West-Changsha South (large-large), Nanchang West-Yichun (large-medium), Nanchang West-Gaoan (largesmall), Xinyu North-Pingxiang North (medium-medium), Pingxiang North-Liling East (medium-small), Gaoan-Liling East (small-small). 3.2 A Passenger Flow Forecasting Method Based on the Artificial Neural Network Artificial neural network is a nonlinear function reflecting the mapping relationship between input and output. This relationship goes through a method similar to black-box processing, and after weight adjustment and the determination of the optimal hidden number, the accuracy of the mapping relationship is guaranteed [14]. At the same time, the training errors of the whole neural model will also be reflected through the mathematical model, so as to facilitate the subsequent adjustment and processing. Assuming that a total of K samples are input into the network, and the output of each sample is N, the overall error of the network can be constructed according to formula (1). K N 2 K=1 n=1 (δkn ) (1) δANN = 2 The number of hidden layer nodes can be determined by formula (2). √ N n1 = nm+ 2
(2)
Generally speaking, a neural network with 3 or more layers of neurons includes input layer(s), hidden layer(s) and output layer(s). The upper and lower layers are fully connected, the generated signals are transmitted forward, and the feedback error signals are propagated backward. The hidden layer(s) can be either single or multiple. For each training sample, the signal is propagated in turn, from input to output to error generation. 3.2.1 Preparation of the Input Data for Passenger Flow Forecast Several factors affecting the passenger flow have been mentioned above: population, economy, tourism and distance. In terms of the population, this study selects the total number of permanent residents along the routes; the most direct reflection of economic level is the regional GDP; in terms of the overall development of tourism, this paper quantifies it with the total tourism revenue; in passenger flow forecasting, distance is different from the other three input indicators, for it is a fixed value. After selecting the high-speed railway section for forecasting, this paper makes overall statistics on the distance between different places. The above four aspects serve as input indexes of the artificial neural model. The transfer process is as shown in Fig. 1.
588
C. Zhang et al. Input layer
Hidden layer
Output layer
Population
x
GDP
Passenger flow
y Tourism revenue
z Percapita disposable income
Fig. 1. Conduction diagram of artificial neural network
There are a total of 7 stations between Nanchang West Station and Changsha South Station. Limited by the length of this paper, between 2015 and 2019, only the input data of year 2019 is taken as the example, and the population, Gross Domestic Product (GDP), per capita disposable income, tourism revenue, and distances between stations are as shown in the table Table 2. Table 2. Input data of cities in the section from Nanchang West Station to Changsha South Station in 2019 Station
Population (ten thousand)
Nanchang West
Per capita disposable income (yuan)
Tourism revenue (100 million yuan)
560 88
5596
44136
1869
449
26790
90
119.6
972
30046
521
Yichun
505.4
2688
25597
1016
Pingxiang North
181.1
930
29019
520
Liling East
105
716
35679
83
Changsha South
839
11574
55211
2029
Gaoan Xinyu North
GDP (ten thousand yuan)
Note: The data in the table come from the statistical communiques of national economic and social development of different regions
Passenger Flow Forecast of the Section of Shanghai-Kunming
589
3.2.2 The Training of the Artificial Neural Network Forecasting Model (1) The determination of hidden layer nodes The hidden layer nodes cover weight parameters. The neural network plays a crucial role in data processing and storage. It is worth exploring how to give the appropriate layer value for the forecast data and forecast model. According to the summary of previous experts and scholars, researchers derive the number of hidden layer nodes mainly through formula method and with their experience. Generally, the number of hidden layers should not be too small, but too many hidden layers will also lead to a longer training time of the model and slower convergence. (2) The choice of the transfer function The transfer function is a mathematical model, which is mainly an algorithm to express the differential equation of the output variables and the input variables in the neural network. This study selects the transig function from the neural network toolbox. This function can improve the nonlinear mapping ability between input and output and enhance the accuracy of the algorithm. (3) The training of artificial neural forecasting models In the second stage, the researcher used the MATLAB artificial neural toolbox to input the original data from 2015 to 2018 to train the artificial neural network model. Population, GDP, per capita disposable income, tourism income and distance between a pair of cities are taken as nine input variables. The number of hidden layers is determined as 3 by the formula mentioned above, and the output variable is the passenger flow. The schematic diagram of the artificial neural network is as shown in Fig. 2.
Fig. 2. Schematic diagram of artificial neural network
Due to the constraint of the length, this paper only lists the passenger flow in 2019 forecast by the neural network model, and compares it with other methods in Table 3.
590
C. Zhang et al.
Table 3. The forecast passenger flow between Nanchang West Station-Changsha South Station in 2019 Nanchang West Nanchang West
Gaoan
Xinyu North
Yichun
Pingxiang North
Liling East
Changsha South
446520
649014
811889
723946
30297
1131922
44141
151390
25309
3305
75573
50673
49017
4049
237401
126311
12134
426250
23571
782000
Gaoan
353332
Xinyu North
612719
55202
Yichun
756952
149252
85626
Pingxiang North
669652
28464
60708
102982
Liling East 30720
3041
4500
8237
15733
Changsha South
81586
269834
434763
887162
1142427
496998 535655
3.3 Passenger Flow Forecast Based on the Exponential Smoothing Method The exponential smoothing algorithm is exactly referred to as the “three exponential smoothing”. Through a time window function, it gradually fits from history to the present to realize the forecast of the future. yt + m = 3yt − 3yt + yt + (6 − 5a)yt − (10 − 8a)yt + (4 − 3a)yt . am a2m2 + yt − 2yt + yt · 2(1 − a)2 2(1 − a)2
(3)
yt = ayt − 1 + (1 − a)yt − 1
(4)
Upper confidence limit: The upper limit of future trends does not exceed this line Lower confidence limit: The lower limit of future trends does not exceed this line Confidence interval: The interval in which future trends fluctuate Trend line: Future trends are most likely to follow the trend of this line At the same time, the three exponential smoothing algorithm can also analyze and follow the volatility and periodicity presented by historical data. It should be noted that seasonality is not necessarily cyclical in human life and work. This study plans to use the passenger flow data of 2015–2018 to forecast the passenger flow data of 2019 and it plans to compare and verify the forecasting effect. In addition, six representative intervals out of twenty OD intervals between Nanchang West and Changsha South will be selected for a case analysis (Tables 4, 5, 6, 7, 8 and 9).
Passenger Flow Forecast of the Section of Shanghai-Kunming
591
Table 4. Passenger flow sample between Nanchang West-Gaoan (large-small) Year
Value
Trend Forecast
Lower Confidence Limit
Upper Confidence Limit
2015
180952
2016
307089
2017
342143
2018 2019
418250
418250
418250
418250
444791
499041
448145
549937
Table 5. Passenger flow sample between Nanchang West-Yichun (large-medium) Year
Value
2015
398515
Trend Forecast
Lower Confidence Limit
Upper Confidence Limit
2016
532044
2017
604617
2018
711889
711889
711889
711889
2019
808374
815322
783469
847174
Table 6. Passenger flow sample between Nanchang West-Changsha South (large-large) Year
Value
Trend Forecast
Lower Confidence Limit
Upper Confidence Limit
2015
750107
2016
861616
2017
937563
2018
1031922
1031922
1031922
1031922
2019
1144047
1125809
1106710
1144908
Table 7. Passenger flow sample between Xinyu North-Pingxiang North (medium-medium) Year
Value
Trend Forecast
Lower Confidence Limit
Upper Confidence Limit
2015
23111
2016
31644
2017
36556
2018 2019
43017
43017
43017
43017
48826
49744
47692
51797
3.4 Passenger Flow Forecast Based on the Time Series Method The passenger flow fitting figure calculated according to the time series is shown in Fig. 3.
592
C. Zhang et al. Table 8. Passenger flow sample between Pingxiang North-Liling East (medium-small)
Year
Value
Trend Forecast
Lower Confidence Limit
Upper Confidence Limit
2015
5694
2016
10959
2017
14442
2018
20571
20571
20571
20571
2019
24111
25064
23999
26129
Table 9. Passenger flow sample between Gaoan-Liling East (small-small) Year
Value
2015
3107
Trend Forecast
Lower Confidence Limit
Upper Confidence Limit
2016
3454
2017
3250
2018
3305
3305
3305
3305
2019
3225
3378
3074
3682
Fig. 3. Passenger flow fitting diagram of time series
According to the fitting analysis of time series, the passenger flow forecasting formula of each representative interval can be obtained as follows: Nanchang West - Gaoan : y = 74695x + 125372 (R2 = 0.949)
(5)
Passenger Flow Forecast of the Section of Shanghai-Kunming
593
Nanchang West - Yichun : y = 101270x + 308593 (R2 = 0.9879)
(6)
Nanchang West - Changsha South : y = 92139x + 664954 (R2 = 0.9949)
(7)
Xinyu North - Pingxiang North : y = 6463x + 17425 (R2 = 0.9886)
(8)
Pingxiang North - Liling East : y = 4811.4x + 888 (R2 = 0.99)
(9)
Gaoan - Liling East : y = 39x + 3181.5 (R2 = 0.1232)
(10)
3.5 Comparative Verification of the Forecasting Effect Taking the section of Shanghai-Kunming High-speed Railway between Nanchang West and Changsha South in 2019 as an example, the specific error calculation results are as shown in Table 10. Table 10. Comparison of passenger flow forecasting errors of the exponential smoothing method, the time series method and the artificial neural network method Interval
Actual value
Forecast value of exponential smoothing method
Relative error
The forecast value of the time series method
Relative error
The forecast value of the artificial neural network method
Relative error
Nanchang West-Gaoan
444791
448145
0.75%
468847
5.41%
446520
0.39%
Nanchang West-Yichun
808374
815321
0.86%
814943
0.81%
811889
0.43%
1144047
1125809
1.59%
1125649
1.61%
1131922
1.06%
Xinyu North-Pingxiang North
48826
49744
1.88%
49740
1.87%
49017
0.39%
Pingxiang North-Liling East
24111
25064
3.95%
24945
3.46%
23571
2.24%
Gaoan-Liling East
3225
3378
4.74%
3377
4.71%
3305
3.1%
Nanchang West-Changsha South
Note: The actual passenger flow in the table comes from the railway operation departments
By comparing the passenger flow forecast data in 2019, it can be seen that the artificial neural network forecasting model is significantly superior to the other two methods. Its error rate is basically controlled within 3%, in line with the standard error rate of engineering forecasting (≤10%). According to Zhang Bomin’s research on the
594
C. Zhang et al.
passenger flow forecast of Shanghai-Nanjing Intercity Railway, it can be seen that the forecast results of this paper are not much different from those in Zhang’s research. The error accuracy can also meet the requirements of operation agencies in passenger transportation organization and marketing [15].
4 Forecast of Passenger Flow in Coming Years Based on the Artificial Neural Network 4.1 Input Data Forecast in the Forecast of Passenger Flow in Coming Years (1) The regression analysis method The regression analysis method is a method used to determine the relationship between independent variables and dependent variables. It determines a suitable mathematical model to approximately express the mean change relationship between variables. It performs simple processing on large amounts of data and establishes correlation equations. If dependent variables are represented in the expression as a first-order function of the independent variables, the equation is called a linear regression equation, otherwise it is called a nonlinear regression equation. According to the number of independent variables and dependent variables, regression can be divided into unary regression or multiple regression. The regression analysis method aims to find a relationship between variables and express this relationship through mathematical expressions. The general model of regression is: y = h(x) = ωx + b + ε
(11)
where, ω is the correlation coefficient, b is the offset, and ε is the error term. When unknown parameters ω and b are obtained, it means that the model is established. That is, given x, the corresponding y (input) value can be forecast. Using the method of regression analysis, this study will forecast the sample values of four input data for each region in the medium (2025) and long (2035) terms. (2) Input data forecast in the forecast of passenger flow in coming years. After randomly selecting four input indicators from 2015 to 2019, this study establishes four corresponding regression models: population, GDP, per capita disposable income, and tourism income. Their analysis of the forecast results are as shown in Table 11. In Table 11, R Error represents Residual Error, S R Error represents Standard Residual Error. Regression statistical results of the model: “With an observed value of 20, the linear regression coefficient was 0.999953, the fitting coefficient was 0.999905”, showing a high degree of fitting. In view of this, this model was used to forecast the input data of the forecast of passenger flow in 2025 and 2035. The forecast results are as shown in Table 12.
Passenger Flow Forecast of the Section of Shanghai-Kunming
595
Table 11. The analysis of prediction results of input index regression model from 2015 to 2019 Observed Value
Forecast R Error S R Y Error
Forecast R Error Y
SR Error
2017 127.7
4.65 1.22 Per capita 2015 27907.4 −114.43 −0.53 −1.48 −0.39 disposable 2016 23702.5 −90.51 −0.42 income 0.28 0.07 2017 27017.3 −129.32 −0.6
2018 484.5
−3.54 −0.93
Population 2015 914.3 2016 109.5
2019 39.5 GDP
Observed Value
2015 237
0.47 7.01
0.09 Tourism −0.33 revenue
−25.1
2017 312.6
−10.62 −0.14
2019 2043.9
97.67
0.45
2019 51024.2 −232.15 −1.07
0.12
2016 1405.1 2018 167
2018 33326.3 2015
1548.3
−5.29 −0.1
2016
98.6
−5.56 −0.1
2017
609.7
−13.69 −0.26
4.97
0.07
2018
136.4
−3.38 −0.06
11.12
0.15
2019
208.6
21.42
0.4
Table 12. Predicted value of input samples of Hangzhou-Changsha high speed railway in 2025 and 2035 Station
Nanchang West Gaoan
Population (ten thousand)
GDP (ten thousand yuan)
Per capita disposable income (yuan)
Tourism revenue (hundred million yuan)
2025
2035
2025
2035
2025
2025
2035
607
682
12247
62297
3881
7246
8135
2035 92910
88
89
750
1345
37735
56146
195
378
Xinyu North
123
129
1431
2207
49710
80924
1386
2630
Yichun
569
587
3966
6106
36119
53875
2059
3821
Pingxiang North
200
210
1015
1052
40641
60213
1029
1846
Liling East
104
102
967
1404
50471
75065
172
320
Changsha South
984
1226
16435
24242
77640
115638
2989
4616
4.2 Results of Passenger Flow Forecast in the Coming Years In the third stage, this study forecast the passenger flow of the section of ShanghaiKunming High-speed Railway from Nanchang West to Changsha South in the middle and long term. The four input indicators have been identified above, and the forecast passenger flows in 2025 and 2035 are as shown in the table below 13 (Tables 13 and 14).
596
C. Zhang et al.
Table 13. Passenger flow forecast OD of the section from Nanchang West to Changsha South in 2025 Station
Nanchang West
Nanchang West
Gaoan
Xinyu North
Yichun
Pingxiang North
Liling East
Changsha South
896947
981161
1176463
1004914
89642
1673852
153862
305199
55217
5085
155066
72603
75069
48294
437030
167196
165350
777483
169011
1460813
Gaoan
738006
Xinyu North
740370
78543
Yichun
1104007
128467
93117
Pingxiang North
1068171
158856
185081
Liling East Changsha South
218917
72599
2972
16430
8858
24513
2002808
150992
470274
926393
1473751
810114 1370406
Table 14. Passenger flow forecast OD of the section from Nanchang West to Changsha South in 2035 Station
Nanchang West
Nanchang West
Gaoan
Xinyu North
Yichun
Pingxiang North
1491838
2156428
2291762
1460454
152597
2370887
256156
591543
108256
10638
281150
81650
84158
76781
671027
228968
225614
978126
159362
2084259
Gaoan
953341
Xinyu North
802187
84795
Yichun
1272821
205987
120876
Pingxiang North
1559070
293177
505512
381671
Liling East
210563
10229
56010
20782
44662
Changsha South
5229507
436082
1198976
1683285
3255387
Liling East
Changsha South
1578272 2628266
5 Conclusion Based on the historical passenger flow data of the section of Shanghai-Kunming HighSpeed Railway from Nanchang West to Changsha South from 2015 to 2018, three exponential smoothing method, time series method and artificial nerve method are used for a comparative study on the methods of passenger flow forecasting. Through error analysis, it is found that the artificial neural network model fits the original passenger flow data
Passenger Flow Forecast of the Section of Shanghai-Kunming
597
from 2015 to 2018 well, and the error between the forecast data in 2019 and the real value is small. This proves the validity of the artificial neural network model in forecasting the passenger flow of the section of Shanghai-Kunming High-Speed Railway from Nanchang West to Changsha South. On this basis, this study uses the artificial neural network model trained by historical passenger flow data from 2015 to 2019 to forecast the passenger flow in the above section in 2025 and 2035, which can serve as a reference for the later expansion of the high-speed railway stations in the above-mentioned cities and the formulation of the high-speed train operation schemes. Due to the numerous influencing factors of passenger flow generation and the complex influencing relationship among each other, the accuracy of high-speed railway passenger flow prediction can be further improved by further subdividing the influencing factors of passenger flow generation in future studies.
References 1. Teng, J., Li, J.Y.: A forecasting method of short-term intercity railway passenger flow considering date attribute and weather factor. China Railway Sci. 41(05), 136–144 (2020). (in Chinese) 2. Abisoye Blessing, O., Umar, A., Abisoye Opeyemi, A.: Challenges of airline reservation system and possible solutions (a case study of overland airways). Int. J. Inf. Technol. Comput. Sci. (IJITCS) 9(1), 34–45 (2017) 3. Bahmani, Z., Ghasemi, M.R., Mousaviamjad, S.S., Gharehbaghi, S.: Prediction of performance point of semi-rigid steel frames using artificial neural networks. Int. J. Intell. Syst. Appl. (IJISA) 11(10), 42–53 (2019) 4. Awad, M., Zaid-Alkelani, M.: Prediction of water demand using artificial neural networks models and statistical model. Int. J. Intell. Syst. Appl. (IJISA) 11(9), 40–55 (2019) 5. Su, C.C.: Passenger Flow OD Prediction of Urban Rail Transit Based on Passenger Flow Trend Characteristics. Beijing Jiaotong University, Beijing (2020). (in Chinese) 6. Peng, J.Y.: Research on time sequence of rail passenger flow based on self-organizing mapping of neural network. Smart City 7(03), 123–124 (2021). (in Chinese) 7. Hassan, M.M., Mirza, T.: Using time series forecasting for analysis of GDP Growth in India. Int. J. Educ. Manage. Eng. (IJEME) 11(3), 40–49 (2021) 8. Vuuren, D.V.: Optimal pricing in railway passenger transport: theory and practice in The Netherlands. Transp. Policy 9(2), 95–106 (2002) 9. Friesz, T.L.: Transportation network equilibrium, design and aggregation: key developments and research opportunities. Transp. Res. Part B 19(5–6), 413–427 (1985) 10. Zhan, S., Wong, S.C., Lo, S.M.: Social equity-based timetabling and ticket pricing for highspeed railways. Transp. Res. Part A: Policy Practice 137, 165–186 (2020) 11. Patriksson, M.: The traffic assignment problem: models and methods. VSP BV, Utrecht, The Netherlands (1994). 1 12. Al-Maqaleh, B.M., Al-Mansoub, A.A., Al-Badani, F.N.: Forecasting using artificial neural network and statistics models. Int. J. Educ. Manage. Eng. (IJEME) 6(3), 20–32 (2016) 13. Mgandu, F.A., Mkandawile, M., Rashid, M.: Trend analysis and forecasting of water level in mtera dam using exponential smoothing. Int. J. Math. Sci. Comput. (IJMSC) 6(4), 26–34 (2020) 14. Ojo, J.S., Ijomah, C.K., Akinpelu, S.B.: Artificial neural networks for earth-space link applications: a prediction approach and inter-comparison of rain-influenced attenuation models. Int. J. Intell. Syst. Appl. (IJISA) 14(5), 47–58 (2022) 15. Zhang, B.M.: On short-term prediction of passenger flow of Shanghai-Nanjing intercity railway. China Railway (09), 29–33+42 (2014), (in Chinese)
The Effect of Labor Rights on Mental Health of Front-Line Logistics Workers: The Moderating Effect of Social Support Yi Chen and Ying Gao(B) School of Business, Sichuan University, Chengdu 610065, China [email protected]
Abstract. Logistics industry is a typical labor-intensive industry. It is of great significance to pay attention to and improve the mental health level of front-line logistics practitioners for the sustainable and stable development of large-scale labor resources in logistics industry. This paper adopts the random sampling method, takes the front-line logistics industry employees in Chengdu, Sichuan Province as the research sample, and collects relevant data through questionnaire survey. Through statistical analysis, it can be concluded that: Firstly, labor rights protection has a positive impact on the mental health of front-line logistics workers. Secondly, labor rights infringement has a negative impact on the mental health of front-line logistics workers. Thirdly, social support has a significant positive moderating effect on the influence of labor rights of front-line logistics workers on their mental health. Based on the research conclusions, it is proposed that frontline logistics practitioners should improve their own cultural quality, enhance their understanding of their own rights, and strengthen their awareness of rights protection. Logistics enterprises should improve their own law-abiding consciousness, optimize the front-line work environment and enhance the management suggestions of enterprise humanism. Keywords: Labor rights · Mental health · Social support
1 Introduction Logistics industry is an important part of our national economy. It goes deep into our daily life. Since the beginning of the 21st century, China’s logistics industry has maintained rapid development. In the first three quarters of 2022, the total amount of social logistics in China reached 247 trillion yuan, an increase of 3.5 percent year-on-year. With the development of the times and the promotion of economic globalization, the service level of Chinese logistics industry has been continuously improved, the technical level has been continuously improved, the development environment has been continuously improved, and the future development of logistics industry is expected. With the rapid development of the logistics industry, it has absorbed a large number of social labor employment, showing as a labor-intensive industry. By the end of 2019, the number of employment reached 774.71 million, among which the number of logistics employees was up to 51.91 million (an increase of 3.6% over 2016), accounting for 6.7%. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 598–607, 2023. https://doi.org/10.1007/978-3-031-36115-9_54
The Effect of Labor Rights on Mental Health
599
In the huge employment group of employees in the logistics industry, for those who work in the office every day, they have a better working environment, regular working hours, wages paid on time, etc., and the protection of labor rights is better and less infringed. However, front-line logistics practitioners are different. Their working environment is usually outdoors in cold and hot summer. Due to the particularity of the logistics industry, front-line logistics practitioners still have high work intensity, long working hours and low salary, which infringes their labor rights, thus affecting their living conditions. Therefore, there may be a direct relationship between the labor rights and mental health of front-line logistics workers. At present, there are less researches on labor rights, social support and mental health. In addition, studies on labor rights and social support have focused less on front-line logistics workers, whose labor rights are often not well protected or infringed due to the particularity of work content and nature. Based on this, this paper takes front-line logistics practitioners as the research object and includes labor rights, mental health and social support into the research model to explore the impact of labor rights on their mental health and whether social support plays a regulating role in this process, and then puts forward relevant management suggestions to help protect the labor rights of front-line logistics practitioners and improve their mental health level. To achieve the sustained, steady and healthy development of largescale labor resources in the logistics industry.
2 Literature Review and Hypothetical Inference 2.1 Labor Rights and Mental Health Zheng Guanghuai built a sociological framework to understand employees’ mental health based on the two dimensions of labor rights-management system and structureaction, and pointed out that the reality of widespread violation of labor rights would undoubtedly affect employees’ mental health [1]. Liu Linping et al. pointed out that labor rights, such as overtime hours, forced labor and working environment, have a significant impact on the mental health of migrant workers [2]. Yuan Huina pointed out that increasing participation in basic medical insurance can effectively guarantee migrant workers’ access to medical resources, protect their labor rights, and thus improve their health status [3]. Zhu Ling pointed out that workers with low hourly wage rate, no labor contract signed, and lack of negotiation rights in terms of wages and labor protection are more likely to work overtime, and overtime work and poor working environment will have a significant adverse impact on the normal health of these workers [4]. Marx’s theory of labor alienation pointed out that in order to obtain more abundant profits, capitalists constantly exploited laborers, and laborers could not feel the happiness of realizing their own value in the process of labor, but were trapped in deep misfortune, which not only damaged their bodies, but also destroyed their spirits [5]. In the process of pursuing the maximization of benefits, capitalists will reduce costs in all aspects, which leads to the overuse of limited labor force, and the labor rights of laborers are difficult to be adequately protected, which ultimately leads to the physical and mental damage of laborers. Based on Marx’s explanation of labor alienation and the research findings of various scholars, we propose the following research hypotheses:
600
Y. Chen and Y. Gao
H1: The labor rights protection of front-line logistics workers is positively correlated with mental health, that is, the higher the degree of labor rights protection, the better the mental state. H2: The violation of labor rights of front-line logistics workers is negatively correlated with mental health, that is, the higher the violation of labor rights, the worse the mental state. 2.2 The Moderating Effect of Social Support Some scholars conducted research based on the overall nature of social support. The study of He Xuesong et al. on migrant children found that the mental health of migrant children was significantly related to social support, and the improvement of social support could not only improve their mental health, but also reduce their depression [6]. He Xuesong et al. also pointed out that social support (such as support from family, friends and fellow villagers) is an important protective factor for mental health and has a positive and significant effect on mental health [7]. Another part of scholars focused on a certain aspect of social support. Liu Linping et al. pointed out that social support, especially social communication, has a significant impact on the mental health of migrant workers [2]. Ding Tenghui, based on the study of residents in five provinces, and pointed out that each dimension of interpersonal relationship has a significant impact on their mental health [8]. Guo Xinghua and Cai Fengwei conducted an empirical study on the new generation of farmers in Beijing and the Pearl River Delta region and pointed out that social interaction factors, especially group interaction, would affect the mental health of the group [9]. Frances Cullen, the author of social support theory, points out that the more developed and powerful a person’s social support network is, the better able he or she is to cope with challenges in the environment. Social support can positively regulate the mental health of individuals to a certain extent. Based on the social support theory and the research findings of various scholars, we propose the following hypothesis: H3: Social support has a positive moderating effect on the influence of labor rights of front-line logistics workers on their mental health.
3 Data Collection 3.1 Questionnaire Design and Chosen Scale On the basis of referring to the existing research and literature, the two variables of “mental health” and “social support” are measured, which are authoritative and proved to be suitable for our national conditions. For the variable of “labor rights”, it is divided into two variables, “labor rights protection” and “labor rights infringement”, neither of which has a systematic and mature scale at present. Therefore, this study compiled the measurement scale of this variable according to relevant research and combined with the relevant regulations of China’s Labor Law. The details are as follows: 1. Labor rights protection scale: The labor rights protection scale consists of two items, namely, the situation of signing labor contract and the situation of social insurance. The scale adopted Likert four-point scoring method, “1 = very bad situation”, “2 = poor situation”, “3 = good situation”, “4 = very good situation”.
The Effect of Labor Rights on Mental Health
601
2. Labor rights infringement scale: The labor rights infringement scale consists of 4 items, including salary, unpaid wages, overtime work and safety and health. The scale scoring rule is the same as 1. 3. Mental health scale: The mental health scale adopted the GHQ-12 scale, which was compiled by Goldberg D. et al. [10] and verified by Yang Tingzhong et al. and applicable to Chinese people, with a total of 12 items [11]. 4. Social support scale: The SSRS scale was developed by Professor Xiao Shuiyuan based on China’s basic national conditions [12], and verified by Liu Jiwen et al., which is suitable for Chinese research [13]. The scale consists of 10 items in three dimensions: objective support, subjective support and utilization of support. In addition, the questionnaire also involved the respondent’s gender, age, marital status, education level, annual income, job type and other personal characteristic variables. 3.2 Respondents This study randomly selected several logistics companies in Chengdu, Sichuan Province, and distributed questionnaires to front-line logistics workers in the companies. In order to improve the completeness of the questionnaire, five questionnaires including “Basic Information Questionnaire for front-line logistics Practitioners”, “Labor Rights protection Questionnaire”, “Labor rights infringement questionnaire”, “GHQ-12” and “SSRS” were combined into one questionnaire in order. Secondly, in order to improve the efficiency of questionnaire distribution, this study adopts the form of online questionnaire distribution, through the company’s project department manager to issue questionnaires to the work group for data collection. A total of 118 questionnaires were collected in this survey, and 107 valid questionnaires were obtained by deleting the non-research subjects and those with outliers, with an effective recovery rate of 90.68%.
4 Data Analysis and Research Hypothesis Testing 4.1 Data Analysis 4.1.1 Reliability and Validity Analysis of Questionnaire Reliability is used to test whether the measured results of the scale have good consistency and reliability. According to the test results in Table 1, the Cronbach’s alpha of the scale is 0.759, 0.882 and 0.842, with values greater than 0.7, so the scale has high reliability. Validity is used to test the extent to which the scale can accurately measure the real situation. KMO and Bartlett spherical tests are carried out on the scale, and the test results are shown in Table 1. The KMO of the scale were all greater than 0.7. Meanwhile, the significance of the Bartlett spherical test was 0.000. Therefore, the scale used in this study was suitable for factor analysis, and the effect was good. Principal component analysis method was used to extract the scale, and Caesar’s normalized maximum variance method was used to rotate them. The analysis results showed that all item loads in the variable scale were higher than 0.5. Therefore, all scales have good validity.
602
Y. Chen and Y. Gao Table 1. Reliability and validity analysis of questionnaire Questionnaire
Cronbach’s alpha
KMO
Significance
Labor rights
0.759
0.770
0.000
Mental health
0.882
0.846
0.000
Social support
0.842
0.789
0.000
4.1.2 Analysis of Correlation In order to further explore the interaction between variables, Pearson correlation analysis was carried out on variables, and the analysis results are shown in Table 2. According to the data in the table, in terms of the score, the relationship between labor rights protection and mental health (r = −0.372, P < 0.01), labor rights infringement and mental health (r = −0.582, P < 0.01), social support and mental health (r = −0.585, P < 0.01) had a significant negative correlation. Table 2. The mean value, standard deviation and correlation of each variable Variable
Mean value
Standard deviation
Labor rights protection
Labor rights infringement
Labor rights protection
6.86
1.47
0.600
Labor rights infringement
12.96
2.03
-
Mental health
17.30
4.80
–0.372**
–-0.582**
9.13
0.300**
0.438**
Social support 41.63
Mental health
Social support
0.770 0.882 –0.585**
0.842
** stands for P < 0.01 (two-tailed test), the bolded data on the diagonal are the Cronbach’s alpha
values of the scale.
4.2 Research Hypothesis Testing 4.2.1 The Relationship Between Labor Rights Protection and Mental Health Model 1 and Model 2 were constructed to conduct linear regression analysis, with gender, age, marital status, annual income, education background and job category as control variables, labor rights protection as independent variable and mental health as dependent variable. The results are shown in Table 3. Model 1 is the relationship between control variables and mental health, and all control variables have no significant impact on mental health. In Model 2, the protection of labor rights was added on the basis of Model 1. The result (R2 ) shows that the protection of labor rights could explain 13.8% of the fluctuation of mental health. In addition, the
The Effect of Labor Rights on Mental Health
603
labor rights protection of front-line logistics practitioners can significantly affect their mental health. In terms of the scores of the two variables, the influence is negative (β = −0.469, P = 0.000). H1 is assumed to be supported. Table 3. Analysis of regression Variable
Dependent variable: mental health Model 1
Model 2
Model 3
Male (Female = 0)
−0.065
−0.091
−0.011
18–35 years old (51–60 = 0)
−0.199
−0.164
−0.119
36–50 years old
0.014
0.007
−0.065
Married (Divorced = 0)
0.291
0.078
0.165
Control variables
Unmarried
0.275
0.066
0.185
40,000 yuan and below (over 100,000 yuan = 0)
0.023
−0.005
0.034
40,000–60,000 yuan
0.093
0.041
0.166
60,000–80,000 yuan
0.055
0.033
0.037
80,000–100,000 yuan
0.009
−0.043
0.013
Junior College (Bachelor’s degree or above = 0)
−0.057
−0.419
−0.120
High school
0.006
−0.234
−0.017
Junior High School and below
−0.213
−0.140
−0.089
General worker (Mechanic = 0)
0.106
0.062
0.093
Independent variables −0.469**
Labor rights protection
−0.582**
Labor rights infringement R2
0.091
R2 F
0.716
0.229
0.389
0.138
0.298
1.952
4.188
** stands for P < 0.01
4.2.2 The Relationship Between Labor Rights Infringement and Mental Health Similarly, Model 1 is the relationship between control variables and mental health, and Model 3 is based on model 1 by adding labor rights infringement. The results in Table 3 show that R2 changes from 0.091 to 0.389, indicating that labor rights infringement can explain 29.8% of mental health fluctuations. In addition, the violation of labor rights and interests of front-line logistics practitioners can significantly affect their mental health. In terms of the scores of the two variables, the influence is negative (β = –0.582, P =
604
Y. Chen and Y. Gao
0.000), that is, the higher the violation of labor rights of front-line logistics practitioners, the worse their mental state. H2 is assumed to be supported. 4.2.3 The Moderating Effect of Social Support According to the analysis results shown in Table 4, as far as the scores of each variable are concerned, labor equity has a significant negative impact on mental health (β = –0.615, P = 0.000), and social support also has a significant negative impact on mental health (β = −0.652, P = 0.000). The interaction variable (the product of labor equity and social support) has a positive moderating effect on the relationship between labor equity and mental health, and the effect is significant (β = 0.166, P = 0.047), so the hypothesis H3 is also supported. Table 4. Analysis of regulating effect Variable
Control variables: gender, age, marital status, salary, education background, job type β
T
P
Labor rights
−0.615
−6.803
0.000
Social support
−0.652
−7.722
0.000
0.166
2.015
0.047
Labor rights*Social support
5 Conclusion and Management Suggestions 5.1 Conclusion and Discussion According to the research results and the above discussion, this paper draws the following conclusions: Conclusion 1, labor rights protection of front-line logistics workers has a significant positive impact on mental health. Conclusion 2, Labor rights infringement of front-line logistics workers has a significant negative impact on mental health. Conclusion 3, social support can significantly affect the mental health of front-line logistics workers. Conclusion 4, social support increases the positive impact of labor rights of front-line logistics workers on mental health. 5.1.1 Labor Rights and Mental Health According to the research results, labor rights protection of front-line logistics workers has a significant positive impact on their mental health, while labor rights infringement has a significant negative impact on their mental health, which is consistent with the research results of scholars such as Liu Linping et al. [2], Sun Zhongwei et al. [14] and Zheng Guanghuai [1]. First of all, the unreasonable salary and salary arrears of front-line logistics workers, on the one hand, will increase their economic burden, so that they have to find ways
The Effect of Labor Rights on Mental Health
605
to maintain a normal life, whether it is to reduce the quality of life or borrow money from friends and relatives; On the other hand, their delayed payment of wages will make them produce a sense of insecurity and anxiety, and the resulting life and psychological pressure will significantly affect their mental health. Secondly, the daily workload of most front-line logistics practitioners is not small, and overtime work will make their body in a long-term fatigue or overload state, their body and mind can not relax, mental condition is difficult to develop well, and even may lead to industrial injuries. In addition, fewer or no labor contracts and social insurance will reduce the sense of belonging and security of front-line logistics workers, which will cause psychological stress and lead to some mental health problems. Therefore, the establishment of a sound social security system to protect the vital interests of front-line logistics practitioners is conducive to improving their mental health. Therefore, based on the discussion of the above results, it is of great significance to put forward relevant management suggestions to improve the labor rights and interests of front-line logistics practitioners. 5.1.2 The Moderating Effect of Social Support The above results confirmed the positive moderating effect of social support, that is, social support increased the positive impact of labor rights on mental health. This is consistent with the research results of scholars such as He Xuesong et al. [7], Zhang Lei and Chang Yuanyuan [15]. According to the buffer model of social support, social support can buffer the intermediate link between stressful events and disease acquisition. Based on the analysis of the research results of this paper, front-line logistics workers will have a certain degree of psychological pressure when their labor rights are violated, which will lead to some mental problems. If they can get some social support at this time, they will underestimate the harm caused by the violation of labor rights, so as to reduce the pressure. Maintain good mental health. Whether at work or at home, giving them support from others, such as help when they are in trouble and an enlightened ear when they are bored, helps them to solve problems and cope with difficulties and challenges. At the same time, encouraging them to interact with colleagues and neighbors may also help them maintain good mental health to some extent. 5.2 Management Suggestions From the perspective of front-line logistics practitioners and logistics enterprises, this paper puts forward the following suggestions: For front-line logistics practitioners, first of all, improve the cultural quality of frontline logistics practitioners. Secondly, improve the group’s understanding of their own rights, and clarify the legitimate rights they should enjoy as workers. In addition, mass media should be used to strengthen the awareness of rights protection of front-line logistics practitioners and provide them with ways and means of rights protection.
606
Y. Chen and Y. Gao
Finally, front-line logistics practitioners should strengthen social communication and communicate with people more. For logistics enterprises, firstly, raise the awareness of enterprises to abide by the law. At the same time, enterprises should also recognize the importance of front-line logistics workers to the enterprise. They are also a member of the company and the creator of enterprise wealth, and their legitimate labor rights as workers should be protected. Secondly, optimize the enterprise’s front-line working environment. This not only benefits employees’ physical and mental health, but also promotes sustainable development. Finally, promote the thought of enterprise humanism. On the one hand, this is beneficial to the work and mental health of front-line logistics practitioners; on the other hand, it also has positive significance for the long-term healthy development of enterprises.
References 1. Zheng, G.: Towards the sociological understanding of employees’ mental health. J. Soc. Sci. Res. 25(6), 201–222+245–246 (2010). (in Chinese) 2. Liu, L., Zheng, G., Sun, Z.: Labor rights and mental health: based on the questionnaire survey of migrant workers in Yangtze River Delta and Pearl River Delta. Sociological Res. 26(04), 164–184+245–246 (2011). (in Chinese) 3. Yuan, H.: Health and income of migrant workers: Evidence from the survey of migrant workers in Beijing. Manage. World 05, 56–66 (2009). (in Chinese) 4. Zhu, L.: Labor hours and occupational health of rural migrant workers. Soc. Sci. China 2009(01), 133–149+207 (2009). (in Chinese) 5. Karl, M.: Economic and philosophical manuscripts of 1844. People’s Publishing House, 50 (2014) 6. He, X., Wu, Q., Huang, F., Xiao, L.: School environment, social support and migrant children’s mental health. Contemp. Youth Res. 09, 1–5 (2008). (in Chinese) 7. He, X., Huang, F., He, S.: Urban and rural migration and mental health: An empirical study based on Shanghai. Sociological Res. 25(01), 111–129+244–245 (2010). (in Chinese) 8. Ding, T., Xiao, H.: The impact of interpersonal relationship on mental health: based on the empirical survey of residents in Hunan, Guangxi, Shanghai, Shandong and Gansu Provinces. Sci. Theory 31, 119–120 (2010). (in Chinese) 9. Guo, X., Cai, F.: Social interaction and mental health of the new generation of migrant workers: an empirical analysis based on the survey data in Beijing and the Pearl River Delta region. Gansu Soc. Sci. 04, 30–34 (2012). (in Chinese) 10. Goldberg, D.: The General Health Questionnaire, pp. 225–237. Oxford University Press, New York (1996) 11. Yang, T., Li, H., Wu, Z.: Study on the suitability of Chinese health questionnaire for screening psychological disorders in Chinese mainland population. Chinese J. Epidemiol. 09, 20–24 (2003). (in Chinese) 12. Xiao, S.: The theoretical basis and research application of the social support rating scale. J. Clin. Psychiatry 02, 98–100 (1994) 13. Liu, J., Li, F., Lian, Y.: Research on reliability and validity of social support rating scale. J. Xinjiang Med. Univ. 01, 1–3 (2008). (in Chinese) 14. Sun, Z., Li, Z., Zhang, X.: Working environment pollution, overtime work and mental health of migrant workers: from the perspective of “second strike” theory. Population Dev. 24(05), 14–23 (2018)
The Effect of Labor Rights on Mental Health
607
15. Lei, Z., Chang, Y.: Social support and mental health: an empirical survey of the new generation of migrant workers in six cities of Guangdong Province. Northwest Population 35(05), 102– 106 (2014). (in Chinese) 16. Kayla, H., Zauszniewski Jaclene, A.: Stress experiences and mental health of pregnant women: The mediating role of social support. Issues in Mental Health Nursing, 2019, 40(7) 17. Nakao, M., Yano, E.: A comparative study of behavioral, physical and mental health status between term-limited and tenure-tracking employees in a population of Japanese male researchers. Public Health 2, 373–379 (2006) 18. Shafiee, N.S.M., Mutalib, S.: Prediction of mental health problems among higher education student using machine learning. Int. J. Educ. Manage. Eng. (IJEME) 10(06), 1–9 (2020)
An Investigation into Improving the Distribution Routes of Cold Chain Logistics for Fresh Produce Mei E. Xie1 , Hui Ye1 , Lichen Qiao2(B) , and Yao Zhang3 1 School of Business Administration, Wuhan Business University, Wuhan, China 2 College of Life Sciences, Northeastern University, Shenyang, China
[email protected] 3 Technical University of Munich Asia Campus, Technical University of Munich (TUM),
Singapore, Singapore
Abstract. In today’s era, with the improvement of people’s living standards and national economy, people have higher requirements for the quality of fresh agricultural products. Unlike other general logistics products, fresh agricultural products (abbreviated as FAP) have unique characteristics that are prone to deterioration and damage, resulting in significant waste. From this point of view, improving the transportation, storage, and distribution of FAP has become an urgent issue for the cold chain logistics industry. Based on the current situation analysis, this article analyzes the characteristics of FAP cold chain logistics. Aiming at specific problems, this paper proposes an optimization scheme and improvement suggestions for the distribution path of FAP cold chain logistics based on mileage saving method. The purpose is to optimize the distribution path of FAP logistics, provide theoretical basis and empirical reference for the improvement of FAP cold chain logistics in China, to ensure the quality of cold chain logistics and improve the competitive characteristics of cold chain logistics. Keywords: Fresh agricultural products (FAP) · Cold chain logistics · Mileage Saving Method · Distribution path
1 Introduction With the growth of the economy and the development of living level, FAP has become an essential part of people’s table. With the increasing material demand of people and the development of the Internet, the traditional channels for purchasing FAP can no longer meet people’s needs. Therefore, the distribution service of FAP is an inevitable trend. First, Alibaba, JD, Suning and other companies entered the fresh market first, and then the Daily Fresh, Prosperous and Preferred, Dingdong to buy vegetables and so on. The distribution market of FAP accounts for a large part of the logistics industry. According to relevant data, the distribution scale of domestic agricultural products will exceed 170 billion yuan in 2022 (see Fig. 1), and the industry has broad prospects in the future [1–3]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 608–617, 2023. https://doi.org/10.1007/978-3-031-36115-9_55
An Investigation into Improving the Distribution Routes
609
Fig. 1. Distribution scale and forecast of agricultural products in China
In 2000s, the improvement of cold chain logistics has shown a significant upward trend. As the latest China Cold Chain Logistics Development Report (2022) shown by the China Federation of Logistics and Purchasing, the desire for cold chain logistics goes strong. As the report shown, the cold chain logistics market size in 2021 will reach 458.6 billion yuan, with a year-on-year growth of 19.65%. This article uses research methods such as literature research, case studies, field investigations, and mileage savings to mainly describe the characteristics of cold chain logistics for FAP, analyze the current situation of the logistics for FAP, and model and optimize the logistics distribution path for FAP. However, China has a large population, a small average cultivated area, and agricultural production and consumption are lower than the world average, making the demand for FAP even more urgent. In this context, the development of the logistics is particularly important [4–6].
2 Research Status at Home and Abroad 2.1 Domestic Research Status For the reason to reduce the losses in the logistics and distribution of FAP, reduce the cold chain logistics costs of FAP, and improve customer satisfaction, many experts have conducted research on the logistics of FAP from various aspects. Ma Shaohua, Zhang Feng, and Du Chengning pointed out that after agricultural products mature, they have not been properly handled in the transportation and distribution process, resulting in a lack of timeliness for agricultural products [7]. Wang Chenglin et al. (2020) designed a seamless connection of cold chain logistics networks from origin, trunk lines, warehousing to distribution through improved genetic algorithms in Research on Location Optimization of Fresh Logistics Distribution Network [8]. In the other aspect, the research on the development of FAP cold chain logistics and distribution network, scholars also studied the optimization of the distribution path of agricultural cold chain logistics. Tu Wenjing (2022), in the Optimization of Agricultural Product Logistics Distribution Path Based on the Mileage Saving Method - Taking S Company as an Example, used the mileage saving method to reduce the logistics distribution cost of FAP, improve the distribution efficiency of FAP, maximize the freshness and edible quality of enterprise FAP, and improve customer satisfaction [9].
610
M. E. Xie et al.
2.2 Research Status Abroad The research in other countries for agricultural logistics is mainly reflected in the influencing factors. Eksoz C conducted a more comprehensive analysis of the needs for the logistics in the judgment and adjustment of supply integration in the strategic partnership of the food chain, and believed that agricultural product production, cold chain circulation rate, etc. are reasons that affect the demand for the logistics of FAP [10]. Developed countries in Europe and the United States have relatively complete the logistics systems that control the wastage rate of FAP at 1% to 2% through differential pressure precooling. David Bogataj David Bogatai uses the MRP model to reduce losses during transportation in a network physical system based on the extended MRP model, and monitors the dynamic changes of agricultural products within their shelf life in real time to make sure the freshness of FAP in the distribution network transportation process and improve customer satisfaction [11]. In the process of reading a large number of documents, it is found that domestic and foreign scholars have conducted certain research on the logistics, but foreign scholars have conducted early research on agricultural the logistics, and their research on FAP cold chain logistics distribution strategies and optimization is comprehensive and indepth. In recent years, as the country attaches great importance to the development of cold chain logistics, China’s research on cold chain logistics distribution of agricultural products is gradually deepening.
3 Characteristics of Cold Chain Logistics of Fresh Agricultural Products There are some differences between FAP and ordinary agricultural products, mainly including fruits, vegetables, meat, aquatic products, dairy products, egg products and other primary unprocessed agricultural products that are easily affected by external factors such as temperature and time. The cold chain logistics system of FAP is shown in the Fig. 2.
Fig. 2. Cold chain logistics system for fresh agricultural products
At present, the development of cold chain logistics of FAP in China is characterized by diversification, which is embodied in the following aspects.
An Investigation into Improving the Distribution Routes
611
3.1 High Requirements for Logistics Efficiency The difference between FAP and ordinary commodities is that the freshness period of FAP is very short. Due to the particularity of FAP, the transportation time of the entire cold chain logistics must be minimized and the distribution must be timely, which puts forward higher requirements for distribution speed. Currently, China’s fresh e-commerce market is basically operated in small batches, so the requirements for time are more stringent. The fresh agricultural product industry also has increasingly high requirements for the timeliness of cold chain logistics [12]. 3.2 Strict Temperature Control Requirements The logistics of FAP is also very vulnerable to temperature during transportation. On the one hand, if the temperature is too high, the product will be heated to breed bacteria, causing corruption and deterioration; On the other hand, the temperature is too low, and some foods are easy to be frozen and cannot be eaten. Therefore, the logistics of FAP has a high temperature control, which requires that the whole logistics activities, from the collection, processing, packaging, loading and unloading of raw materials to the transportation and distribution of products, should be kept within a constant temperature control range [13]. 3.3 Wide Distribution of Supply Chain The logistics of FAP in China is large in scale, various in types, and relatively dispersed in production and sales regions. Fresh agricultural product production workers are mainly concentrated in rural areas or urban suburbs, and the distribution of personnel is relatively scattered. However, a large amount of consumption demand for FAP is mainly concentrated in cities, and the distance between production and marketing may even exceed 1000 km. There are many links and long distances in the whole supply chain, which greatly increases the difficulty of cold chain logistics transportation and management of cold raw and FAP, and is easy to cause large product loss in each circulation link [14]. 3.4 High Cost FAP are divided into: first, frozen and refrigerated products. FAP need to be stored at low temperature during transportation to ensure freshness, so the requirements for storage and transportation conditions are also higher. The second is perishable goods. This kind of goods usually have a short shelf life, so attention should be paid to preservation and preservation. The third is volatile products. The value of such agricultural products is relatively low, and they are vulnerable to losses due to the impact of the natural environment. Refrigeration equipment is required in all links of the cold chain logistics of FAP, so the construction and operation costs of the logistics system are far higher than those of ordinary logistics [15].
612
M. E. Xie et al.
4 Current Situation Analysis of Cold Chain Logistics of Fresh Agricultural Products Now that China has become the world’s second largest economy, the food issue is the top priority, so people are beginning to pay more attention to the quality of FAP. 4.1 Lack of Perfect Cold Chain Logistics System The entire process of FAP from origin to circulation, processing, packaging, storage, transportation, and retail needs to be in a suitable low-temperature control environment. Currently, only a few enterprises in China can independently carry out cold chain integrated logistics services such as warehousing, transportation, and distribution. Although some small and medium-sized enterprises have a certain number of refrigeration configurations and refrigeration transportation fleets, they have not yet formed an overall logistics service network, unable to meet the needs of the Chinese market. 4.2 Lack of Specialized Talents in Cold Chain Logistics At present, most internal employees in the cold chain industry are lack of professional knowledge. People engaged in the logistics industry need not only knowledge of logistics management, but also knowledge of agricultural product refrigeration technology and food characteristics. Nowadays, colleges and universities pay more attention to theoretical teaching, which leads to a certain gap between the professional knowledge and practical skills mastered by many students and the actual needs of enterprises, and cannot meet the needs of cold chain logistics management. The development of cold chain logistics needs the support of various new professional talents. 4.3 Backward Cold Chain Infrastructure The FAP cold chain logistics is a new way of logistics development. Fresh cold chain logistics started late, developed for a short time, facilities and equipment are relatively backward, and the technical level is low, which restricts the development of China’s market economy. However, cold chain logistics enterprises need to strictly control the storage environment of agricultural products and keep the temperature stable, otherwise it is easy to cause losses and economic losses. At present, the infrastructure and technology level of cold chain logistics in China is lower than that of developed countries significantly.
5 Optimization of Cold Chain Logistics Distribution Path of Fresh Agricultural Products 5.1 Problem Description Generally, FAP enterprises need to have cold chain distribution centers for product distribution. The distribution centers dispatch different types of cold chain logistics vehicles to deliver the products required by each distribution point. Assumption: The distribution
An Investigation into Improving the Distribution Routes
613
center can meet the different needs of each distribution point for cold goods and frozen goods; The geographical location of the distribution point remains unchanged, and the demand has been confirmed without change; On the basis of meeting the requirements of load capacity, demand and time, a logistics distribution path optimization model based on mileage saving method is established, and the feasibility of the algorithm is verified by an example. 5.2 Using Mileage Saving Method to Optimize Distribution Path Specific example: A fresh agricultural product distribution center needs to distribute products to six sales regions, and the P0 logistics center is the starting point for distribution operations to the six distribution points of ABCDEF.
Fig. 3. Distribution Center and Distribution Points
Assumption: the load capacity of the vehicle is 10 tons, and the distribution distance limit is 80 km. The following figure shows the route distance (data on the straight line) between P0 and six distribution points, and the demand of each distribution point (data in parentheses), presented in Fig. 3. Initial plan: Logistics distribution center P0 dispatches vehicles to set out and return to P0 after arriving at 6 distribution points respectively. The initial scheme is shown in Fig. 4. There are 6 distribution lines in total, the total distribution mileage is 120 km, and the number of vehicles used is 6 (10 ton vehicles).
Fig. 4. Initial circuit diagram
614
M. E. Xie et al.
Establish the model, optimize the distribution path through mileage saving method, and formulate the optimal distribution scheme. Step 1: calculate the shortest distance between P0 and six distribution points of the FAP logistics distribution center and between each distribution point, and list the shortest distance table as shown in Table 1. Table 1. Transportation Odometer P P
A
B
C
D
E
F
21
15
17
23
25
19
16
38
44
46
38
26
38
40
34
16
24
36
8
23
A B C D E
15
F
Step 2: Use the mileage saving method to calculate the mileage saving between distribution points. For example, it can be seen from Table 2. LAB = PA + PB − AB = 21 + 15 − 16 = 20
Table 2. Saving Odometer A A B C D E
B
C
D
E
F
20
0
0
0
2
6
0
0
0
24
18
0
40
19 29
F
The mileage saving meter is listed. Step 3: Arrange the data of mileage saving meter in descending order, and list the mileage saving sequence table, as shown in Table 3. Step 4: Use mileage saving method to optimize the distribution route and loading capacity, and draw the final route map, shown in Fig. 5.
An Investigation into Improving the Distribution Routes
615
Table 3. Mileage saved is arranged in descending order No
Mileage
Saving mileage
No
Mileage
Saving mileage
1
D–E
40
5
D–F
19
2
E–F
29
6
C–E
18
3
C–D
24
7
B–C
6
4
A–B
20
8
A–F
2
Fig. 5. Mileage saving and route optimization
The results are as follows: Line 1: traffic volume = 2.1 + 1.2 + 3.2 + 3.5 = 10T; Driving mileage = 19 + 15 + 8 + 16 + 17 = 75 km, Then, 10t vehicle is used for loading and distribution, and the saved distance is: 40 + 29 + 24 = 93 km. Line 2: traffic volume = 5.8 + 3 = 8.8T; Driving mileage = 15 + 16 + 21 = 52 km, Then, 10t truck will be used for loading and distribution, and the saved distance is 20 km.
6 Conclusion The cold chain logistics system in our country is still at a low level and lacks market competitiveness. To achieve a higher position in the competition, we need a complete, efficient and modern cold chain logistics system. Although the total amount of agricultural production and consumption in China has increased rapidly in recent years, the seasonal and regional characteristics of agricultural products have led to large fluctuations in the price of agricultural products, and the characteristics of agricultural products are not easy to preserve, making it put forward higher standards for the timeliness of transportation. On the other hand, in order to achieve quality and safety of the product, customers also expect the quality and freshness of fresh products, which greatly
616
M. E. Xie et al.
increases the circulation cost. From the perspective of the development of FAP cold chain logistics in China, several suggestions are put forward: (1) Strengthen the construction infrastructure, establish a sound refrigerated transportation system, and achieve “seamless” distribution. (2) Accelerate the technological innovation of cold chain logistics, develop new refrigeration equipment, and adopt advanced refrigeration systems to ensure the freshness of food. (3) Increase the support for third-party logistics enterprises, encourage the development and growth of large third-party logistics companies, and provide certain policy support. (4) The government should introduce relevant laws and regulations to standardize the market order and create a good environment for the development of fresh food e-commerce. Acknowledgment. This paper is supported by two projects: (1) The Cooperative Education Project of the Ministry of Education “Intelligent Logistics Planning and Designer Training” (220600924215841), and (2) National Social Science Foundation Program “Research on the Collaborative Development Model of Agricultural Industrial System Based on Agricultural Logistics Park” (18BJY138).
References 1. Cui, Z.: Combining point and area with precise strategy to move towards the new journey of the “Fourteenth Five Year Plan” - Interpretation of the “Fourteenth Five Year Plan” cold chain logistics development plan. China Logist. Procurement 01, 24–26 (2022) 2. Talib, F., Josaiman, S.K., Faisal, M.N.: An integrated AHP and ISO14000, ISO26000 based approach for improving sustainability in supply chains. Int. J. Qual. Reliab. Manag. 38(6), 99–108 (2021) 3. Huang, Y., Cao, H.: Improved artificial bee colony algorithm to optimize support vector machine and its application. Comput. Appl. Softw. 38(02), 258–267 (2021) 4. Gan, W., Zhang, T., Zhu, Y.: On RFID application in the information system of rail logistics center. Int. J. Educ. Manag. Eng. (IJEME), 02(28), 52–58 (2013) 5. Luo, Q., Zhang, L.: Theoretical interpretation and realization path of high-quality development of cold chain logistics of agricultural products. China’s Circulat. Econ. 35(11), 3–11 (2021). (in Chinese) 6. Anitha, P., Malini, M.P.: A review on data analytics for supply chain management: a case study. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 09(08), 30–39 (2018) 7. Ma, S., Zhang, F., Du, C.: Research on the current situation of cold chain logistics development of fresh agricultural products in China. China Storage Transp. 09, 199–200 (2022) 8. Wang, C., Zheng, Y., Huangfu, Y., Hao, H.: Research on location routing optimization of fresh food logistics distribution network. Pract. Underst. Math. 50(10), 33–43 (2020) 9. Tu, W.: Optimization of agricultural product logistics distribution path based on mileage saving method-take S company as an example. China Storage Transp. 06, 89–90 (2022). (in Chinese) 10. Eksoz, C., Mansouri, S.A., Bourlakis, M., et al.: Judgmental adjustments through supply integration for strategic partnerships in food chains. Omega 87(9), 20–33 (2019)
An Investigation into Improving the Distribution Routes
617
11. David, B., Marija, B., Domen, H.: Reprint of mitigating risks of perishable products in the cyber-physical systems based on the extended MRP model. Int. J. Prod. Econ. 10(194), 113– 125 (2017) 12. Zine, B., Ghalem, B., Abdelkader, N.: A cost measurement system of logistics process. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 09(08), 23–29 (2018) 13. Zou, X., Pang, T., Zhou, H.: Research on optimization of fresh agricultural product supply chain under two-way revenue sharing cost sharing contract. Southwest Univ. J. Sci. (Nat. Sci. Ed.) 43(11), 122–130 (2021). (in Chinese) 14. Zheng, Q., Fan, T., Zhang, L.: Income sharing contract of fresh agricultural products under the mode of “agricultural supermarket docking.” J. Syst. Manag. 28(04), 742–751 (2019) 15. Ting, P.-H.: An efficient and guaranteed cold-chain logistics for temperature-sensitive foods: applications of RFID and sensor networks. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 12(18), 1–5 (2013)
Development of Vulnerability Assessment Framework of Port Logistics System Based on DEMATEL Yuntong Qian(B) and Haiyan Wang School of Transportation and Logistics Engineering, Wuhan University of Technology, Wuhan 430070, China [email protected]
Abstract. Port is an important part of the international maritime trade and supply chain. However, in recent years, various emergencies have exposed to the port logistics system, triggered the vulnerability in it. Most of the existing studies involve the vulnerability of ports to specific threats. In order to comprehensively identify and evaluate the bottlenecks exposed by the port logistics system while facing various emergencies, this paper uses the DEMATEL (Decision Making Trial and Evaluation Laboratory) to develop a comprehensive port logistics system vulnerability assessment framework, identify the main vulnerability factors from the qualitative and quantitative perspective. A real case of port in China is investigated. The research results show that the port infrastructure system and port operation management system are the main in ternal factors to trigger the vulnerability of the port logistics system in sequence. Such findings provide useful insight to strengthen the resilience of port logistics system. Keywords: Port logistics · Vulnerability assessment · DEMATEL · Resilience
1 Introduction As an important node of international maritime trade, ports play an important role in ensuring the normal operation of the world economy [1]. Once affected by emergencies, seaports often fail to operate and fall into the dilemma of delay, deviation, interruption and even shutdown [2]. At the same time, due to its special terrain and key socioeconomic functions, seaports are vulnerable to natural and man-made disasters from the sea side and the land side [3]. Any interruption of the seaport will have a direct impact on the supply chain to which the seaport belongs, and will be transmitted to the supply chain network to have an indirect impact on the whole industry [4]. Therefore, with the rapid development of trade globalization, as the “engine” of economic development, port logistics has attracted more and more attention. The research on the vulnerability of port logistics system will also help us find the bottleneck that restricts the efficiency of port logistics, thus promoting the development of the global logistics industry chain. With the occurrence of a series of port accidents caused by natural or human factors, and the impact of various events on the port supply chain, the research on port security © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 618–627, 2023. https://doi.org/10.1007/978-3-031-36115-9_56
Development of Vulnerability Assessment Framework
619
risk and logistics vulnerability has received more and more attention. Mature research methods on vulnerability from ecology and other fields have also been applied in the port logistics field. Qualitative and quantitative research methods based on vulnerability identification have also been proposed by many scholars and have been well applied in the study of specific threats, and have achieved certain research results. However, most of the existing researches on the vulnerability of port logistics are based on specific threats, such as earthquake, fire, tsunami, explosion, epidemic situation, terrorist attacks, etc. the established indicator system is also based on specific threats, lacking a universal and comprehensive indicator system to measure the vulnerability of port logistics [5]. Therefore, the future research direction will focus on the research on the universality and vulnerability of port logistics, which can include the consequences of various threats faced by port logistics and give a comprehensive and objective evaluation. Hence, this paper aims to comprehensively evaluate the vulnerability of port logistics by proposing a universal vulnerability assessment framework. The framework can comprehensively assess the impact of each sub part of the port logistics system on the vulnerability of the entire logistics system, and find out the bottleneck that restricts the efficiency improvement of the port logistics system, highlight the vulnerability of the port logistics system from the side. The vulnerability indicator system that affects port logistics is first constructed, followed by determination of index weight and vulnerability assessment. The other part of this paper is organized as following. Section 2 contains a literature review which focuses on vulnerability and port logistics impact factors, port vulnerability concept, influence of various factors on port logistics vulnerability, and vulnerability identification and research methods. Section 3 introduces the proposed port logistics vulnerability assessment framework. Section 4 presents a case study, results and discussions are also shown in this section. Finally, Sect. 5 concludes the paper with main contributions.
2 Literature Review 2.1 Vulnerability The concept of vulnerability was first proposed in the research on ecology in the 1970s, and then applied to other fields [6]. In recent years, the research on “vulnerability” has become more prominent in the field of information and engineering [7–9]. However, the definition of “vulnerability” has not yet been clearly identified. It is still confused with reliability, risk, accessibility, and so on, even if some researchers have tried to untangle those relationships [10]. In a complex system, vulnerability refers to the collapse of a part (system) caused by internal and external interference factors of the system, while other parts (system) or the whole complex system are directly or indirectly affected, leading to the collapse. When a vulnerability factor or vulnerability source in the system is activated, it will rely on network mobility and transmissibility, leading to other vulnerability sources in a disordered state, and the vulnerability of the system is activated [11]. Therefore, the vulnerability of a complex system is reflected in each part of the system. The vulnerability of the system should be comprehensively evaluated and analyzed, and the dynamic characteristics and driving factors of the vulnerability should be explored.
620
Y. Qian and H. Wang
Although vulnerability is proposed on the basis of risk analysis, they have different emphasis. Risk analysis often focuses on human, environmental and property losses or impacts caused by events [12]. Risk analysis often needs to find out the causes, possibilities and consequences of risk events [13]. When analyzing the vulnerability of the system, we should focus on “the extended set of threats and results”, “reducing the risk of the system and restoring the system to a new stable state”, “the chain breaking time before the system establishes a new stability”. It should be noted that the vulnerability of the system includes the extent to which a system is exposed to external environmental disturbances, the sensitivity of the system, and the self-adaptability to external disturbances. The self-adaptability changes with the changes of external disturbances. 2.2 Port Logistics System Port logistics system refers to the development of a comprehensive port service system with the characteristics of covering all links of the logistics industry chain by a central port city by making use of its own port advantages and relying on advanced software and hardware environment, strengthening its radiation capacity to logistics activities around the port, highlighting the expertise of the port in cargo collection, inventory, and distribution, taking the port-based industry as the basis, information technology as the support, and optimizing the integration of port resources as the goal [14]. Its basic elements include warehousing facilities, terminal equipment and facilities, collection and distribution equipment, handling equipment, transport vehicle equipment, communication networks and facilities, port supporting service facilities and port logistics park, port operation and management personnel, logistics distribution facilities, collection and distribution facilities and information systems. The port realizes the function of modern logistics center with compound advantage, the port multiple identities have strategic status in the international logistics, and the port provides value-added services through the logistics system. To sum up, the port logistics system is a highly integrated and complex system. The port undertakes the conversion task of multiple transportation modes, and the internal function modules of the port are also various. Therefore, the factors that affect the development of port logistics are also diverse. 2.3 Vulnerability of Port Logistics System Based on the application of vulnerability theory in other disciplines, many experts put forward their own views on the vulnerability of ports. Port logistics is not only a relatively complex system, but also closely connected with the external environment, which makes the study of port logistics vulnerability more difficult. Assessing the vulnerability of a port is very challenging, as shown in the following aspects: First, there are multidimensional definitions of port vulnerability [15], and different experts focus on different latitudes; Second, there is no statistics on the critical threshold for the occurrence of largescale disasters in ports [16]; Third, how to construct vulnerability indicators [17]. Therefore, this paper can give the concept of port logistics vulnerability: that is, under the disturbance and interference of internal and external factors, the port logistics system loses all or part of its operational capacity due to the instability and sensitivity
Development of Vulnerability Assessment Framework
621
of its own system, resulting in the decline or stagnation of the efficiency of the logistics system. 2.4 Vulnerability Assessment Methods Vulnerability factors usually have strong concealment and fuzziness, and there are relatively large difficulties and challenges in collecting vulnerability factors. From domestic and foreign research literature, vulnerability identification methods in the transportation field usually refer to risk identification methods. Traditional identification methods include Delphi method, questionnaire survey method, scenario analysis method and literature sorting method. At the beginning, qualitative analysis methods were mainly used for vulnerability identification, but these traditional methods usually have complex processes and heavy tasks, The results often lack systematic ness and comprehensiveness. Later, scholars began to slowly explore the vulnerability of the system measured by quantitative methods. Quantitative vulnerability methods were first developed in the field of ecology. Me. Bar et al. [18] believed that vulnerability is the level of the critical value of disasters. They proposed an index weighted vulnerability evaluation method, gave the absolute vulnerability of a standard reference event, calculated the absolute vulnerability of the research event, and then calculated the relative vulnerability of the research event. This evaluation method is relatively reasonable. Liu et al. [19] further improved the FMEA (Failure Mode and Effects Analysis) method by combining fuzzy logic and Bayesian Network [20]. In order to model the complex environment in supply chain security management and deal with uncertain information, Yang et al. [21] further extended the classic fuzzy rule-based system by incorporating the concept of confidence into the subsequent part of the traditional IF-THEN rule. Shieh et al. [22] integrated three system analysis methods of DEMATEL, ISM (Interpretative Structural Modeling Method) and ANP (Analytic Network Process) to identify the vulnerability factors affecting the transportation system, effectively integrating the characteristics of vulnerability factors and the interaction between vulnerability factors. It can be seen that the combination of quantitative and qualitative methods can effectively overcome the shortcomings between the two, which is also the mainstream method of vulnerability research. Although there have been many research achievements in the vulnerability of ports, most of the existing studies are aimed at the identification and evaluation of port logistics vulnerability under specific threats, lacking a research framework to explore the vulnerability of port logistics system from the overall perspective and the coupling and cooperation between subsystems. In order to fill the research gap, this paper develops a port logistics system vulnerability assessment framework to identify the potential vulnerability sources of the port logistics system, find out the important bottlenecks that affect the improvement of the port logistics efficiency, find out the vulnerable links in the port logistics system and put forward suggestions, so as to reduce the vulnerability of the system and enhance the system resilience.
622
Y. Qian and H. Wang
3 Port Logistics Vulnerability Assessment Model 3.1 Assessment Framework Based on the DEMETAL method, this paper developed a framework for port logistics vulnerability assessment. Firstly, considering that the port logistics system is a complex system, and it is difficult to collect data and distort data by adopting quantitative index method, this paper establishes a vulnerability evaluation system of port logistics system based on interviews with many experts in the port field and their relevant managers (Table 1). Table 1. Vulnerability evaluation index system of port logistics system. Subsystem Bi
Vulnerability influencing factors C i
Natural and geographical conditions B1
The depth of water and channel conditions C1 Port construction conditions C 2 Frequency of natural disasters C 3
Infrastructure conditions B2
Status of port handling facilities C 4 Port storage conditions C 5 Port berth status C 6 Status of traffic facilities in the port C 7 Quality supervision and management level C 8 Average time of ship in port C 9 Personnel management ability and stuff quality C 10
Logistics information system B3
Cargo information management level C 11 Management level of ship mobilization C 12 Customer relationship management level C 13
Port logistics support system B4
Supporting facilities around the port C 14 Port collection and distribution capacity C 15 Development status of port industry C 16
Policy and coordination factors B5
Port administration level C 17 Government supervision and coordination C 18
Secondly, the DEMATEL method is used to construct the overall influence matrix of the vulnerability factors of the port logistics system, which is used to characterize the comprehensive influence relationship among various factors. By using the DEMATEL method, identify the key vulnerability factors in the port logistics system, and identify the vulnerability degree of the port subsystem. Finally, reasonable suggestions are given to reduce the vulnerability of port logistics system.
Development of Vulnerability Assessment Framework
623
3.2 DEMATEL Method DEMATEL is a method of systematic factor analysis using graph theory and matrix tools. The specific steps are as follows: First, determine each factor with Delphi method, brainstorming method or expert interview method. Second, determine the degree of direct influence between elements. First, use the expert scoring method to determine the direct influence matrix of vulnerability influencing factors. The relationship between the factors is divided into five grades. 0 indicates no influence relationship, 1 indicates weak influence, and 2 indicates relatively weak influence k. A score of 3 indicates a strong influence, while a score of 4 indicates a strong influence. A direct influence matrix A is established. The variable aij indicates the elements in the row i and column j of the matrix, and the nj=1 aij expression indicates the sum of rows in the matrix. Third, normalization directly affects the matrix. Using Eq. (1) calculate the normalized direct impact matrix G. G=
max
1 0,λmax > n,Smaller CI means better consistency. In practice, we usually compare the CI with the average random consistency index RI. See Table 5 for RI values: Table 5. RI value table Matrix order
1
2
3
4
5
6
7
8
9
RI
0.00
0.00
0.58
0.90
1.12
1.24
1.32
1.41
1.45
The first-order and second-order matrices are always consistent, and when the judgment matrix order is greater than 2, the random consistency ratio of the judgment matrix needs to be calculated according to Table 5. When RI < 0.10, we consider that the judgment matrix has satisfactory consistency, otherwise, we need to readjust the judgment matrix for recalculation. CR =
CI RI
(4)
3.3.2 Fuzzy Hierarchical Evaluation Steps (1) Establish the index system set. The evaluation index system of emergency organization capability of cruise ship major emergencies is shown in Table 7 in Sect. 4.1. The overall target is recorded as A, and the set of A includes three elements B1 , B2 and B3 , that is, A = {B1 ,B2 ,B3 }. And each secondary target can be further divided into different sets of C 1 , C 2 , C 3 , C 4 , C 5 , etc.
Research on Cruise Emergency Organization Based on Improved AHP-PCE Method
675
(2) Establishing a fuzzy evaluation set. It is difficult to describe the evaluation of the emergency management capability of a cruise ship, and the only way to determine the strengths and weaknesses of the emergency management capability is to use a rubric. The emergency management capability of cruise ships can be divided into four levels: “good”, “better”, “average” and “poor”. “This means that the evaluation set V = {good, better, fair, poor} = (V 1 , V 2 , V 3 , V 4 ). (3) Determination of indicator system weights. Determine the weight value of each indicator using hierarchical analysis. (4) Determine the affiliation. Determination of affiliation. Qualitative analysis of each tertiary indicator is carried out by means of expert consultation and questionnaires, and the evaluation level of each indicator is also given. Based on the results of the survey, a fuzzy relationship matrix T between the evaluation indicators and the evaluation set is established, so as to obtain the fuzzy relationship from A to V. tkj =
The K tertiary index is evaluated as the number of people in the level Total number of surveys ⎤ ⎡ t11 t12 . . t1n ⎢t t . . t ⎥ ⎢ 21 22 2n ⎥ ⎥ ⎢ Tk = ⎢ . . . . . ⎥ ⎥ ⎢ ⎣ . . .. . ⎦ tn1 tn2 . . tnn
(5)
(6)
(5) Fuzzy Comprehensive Evaluation 1) The third level of fuzzy comprehensive evaluation is based on the secondary indicator weights Rij and the tertiary indicator affiliation judgment matrix T k . Aij = Rij Tk
(7)
2) Based on the evaluation results of the first step, the second level indicators are evaluated by fuzzy synthesis. Ai = Ri Aij
(8)
3) Based on the evaluation results of the second step, a fuzzy integrated evaluation of the first level indicators was carried out. A = RAi
(9)
where A is denoted as the cruise ship major emergency response organization comprehensive evaluation results affiliation matrix. The fuzzy comprehensive evaluation results are denoted by K, then: K = AV T
(10)
On the final result of K, different grades of the evaluation object can be obtained.
676
L. Zhang and Z. Li
4 Cruise Ship Major Emergency Response Organization Evaluation 4.1 Calculation of Indicator Weights This chapter is mainly to evaluate the emergency response organization of major cruise ships.The weights of first-level indicators are shown in Tables 6 and 7. Table 6. Judgment matrix and weights of first-level indicators A
Organizational functional complexity
Degree of information conversion
Organizational structure complexity
weight
Organizational functional complexity
1.00
5.00
3.00
0.53
Degree of information conversion
0.20
1.00
0.50
0.15
Organizational structure complexity
0.34
2.00
1.00
0.32
Therefore, the maximum eigenvalue λmax = 5.006, CR = 0.019 < 0.1, CR = 0.025 < 0.1 conformity testing. Table 7. A1 judgment matrix and weight A1
Functional coupling degree
Degree of functional polymerization
Hierarchical functional span
Weights
Functional coupling degree
1.00
2.00
0.34
0.31
Degree of functional polymerization
0.50
1.00
0.250
0.17
Hierarchical functional span
3.00
4.00
1.00
0.52
The weights of the secondary indicators are shown in Tables 8, 9 and 10.
Research on Cruise Emergency Organization Based on Improved AHP-PCE Method
677
Largest eigenvalue is λmax = 5.006, CR = 0.019 < 0.1, conformity testing. Table 8. A2 judgment matrix and weights A2
Information feedback
Communication network
Information channel
Weights
Information feedback
1.00
2.00
0.34
0.32
Communication network
0.50
1.00
0.20
0.17
Information channel 3.00
5.00
1.00
0.52
Largest eigenvalue is λmax = 4.923, CR = 0.017 < 0.1, conformity testing. Table 9. A3 judgment matrix and weights A3
Management level
Management range
Rank relation
Weights
Management level
1.00
5.00
6.00
0.51
Management range
0.20
1.00
3.00
0.31
Rank relation
0.17
0.34
1.00
0.18
Largest eigenvalue is λmax = 4.865, CR = 0.015 < 0.1, conformity testing. Through the calculation of the weight of the first-level indicators, we can determine the comprehensive weight of the second-level indicators. The Table 10 is the comprehensive weight table of the second-level indicators: Table 10. Comprehensive weight table of secondary indicators Organizational functional complexityB1
Degree of information conversionB2
Organizational structure complexityB3
0.532
0.145
0.313
Comprehensive weight
Function coupling 0.310 degree C1
0.103
Functional degree 0.169 of polymerization C2
0.056
(continued)
678
L. Zhang and Z. Li Table 10. (continued)
Hierarchical functions span C3
Organizational functional complexityB1
Degree of information conversionB2
Organizational structure complexityB3
0.532
0.145
0.313
0.521
Comprehensive weight
0.173
Information feedback C4
0.323
0.107
Communication network C5
0.165
0.055
Information channel C6
0.521
0.173
Management level C7
0.511
0.170
Management amplitude C8
0.308
0.102
The grade is C9
0.181
0.061
4.2 Fuzzy Evaluation of Emergency Organization for Cruise Major Emergencies In this section, based on the calculation of the first-level indicators and second-level indicators in Sect. 4.1, combined with the scoring results of the questionnaire feedback, the distribution of each indicator is statistically obtained. According to the formula (5) in Sect. 4, the three-level indicators are normalized to obtain the affiliation matrix AI, and the major emergencies of the cruise ship are analyzed according to the formula (6) ~ formula (10). The emergency organization conducts a fuzzy comprehensive evaluation. The evaluation process is shown in Table 11. Table 11. Fuzzy comprehensive evaluation process of major cruise accidents Secondary index evaluation results First level index evaluation results Evaluation result (0.16,0.16,0.35,0.33)
(0.18,0.22,0.30,0.30)
(0.18,0.25,0.32,0.25)
(0.19,0.17,0.33,0.31)
(0.18,0.25,0.32,0.25)
(0.10,0.24,0.35,0.31) (0.30,0.27,0.23,0.20) (0.16,0.20,0.35,0.29) (0.27,0.20,0.27,0.26) (continued)
Research on Cruise Emergency Organization Based on Improved AHP-PCE Method
679
Table 11. (continued) Secondary index evaluation results First level index evaluation results Evaluation result (0.14,0.13,0.40,0.33) (0.16,0.30,0.29,0.25)
(0.19,0.29,0.30,0.12)
(0.18,0.25,0.32,0.25)
(0.27,0.35,0.20,0.18) (0.14,0.13,0.40,0.33)
Assuming that the comment level V = (V1 , V2 , V3 , V4 ) = {poor, general, better, good}, and taking the values{5, 6, 7, 8} respectively, it can be concluded that the emergency The organizational fuzzy comprehensive evaluation result K = 6.32. It can be seen from this that the emergency organization system for major emergencies of the cruise ship is between average and good, and it can be considered that the emergency organization system is basically perfect.
5 Conclusion Cruise tourism is an indispensable form of travel in modern tourism, and cruise safety issues have become the most concerned aspects of passengers, cruise companies, and cruise management departments. The occurrence of various large and small cruise accidents not only brought huge losses to the lives and property of passengers, but also made people afraid of cruise tourism, a modern way of travel, and dare not try it. Whether the emergency management of major cruise emergencies is perfect and reasonable is an important support for maintaining the rapid development of the cruise economy. This paper mainly focuses on the relevant research on the emergency organization of major emergencies of cruise ships. Through the investigation and analysis of relevant factors affecting the emergency organization of major emergencies of cruise ships, the relevant indicators of emergency organization are put forward, and the evaluation model of emergency organization for major emergencies of cruise ships is constructed at the same time.
References 1. Wang, C., Guo, J., Kuo, M.: The building of social resilience in Sichuan after the Wenchuan earthquake: a perspective of the social-government interactions. Saf. Sci. (C) (2020) 2. Harrison, S., Johnson, P.: Challenges in the adoption of crisis crowd sourcing and social media in Canadian emergency management. Gov. Inf. Q. 8(3), 3–4 (2019) 3. Zaw, T.N., Lim, S.: The military’s role in disaster management and response during the 2015 Myanmar floods: a social network approach. Int. J. Disaster Risk Reduct. 18(10), 6–8 (2017) 4. Espinoza, A.E., Osorio-Parraguez, P., Quiroga, E.P.: Preventing mental health risks in volunteers in disaster contexts: the case of the Villarrica Volcano eruption, Chile. Int. J. Disaster Risk Reduct. 21(13), 12–14 (2018)
680
L. Zhang and Z. Li
5. Whittaker, J., McLennan, B., Handmer, J.: A review of informal volunteerism in emergencies and disasters: definition, opportunities and challenges. Int. J. Disaster Risk Reduct. 15(8), 6–7 (2015) 6. Bundy, J., Pfarrer, M.D., Short, C.E., Coombs, W.T.: Crises and crisis management: integration, interpretation, and research development. J. Manag. 11(6), 31–33 (2017) 7. Bisri, M.B.F., Beniya, S.: Analyzing the national disaster response framework and interorganizational network of the 2015 Nepal/Gorkha earthquake. Procedia Eng. 21(11), 12–14 (2016) 8. Jiang, P., Wang, Y., Liu, C., Hu, Y., Xie, J.: Evaluating critical factors influencing the reliability of emergency logistics systems using multiple-attribute decision making. Symmetry 13(4), 6–7 (2019) 9. Liu, D., Li, Y., Fang, S., Zhang, Y.: Influencing factors for emergency evacuation capability of rural households to flood hazards in western mountainous regions of Henan province, China. Int. J. Disaster Risk Reduct. 8(3), 16–17 (2017) 10. Eisenman, D.P., Long, A., Setodji, C., Hickey, S., Gelberg, L.: Differences in individual-level terrorism preparedness in Los Angeles County. Am. J. Prev. Med. 10(2), 6–7 (2012) 11. Hearst, M.A., Scholkopf, B., et al.: Trends and controversies-support vector machines. IEEE Intell. Syst. 16(5), 24–34 (2005) 12. Paciarotti, C., Cesaroni, A., Bevilacqua, M.: The management of spontaneous volunteers: a successful model from a flood emergency in Italy. Int. J. Disaster Risk Reduct. 11(3), 12–14 (2018) 13. Yue, H., Zhu, T.L.: Improvement of evaluation method of elderly family medical product design based on AHP. Math. Probl. Eng. 11(8), 6–7 (2022) 14. Dai, X., Wu, X.: Safety and stability evaluation of the uranium tailings impoundment dam: based on the improved AHP-cloud model. J. Radiat. Res. Appl. Sci. 15(1), 21–31 (2022) 15. Niloufar, V., Mehdi, G.: Preference of hybrid steel frame with exclusive seismic performance using the analytic hierarchy process. J. Earthquake Eng. 26(10), 12–14 (2022) 16. Jiang, L., Li, Y., Jiang, C.: Employment competitiveness of college students based on improved AHP. Eurasia J. Math. Sci. Technol. 13(8), 3–4 (2017) 17. Zhang, J., Xiao, B.: Application of analytic hierarchy process in abnormal vibration of Marine machinery. J. Phys: Conf. Ser. 17(6), 11–12 (2021) 18. Han, B., Ming, Z.: Comprehensive risk assessment of transmission lines affected by multimeteorological disasters based on fuzzy analytic hierarchy process. Int. J. Electric. Power Energy Syst. 13(3), 6–7 (2021) 19. Dhanapal, S., Panneer, D.D., Sarit, M.: Composite techniques of structural equation modeling and analytic hierarchy process for information technology vendor selection. Int. J. Inf. Technol. Decis. 20(4), 11–15 (2021) 20. Nie, H.: Fuzzy evaluation model of the teaching quality in colleges and universities based on analytic hierarchy process. Basic Clin. Pharmacol. Toxicol. 127(3), 189–190 (2020)
Fresh Agricultural Products Supplier Evaluation and Selection for Community Group Purchases Based on AHP and Entropy Weight VIKOR Model Gong Feng1 , Jingjing Cao1(B) , Qian Liu1 , and Radouani Yassine1,2 1 School of Transportation and Logistics Engineering, Wuhan University of Technology,
Wuhan 430072, China [email protected] 2 International University of Rabat, 999055 Rabat, Morocco
Abstract. The outbreak and persistence of COVID-19 and the lack of effective management of the agricultural product supply chain have made the green and safe supply of fresh agricultural products an increasingly prominent issue. As a derivative of the new retail model, the community group purchases platforms have become the main spots for Chinese residents to consume fresh agricultural products. Therefore, the selection of safe and reliable suppliers of fresh agricultural products has become an inescapable focal issue in the development of community group purchases platforms. This paper proposes a three-stage evaluation index system for sustainable supplier selection of fresh agricultural products from the perspective of safety and sustainability, taking into account several aspects of suppliers such as product quality, service, development, technology and green. Then combine AHP with the entropy-weighted VIKOR method to calculate the combined weight of each indicator and prioritize candidate suppliers. Meanwhile, an numerical example is presented to verify the effectiveness of the proposed evaluation model. This paper aims to provide a reference for relevant community group purchases platforms and suppliers to form their own core competitiveness for better development. Keywords: Fresh agricultural products · Supplier evaluation and selection · Community group purchases platforms · AHP · Entropy Weight · VIKOR
1 Introduction Recently, with the development of the national economy and the increase in consumer income, ecological pollution has intensified [1]. And COVID-19 is still active. Consumer’s expectations for fresh agricultural product have shifted these days [1]. During COVID-19, developing community group purchasing platforms played a unique role in safeguarding people’s livelihoods, as traffic control and community lockdown pose strong impact on the public’s normal life. Consumer consumption habits and patterns of © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 681–695, 2023. https://doi.org/10.1007/978-3-031-36115-9_61
682
G. Feng et al.
reliance persist in today’s post-epidemic environment, providing a chance for community group purchasing platforms that focus on fresh agricultural product to grow fast in China. The selection of fresh food suppliers is essential for a community group purchases platform’s procurement system, which affects the quality of the products and the company’s procurement costs [3]. And directly determines a community group purchases platform’s market competitiveness [4]. Research on suppliers of fresh agricultural products in China started late and hasn’t formed a mature theoretical and methodological system yet [5]. Now, homogenization of major community group purchases platform is serious. This has led to problems such as the poor competitiveness of community group purchases platforms and the low quality of fresh agricultural products provided [6, 7]. Therefore, there is an urgent demand to establish an agricultural products supplier selection and evaluation system to reduce the procurement costs of community group purchases platforms effectively and guarantee the quality of fresh agricultural products. This paper proposes the AHP and Entropy-weight VIKOR method to construct a new evaluation system for the selection of suppliers for community group purchases platforms, the contributions and innovations of this paper are as follows: 1) We established a supplier evaluation index system in five aspects: quality criteria, service criteria, technical criteria, development criteria and green criteria, and construct an indicator system including 15 indicators. 2) We proposed an integrated hybrid approach of AHP and Entropy weights VIKOR to select suppliers for community group purchases platforms. The weights of the indicators consider both subjective and objective factors; the VIKOR judging method reduces the tendency of negative indicators’ being overlooked, maximizes group benefits and minimizes individual regret. 3) We concentrated on community group purchases platforms, which has recently emerged and is a derivative of the new retail model. This paper can fill the gap in the research on the selection of suppliers for fresh products in community group purchases platforms to a certain extent and has certain significance for relevant enterprises. The paper is organized as follows. In Sect. 2 the indicators related to the selection of community group purchases platforms suppliers are described, and the evaluation system is determined. The data processing and calculation process of the AHP-Entropy VIKOR method is described in Sect. 2.4, following by a numerical application and case study of a community group purchases platform in Sect. 3. Recommendations on the management issues involved are provided in Sect. 4. In Sect. 5, the relevant research on green suppliers in China and abroad is reviewed. Finally, in Sect. 6 concluding remarks and future research directions are presented. 1.1 Evaluation System of Green Suppliers of Fresh Agricultural Products Considering the background of the post-epidemic era, the relevant characteristics of fresh agricultural products supply chain and the traditional evaluation factors of suppliers, the evaluation indicators of green suppliers of fresh agricultural products for community group purchase platform are determined, and an evaluation system composed of target
Fresh Agricultural Products Supplier Evaluation and Selection
683
layer, constraint major element layer, constraint sub-element layer and program layer is constructed from top to bottom, as shown in Fig. 1. In order to display the evaluation indicators for the selection of community group purchases platform suppliers better, this paper starts from five aspects: quality criteria, service criteria, technology criteria, development criteria and green criteria, and further decomposes them into 15 sub-elements, aiming to make a comprehensive, specific and objective evaluation of fresh agricultural products suppliers of community group purchases platforms.
Fig. 1. Community group purchases fresh agricultural products supplier evaluation system
2 Methodology In this work, a novel three-stage approach is proposed for the evaluation and selection of suppliers for Community Group Purchases platforms. In the first stage, the subjective weights of each criterion and indicator are determined based on the scoring of relevant criteria and indicators by professional evaluators using the AHP method. In the second stage, the objective weights of each indicator are determined by entropy weighting method based on the supplier’s raw data. Then the subjective and objective weights are combined to derive a composite weight. Lastly, the VIKOR method was used to prioritize the community group Purchases platform suppliers. 2.1 Hierarchical Analysis to Determine Subjective Weights Hierarchical analysis (AHP) provides a simple decision-making method for complex decision-making problems with multiple objectives, multiple criteria or unstructured
684
G. Feng et al.
characteristics by mathematizing the decision-making process with less quantitative information based on an in-depth analysis of the nature of complex decision-making problems, influencing factors and their intrinsic relationships. The basic steps are as follows. 1) Construction of the judgment matrix. The importance levels between the two indicators are shown in the Table 1. ⎡ ⎤ a11 · · · a1m ⎢ ⎥ (1) A = (aij )n×m= ⎣ ... . . . ... ⎦ an1 · · · anm
Table 1. Importance level and scale value Scale
Meaning
1
Both are equally important
3
The former is slightly strong
5
The former is strong
7
The former is obviously strong
9
The former is absolutely strong
2, 4, 6, 8
The effect of the former is between above two adjacent levels
1, 1/2, …, 1/9
The ratio of influence is exactly the reciprocal of aij
2) Solve the judgment matrix A to obtain the maximum eigenvalue. λmax =
1 n (Aw)i , i = 1, 2, ...n, w = (w0 , w1 , w2 , w3 , w4 ) i=1 wj n
(2)
3) Calculate consistency indicator CI, random consistency indicator RI and consistency ratio CR for consistency testing. CI =
CI λmax − n , CR = . n−1 RI
(3)
When CR = 0, there is perfect consistency; when CR < 0.1 consider A to be a satisfactory consistency matrix. CI、CR is given in the (3), RI is obtained by looking up the Table 2 according to the order of the matrix; when CR < 0.1, the eigenvector sought is the weight value of each criterion layer factor, otherwise reconstruct the judgment matrix A. 2.2 Entropy Weighting Method to Determine Objective Weights The entropy weight method is an objectively determined weight method. In information theory, the amount of information provided by data can be reflected by entropy. When
Fresh Agricultural Products Supplier Evaluation and Selection
685
Table 2. Average random consistency index n
3
4
5
6
7
8
9
RI
0.52
0.89
1.12
1.26
1.36
1.41
1.46
the data of each plan of an indicator differs greatly, the entropy is low and the amount of information provided is large, while the opposite is true when the data difference is small. The basic steps are as follows. 1) According to the raw data of each indicator, an evaluation matrix of type can be constructed based on the m standardized indicator values of the given n evaluation objects, and the matrix is as follows. 2) Normalize each element in the evaluation matrix. The evaluation indexes are divided into two categories, one is the benefit index and the other is the cost index. The larger the benefit index value, the better, and the smaller the cost index value, the better. ⎤ ⎡ x11 · · · x1n ⎥ ⎢ (4) X = ⎣ ... . . . ... ⎦
yij =
⎧ ⎪ ⎨ ⎪ ⎩
xm1 · · · xmn
xij −min(xi ) max(xi )−min(xi ) ,
benefit − type indicators
max(xi )−xij max(xi )−min(xi ) ,
(5)
cost − type indicators
max(xi ), min(xi ) are the maximum and minimum values of the ith indicator respectively. The standardized raw data can be formed into a new evaluation matrix of type Y = (yij )m×n and the elements of the matrix are the standardized data. 1) Calculate the entropy value for indicator.
n yij j=1 Pij lnPij , Pij = n Ei = − ln(n) j=1 yij
(6)
2) Calculate the weights from the calculated entropy value. wi =
1 − Ei
m− m i=1 Ei
(7)
2.3 Calculate the Composite Weights Different weighting methods may result in different final optimal solution. The subjective weighting method can reflect the decision maker’s intention, but the evaluation results are more subjective; while the objective weighting method is more theoretical, but doesn’t consider the decision maker’s intention. Thus, we use a combination of subjective and objective weighting methods to determine the weights.
686
G. Feng et al.
Assuming that the indicator weight determined by the hierarchical analysis method is wi and the entropy weighting method determines a weight of i , the composite weight is λi . A common formula for a combined subjective-objective approach to calculate composite weight is: wi ×i i = 1, 2, 3, ..., m λi = m i=1 wi ×i
(8)
Song et al. improved the formulae commonly used for the combined subjectiveobjective assignment method [8]. And one of them can be descripted as follows: wi +i λi = m i = 1, 2, 3, ..., m i=1 wi +i m m m (wi +i ) = wi + i = 2 i=1
i=1
i=1
(9) (10)
Calculate Eq. 10 to get Eq. 11 λi = 0.5 × wi + 0.5 × i
(11)
2.4 VIKOR Method Judging Model VIKOR is a method for optimizing compromise solutions in multi-attribute decision making. The basic idea of VIKOR is to define the ideal solution and the negative ideal solution first and then prioritize the alternatives according to how close they are to the ideal solution in terms of their evaluated values. The basic steps are as follows. 1) Calculate the positive and negative ideal solutions yi+ ,yi− + − yi , yi =
+ + maxj yi+, minj yi + , ∀i, Benefit type minj yi , maxj yi , ∀i, Cost Type
2) Calculate utility value and individual regret value Ri . n λj (yj+ − yij ) λj (yj+ − yij ) Si = , Ri = max j=1 y + − y − j yj+ − yj− j j
(12)
(13)
where yj+ denotes positive ideal value, yj+ = max yij
(14)
yj− denotes negative ideal value, yj− = min yij
(15)
j
j
3) Calculate the compromise value Qi Qi = ε
Si − S − Ri − R− + (1 − ε) + + − S −S R − R−
(16)
Fresh Agricultural Products Supplier Evaluation and Selection
687
where ε is compounding factor, and S + = max {Si }, S − = min{Si }
(17)
R+ = max {Ri }, R− = min{Ri }
(18)
i
i
i
i
Ranking of alternative suppliers. Firstly, the ranking is obtained by increasing the value of Qi : A(1) , A(2) , A(2) , A(3) , . . . , A(i) . If A(1) is the scheme ranked first according to the value of Qi and satisfies the following conditions. Q(A(2) ) − Q(A(1) ) ≥
1 . n−1
Scheme A(1) is still optimal according to Si Ri ordering. Then A(1) is the most stable optimal solution in the decision process. If the two conditions above do not hold simultaneously, there are two situations to obtain a compromise solution. If condition 1 is not satisfied, the compromise is A(1) , . . . , A(i) , A(i) which is the 1 . maximised i-value determined by Q(A(1) ) < n−1 If condition 2 is not satisfied, the compromise solution is A(1) , A(2) .
3 Empirical Research A community group Purchases platform has three fresh agricultural products suppliers which are evaluated to determine the best supplier. Based on the actual situation of the three suppliers we have compiled data on the performance of the three suppliers in various indicators, which is presented in Table 3. Table 3. Supplier raw data Item
Supplier 1
Supplier 2
Supplier 3
Indicator Type
Criterion
Indicators
Quality
C11
80
95
85
Benefit type
C12
80
95
85
Benefit type
C13
80
90
85
Benefit type (continued)
3.1 Determination of Indicator Weights After consulting a large number of relevant documents and consulting relevant experts and scholars, according to the basic principles of AHP, we constructed the judgment
688
G. Feng et al. Table 3. (continued)
Item
Supplier 1
Supplier 2
Supplier 3
Indicator Type
Criterion
Indicators
Services
C21
85
80
85
Benefit type
C22
90
75
70
Benefit type
Technology
Development
Green
C23
85
80
90
Benefit type
C31
75
95
80
Benefit type
C32
80
90
85
Benefit type
C33
85
95
75
Benefit type
C41
90
85
95
Benefit type
C42
70
95
85
Benefit type
C43
85
90
95
Benefit type
C51
85
80
90
Benefit type
C52
85
75
70
Cost type
C53
80
90
85
Benefit type
matrix O-B in Table 4. For the target layer corresponding to the criterion layer and the judgment matrix B1-C, B2-C, B3-C, B4-C and B5-C for the criterion layer corresponding to the indicator layer (Table 5). Then we used Excel to calculate the weights and conduct consistency checks. Finally, we calculate the subjective weight of the target layer corresponding to the index layer in In the same way, the judgement matrix B2-C, B3-C, B4-C, B5-C is constructed and the related metrics can be calculated. Then the hierarchical indicators is sorted and ranked as Table 6. Table 4. Judgement matrix O-B Judgement Matrix O-B and Consistency Test B2
B3
B4
Wi
λmax
5
0.5127
5.0040
2
0.1907
Criterion
B1
B5
B1
1
3
5
5
B2
1/3
1
2
2
B3
1/5
1/2
1
1
1
0.0989
B4
1/5
1/2
1
1
1
0.9889
B5
1/5
1/2
1
1
1
0.0989
CI = 0.0010 CR = 0.009 RI = 1.12
In the same way, the judgement matrix B2-C, B3-C, B4-C, B5-C is constructed and the related metrics can be calculated. Then the hierarchical indicators is sorted and ranked as Table 6.
Fresh Agricultural Products Supplier Evaluation and Selection
689
Table 5. Judgement matrix B1-C Judgement Matrix B1 -C and Consistency Test Quality criterion
C11
C12
C13
Wi
λmax
C11
1
1/3
4
0.2499
3.0092
C12
3
1
9
0.6813
C13
1/4
1/9
1
0.0688
CI = 0.0046 CR = 0.0088 RI = 0.52
Table 6. Composite ranking of hierarchical indicators Indicator
Quality
Services
Technology
Development
Green
0.5127
0.1907
0.0989
0.0989
0.0989
W
C11
0.2499
–
–
–
–
0.1282
C12
0.6813
–
–
–
–
0.3493
C13
0.0688
–
–
–
–
0.0353
C21
–
0.2297
–
–
–
0.0438
C22
–
0.6483
–
–
–
0.1236
C23
–
0.1220
-
–
–
0.0233
C31
–
–
0.6370
–
–
0.0630
C32
–
–
0.1047
–
–
0.0104
C33
–
–
0.2583
–
–
0.0255
C41
-
–
–
0.6483
–
0.0641
C42
-
–
–
0.1220
–
0.0121
C43
-
–
–
0.2297
-
0.0227
C51
-
–
–
–
0.1095
0.0108
C52
-
–
–
–
0.3090
0.0306
C53
-
–
–
–
0.5816
0.0575
The entropy and entropy weights of each evaluation indicator are calculated according to (4) to (7), where environmental protection expenditure is a cost-based indicator and the rest of the indicators are benefit-based indicators. The entropy weights for each indicator were calculated as shown in Table 7. Based on the weights of each indicator derived from the hierarchical analysis and the entropy weighting method, the combined weights were calculated according to (11) in Tables 7 and 8.
690
G. Feng et al. Table 7. Entropy weighting calculation results
Criterion
Indicator
Entropy value ei
Coefficient of variation ∂i
Entropy weights wi
Quality
C11
0.5119
0.4881
0.0741
C12
0.5119
0.4881
0.0741
Services
Technology
Development
Green
C13
0.5794
0.4206
0.0639
C21
0.6309
0.3691
0.0561
C22
0.4555
0.5445
0.0827
C23
0.5794
0.4206
0.0639
C31
0.4555
0.5445
0.0827
C32
0.5794
0.4206
0.0639
C33
0.5794
0.4206
0.0639
C41
0.5794
0.4206
0.0639
C42
0.6022
0.3978
0.0604
C43
0.5794
0.4206
0.0639
C51
0.5794
0.4206
0.0639
C52
0.6126
0.3874
0.0588
C53
0.5794
0.4206
0.0639
Table 8. Composite weight calculation results Serial number Indicator Subjective weights θ i Entropy weights wi Composite weights λi 1
C11
0.1282
0.0741
0.1011
2
C12
0.3493
0.0741
0.2117
3
C13
0.0353
0.0639
0.0496
4
C21
0.0438
0.0561
0.0499
5
C22
0.1236
0.0827
0.1032
6
C23
0.0233
0.0639
0.0436
7
C31
0.0630
0.0827
0.0728
8
C32
0.0104
0.0639
0.0371 (continued)
3.2 VIKOR Method to Determine the Optimal Option The positive and negative ideal solutions for each indicator can be derived from Eqs. (4), (5) and (12). yi+ = (1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1); yi− = (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0)
Fresh Agricultural Products Supplier Evaluation and Selection
691
Table 8. (continued) Serial number Indicator Subjective weights θ i Entropy weights wi Composite weights λi 9
C33
0.0255
0.0639
0.0447
10
C41
0.0641
0.0639
0.0640
11
C42
0.0121
0.0604
0.0362
12
C43
0.0227
0.0639
0.0433
13
C51
0.0108
0.0639
0.0374
14
C52
0.0306
0.0588
0.0447
λ = (0.1011, 0.2117, 0.0496, 0.0499, 0.1032, 0.0436, 0.0728, 0.0371, 0.0447, 0.0640, 0.0362, 0.0433, 0.0374, 0.0447, 0.0607)
The S, R and Q values for each of the 3 suppliers were calculated and ranked according to Eqs. (13), and the results are shown in Table 9. Table 9. S, R, Q for the three suppliers Suppliers
S
R
Q
Value
Ranking
Value
Ranking
Value
Ranking
Supplier 1
0.7074
3
0.2117
3
1
3
Supplier 2
0.3236
1
0.0774
1
0
1
Supplier 3
0.5439
2
0.1411
2
0.5244
2
Firstly, the ranking is obtained by decreasing value: Supplier 2, Supplier 3, Supplier 1. This is the option ranked first according to the value of Q and satisfies the following conditions. 1) Q(A(2) ) − Q(A(1) ) = 0.5244 ≥ 0.5 2) Option A(1) is still the best option according to the S and R ranking. So the supplier selection ranking is: Supplier 2 > Supplier 3 > Supplier 1.
4 Analysis and Suggestions Through hierarchical analysis, an evaluation system consisting of a target layer, a constraint major element layer, a constraint sub-element layer and a program layer was constructed. By the case studied in Sect. 4, we found that in the supplier selection process of the community group purchases platforms, indicators such as product freshness, epidemic prevention and safety, emergency response capability and cold chain logistics and distribution technology have a greater composite weight. This also reflects consumer preference for quality and safety of fresh agricultural products in the post-epidemic era. By theoretical analysis, it is clear that whether fresh
692
G. Feng et al.
agricultural products carry or spread COVID-19 during processing and transportation, and whether the freshness of the products can be guaranteed are crucial. It affects the competitiveness of the community group purchases platforms directly. If community group purchases platforms want to obtain further development and better foothold in the market, they should take measures from the following aspects. 1) Combine subjective judgement and objective data to make supplier selections. The issue of supplier selection contains both qualitative and quantitative elements. Community group purchases platforms should be both subjective and objective to improve the accuracy of comprehensive evaluation when making decisions on supplier selection. The research results in this paper show that the epidemic prevention and safety, freshness of the products, the emergency response capability and financial stability of the suppliers are given greater weight. This reflects that in the special era, consumers’ preference for the safety and quality of fresh agricultural products, and enterprises should focus on these influencing factors when determining supplier evaluation indicators. 2) Construct a scientific and perfect supplier evaluation system. Determining reasonable supplier evaluation indicators and building a scientific supplier evaluation system can largely reduce enterprise procurement, logistics costs and the potential risks in business management, while ensuring the quality of products and increasing the profit point of enterprises. When determining evaluation indicators, enterprises can consider traditional supplier evaluation and selection indicators such as development and service, but also need to consider the needs of the enterprise’s own development and changes in the social environment.
5 Related Work Research on supplier selection focuses on the determination of supplier evaluation indicators and the establishment of the system and the selection of evaluation methods. In this section we will first review relevant research on fresh products supply chains, and then we will review the current status of research on evaluation indicators and systems and evaluation methods at home and abroad. Previous research on supplier evaluation and selection systems has mainly focused on social performance indicators, economic performance indicators and environmental performance indicators. In recent years, under the influence of epidemics and other related factors, the relevant indicators for supplier evaluation should be adjusted and improved. At the same time, with more and more indicators included in the supplier evaluation index system, the supplier evaluation methods are becoming more and more complex. 5.1 Determination of Indicator Weights Companies must use green supplier selection strategies to respond to market pressures [4]. Mousak hani et al. consider the selection of green suppliers based on environmental competence to be one of the most important issues facing companies [9]. Parkouhi et al. argue that green supplier selection is one of the core issues in green supply chain management and contains a large number of qualitative and quantitative factors as an
Fresh Agricultural Products Supplier Evaluation and Selection
693
MCDM issue [10]. At present, China has not yet developed a mature theoretical and methodological system for the establishment of an evaluation system for the selection of fresh produce suppliers [11]. Community group purchases and other e-commerce platforms have formed a new supply chain operation model for fresh products e-commerce, and how suppliers are evaluated and selected has become extremely important [3]. As an innovative new retail model, community group purchases are booming. However, as a new e-commerce model, the community group purchases platforms have problems such as serious homogeneity and poor product quality [6]. In terms of bringing convenience to consumers, the social e-commerce model has great advantages and potential, but there are still major problems in terms of product quality and product traceability [8]. 5.2 Evaluation Criteria Lo et al. identified supplier evaluation indicators as green performance, product quality, etc., based on the attributes of products purchased by consumers [4]. Mousak hani et al. divided the evaluation criteria of cost, quality, delivery, technology, environmental competence, organization and green image into green and general criteria for the evaluation of green suppliers [9]. Zhang et al. propose environmental performance, economic performance and social performance as integrated evaluation criteria for sustainable suppliers [12]. Rong et al. proposed evaluation indicators based on the impact of the epidemic such as epidemic safety and quality traceability techniques [2]. Incorporating environmental factors into supplier selection leads to higher margins and greater customer satisfaction [13]. Freshness of the product is an important factor in the quality of fresh agricultural products [14]. 5.3 Evaluation Methods Hou et al. used the entropy weighting method to obtain the weights of each evaluation index, and then used the improved TOPSIS method to select the appropriate green supplier [15]. Chandra et al. proposed a comprehensive evaluation method for integrated green suppliers with BWM and VIKOR [16]. Yuan et al. proposed an integrated evaluation method combining SWOT strategic analysis and entropy-weighted fuzzy synthesis evaluation [17]. Wang et al. used hierarchical analysis to select a low-carbon supplier of fresh produce [11]. Agarwal et al. adopted AHP method for supplier selection in dynamic environment [18]. Zhou et al. proposed an evaluation research method combining intuitionistic fuzzy sets and TOPSIS method [19]. As supplier selection is a Multi-Criteria Decision Making problem, Sahai et al. usaged Data Envelopment Analysis to measure supplier performance [20]. Liao et al. used the AHP and TOPSIS composite method to determine the weight of each index for the evaluation and selection of green suppliers for paper manufacturing enterprises [21]. Singh et al. proposed a new hybrid approach by using logarithmic fuzzy preference programming (LFPP) and artificial neural network (ANN) to generate requirement prioritization of the washing machine supplier [22].
694
G. Feng et al.
6 Conclusion and Future Work This paper utilized the AHP-Entropy VIKOR method to evaluate suppliers quantitatively by introducing five criteria: quality, service, technology, development and green, and refining them into 15 secondary indicators. When selecting suppliers, enterprises can refer to the evaluation system established in this paper and calculate the performance of suppliers by combining the composite weights of the calculated secondary indicators; the calculated weights reflect that fresh products suppliers should mainly focus on freshness, epidemic prevention, safety and other product quality indicators. They should also focus on their own financial stability and information technology level. The improvement of these aspects will allow them to have their own unique advantages and increase their competitiveness. Meanwhile, the community group purchases platform that chooses such suppliers has stronger market competitiveness. The three suppliers analyzed in this paper each have their own strengths in product quality, technology or low carbon green, and the relevant indicators meet certain standards, but none of the suppliers has consider all the five aspects of quality, service, technology, development and green, which reflects the current situation of most suppliers to a certain extent. The final selected supplier 2 has a higher priority due to the high quality of the fresh products provided and its technical capabilities, but there is still room for improvement in service guidelines. In future research, we will investigate the development of fresh agricultural product suppliers on community group purchases platforms in greater depth, so as to identify more accurate and comprehensive evaluation indicators and establish a more detailed and complete evaluation system. In terms of evaluation methods, we will seek more scientific and reasonable evaluation models. For example, we will consider the complexity of objective things and the fuzziness of human thinking, and conduct a multi-attribute decision-making process in a fuzzy environment. Acknowledgment. This project is supported by the National Natural Science Foundation, China (No. 61502360).
References 1. Du, Y., Zhang, D., Zou, Y.: Sustainable supplier evaluation and selection of fresh agricultural products based on IFAHP-TODIM model. Math. Probl. Eng. 1–15 (2020) 2. Rong, L., Wang, L., Liu, P.: Supermarket fresh food suppliers evaluation and selection with multigranularity unbalanced hesitant fuzzy linguistic information based on prospect theory and evidential theory. Int. J. Intell. Syst. 37(3), 1931–1971 (2021) 3. Dan, H., Jing, M.: The research on the factors of purchase intention for fresh agricultural products in an E-Commerce environment. IOP Conf. Ser. Earth Environ. Sci. 100, 0123173 (2017) 4. Lo, H.-W., Liou, J.J.H., Wang, H.-S., et al.: An integrated model for solving problems in green supplier selection and order allocation. J. Clean. Prod. 190, 339–352 (2018) 5. Deng, Y.: Selection of cold chain logistics suppliers for fresh agricultural products based on AHP. Logist. Technol. 04, 91–93 (2017). (in Chinese) 6. Li, M., Fang, L.: Business model analysis and development prospect of community Groupbuying platform. Econ. Res. Guid. 03, 83–86 (2022). (in Chinese)
Fresh Agricultural Products Supplier Evaluation and Selection
695
7. Cai, G.: Research on the current situation of agricultural product supply chain considering fresh-keeping efforts in the context of social e-commerce. China Logist. Procurement 12, 95–97 (2022). (in Chinese) 8. Song, H., Wang, Z.: Weighing between objective weight and subjective weight. Tech. Econ. Manag. Res. 03, 62 (2003). (in Chinese) 9. Mousakhani, S., Nazari-Shirkouhi, S., Bozorgi-Amiri, A.: A novel interval type-2 fuzzy evaluation model based group decision analysis for green supplier selection problems: A case study of battery industry. J. Clean. Product. 168, 205–218 (2017) 10. Parkouhi, S.V., Ghadikolaei, A.S.: A resilience approach for supplier selection: using fuzzy analytic network process and grey VIKOR techniques. J. Clean. Product. 161, 431–451 (2017) 11. Wang, Y., Zhang, P., Chen, X.: Empirical analysis of low-carbon fresh agricultural product supplier evaluation. Bus. Econ. Res. 05, 127–130 (2018). (in Chinese) 12. Jing, Z., Dong, Y., Qiang, L., et al.: Research on sustainable supplier selection based on the rough DEMATEL and FVIKOR methods. Sustainability 13(1), 88 (2020) 13. Verma, M., Prem, P.R., Ren, P., et al.: Green supplier selection with a multiple criteria decisionmaking method based on thermodynamic features. Environ. Dev. Sustain. 1–33 (2022) 14. Ma, X., Wang, S., Islam, S.M.N., et al.: Coordinating a three-echelon fresh agricultural products supply chain considering freshness-keeping effort with asymmetric information. Appl. Math. Model. 67, 337–356 (2019) 15. Bin, H., Wang, Y.: Supplier evaluation and selection based on improved TOPSIS method in green supply chain. J. Hunan Univ. Technol. 28(02), 81–86 (2014) 16. Garg, C.P., Sharma, A.: Sustainable outsourcing partner selection and evaluation using an integrated BWM–VIKOR framework. Environ. Dev. Sustain. 22(2), 1529–1557 (2020) 17. Ying, Y., Yi, Z., Long, D., et al.: Research on large supermarket fresh food supplier evaluation and selection based on SWOT-entropy weight fuzzy comprehensive model. In: 2015 International Conference on Advanced Mechatronic Systems, pp. 15–19. IEEE (2015) 18. Prince, A., Manjari, S., Vaibhav, M., et al.: Supplier selection in dynamic environment using analytic hierarchy process. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 6(4), 20–26 (2014) 19. Zhou, Q., Wang, Q., Chen, L.: Research on green supplier selection based on intuitive fuzzy set-TOPSIS. J. Syst. Sci. 01, 94–98 (2017). (in Chinese) 20. Manjari, S., Prince, A., Vaibhav, M., et al.: Supplier selection through application of DEA. Int. J. Eng. Manuf. (IJEM) 4(1), 1–9 (2014) 21. Liao, J., Zhang, R.: Research on the evaluation and selection of green suppliers in papermaking enterprises based on AHP-TOPSIS method. Logist. Eng. Manag. 02, 91–93+84 (2020). (in Chinese) 22. Yash, V.S., Bijendra, K., Satish, C.: A hybrid approach for requirements prioritization using LFPP and ANN. Int. J. Intell. Syst. Appl. (IJISA) 11(1), 13–23 (2019)
Logistics of Fresh Cold Chain Analysis of Joint Distribution Paths in Wuhan Weihui Du1 , Donglin Rong2(B) , Saipeng Xing3 , and Jiawei Sun4 1 School of Logistics, Wuhan Technology and Business University, Wuhan 430065, China 2 National Engineering Research Center of Geographic Information System, China University
of Geosciences, Wuhan 430074, China [email protected] 3 School of Management, Wuhan Technology and Business University, Wuhan 430065, China 4 Technical University of Munich Asia Campus, Technical University of Munich (TUM), Singapore, Singapore
Abstract. In the process of transportation and circulation of fresh products, the loss rate is an important reason for the high cost of logistics and the decline of product quality. Cold chain joint distribution of fresh products is an important method and means to reduce the loss rate. It is of great significance to study cold chain joint distribution for fresh cold chain logistics. The research content of this article is a common distribution route optimization fresh cold chain logistics in Wuhan city. First of all, the general situation of fresh cold chain logistics in our country is summarized, and the related concepts and connotations of cold chain logistics are deeply interpreted. Secondly, the existing problems of fresh cold chain logistics in Wuhan are analyzed and discussed. Finally, taking Wuhan Baishazhou cold chain logistics distribution center as an example, this paper uses the mileage saving method to study the route optimization of cold chain logistics distribution, so as to provide solutions to the high loss and high cost problems of fresh cold chain logistics joint distribution. Keywords: Cold chain logistics · Joint distribution · Path optimization · Mileage saving method
1 Introduction Compared with other countries such as the United States and Japan, China’s fresh food cold chain logistics started a little later, so it is relatively backward in terms of cold chain transportation equipment and transportation skills, and there are more problems in the whole cold chain logistics transportation. At this stage, most of the food in China is still in accordance with the traditional distribution method of room temperature logistics to carry out the transportation of fresh food, the high cost, difficulty and strict time requirements of distribution have become important problems faced in the process of fresh cold chain logistics transportation. In cold chain logistics, choosing the optimal distribution route can save a lot of resources, reduce costs and speed up the circulation of fresh food, thus increasing the company’s revenue [1]. For the company, optimizing © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 696–707, 2023. https://doi.org/10.1007/978-3-031-36115-9_62
Logistics of Fresh Cold Chain Analysis of Joint Distribution Paths
697
the common distribution route in fresh food cold chain logistics greatly reduces the transport time of fresh food, reduces the loss of fresh food, reduces the costs consumed in transit, and improves the sales efficiency and customer satisfaction of new products. For the customer, the optimization of the common distribution route of the fresh food cold chain logistics will result in a reduction in the cost of the fresh food purchased by the customer, making the customer satisfied with our service and thus making the customer more inclined to choose and purchase fresh food cold chain products. The optimization of the cold chain logistics distribution route is the key to achieving full process optimization. The study of supply chain optimization for cold chain logistics companies has contributed to their development, effectively reducing costs, saving money as well as increasing revenue [2].
2 Relevant Conceptions 2.1 Theory of Cold Chain Logistics Cold chain logistics, also known as low temperature logistics, is a way of freezing, storing, transporting, storing and transporting goods. The temperature of each of his logistics links is kept at a specific temperature through the use of various refrigeration technologies and refrigeration equipment, so that the environment they are in is maintained at a specific temperature all the time, which can not only protect the quality of food, but also reduce unnecessary losses and reduce the consumption of logistics costs [3, 4]. Since the 21st century, with the rapid development of science and economy, people’s standard of living has gradually begun to improve. People are not just chasing after filling their stomachs, but are beginning to pursue a high quality of life, and fresh food is increasingly being enjoyed by most people as an indispensable delicacy on their tables. Due to the special nature of fresh food, the number of goods that need to be delivered through cold chain logistics is also increasing year by year. The requirements of cold chain logistics are relatively high, so if we want to promote the rapid development of cold chain logistics, we need to vigorously develop our refrigeration technology. In addition, relevant policies from the national government are needed to support this [5]. 2.2 Joint Distribution Joint Distribution is the combination of a number of companies in an area in order to improve the efficiency of the logistics of the company. The company delivers the fresh food to a professional third party logistics company, which organises the information and arranges the transport, so that the fresh food is transported by different companies to different sorting centers, and then delivers the different fresh food products to the corresponding customers according to their needs. The main objective of joint distribution is to rationalise distribution routes, save costs, reduce loss of goods and increase customer satisfaction [7].
698
W. Du et al.
2.3 Vehicle Distribution Routing Issues The logistics and distribution vehicle route optimization problem refers to the use of limited vehicles or resources to meet the needs of our known customers, which in a nutshell means using minimal resources to maximize our customers’ needs. In practical terms: we need to make good use of each vehicle, taking their on-board rated weight, maximum mileage and other issues into account, and then develop a distribution route with the lowest cost [8].
3 Analysis of the Current Situation and Problems of Common Distribution of Fresh Food Cold Chain Logistics in Wuhan 3.1 Fresh Cold Chain Logistics Development in Wuhan At the algorithm level, vehicle routing problem (VRP) can be abstractly solved in digraph, undirected graph, connected graph and network graph. The cities and warehouses arriving in the logistics distribution route are represented by points, and the connection between the points represents the water, land and air routes between the two cities, which can clearly reflect the connection between each point and point. The research models of vehicle routing problem (VRP) mainly include mathematical model and network graph model. The mathematical model has the advantages of large capacity, high flexibility and strong versatility than network graph model, so the mathematical model is often used [9, 10]. 3.1.1 Growth of Cold Chain Market Demand in Wuhan Information shows that the current frozen food market in China, the annual growth rate is about 10%. Wuhan cold storage development potential is great, Wuhan is located in the “nine provinces through” the geographical location, with radiation around the provinces and the role of the local. Wuhan consumes 800,000 tons of frozen products every year, of which the demand for meat, seafood and other frozen products is about 350,000 tons, of which the demand for freshwater fish is about 300,000 tons. There are also markets for vegetables and fruit, all at a low temperature of 0–5 °C. Tens of thousands of tonnes of fresh food are shipped out every day from cold stores across Wuhan. According to statistics, about 50% of the food consumed by Wuhan residents on a daily basis belongs to fresh products. With that in mind, in order to meet the daily needs of the residents, it is necessary to ensure that 1 million tonnes of fresh food can be stored to meet the rapid growth in demand. The people of Wuhan are constantly improving their quality of life and pursuing a more nutritious and healthy lifestyle in parallel with technological progress. As a result, people also have higher demands for the quality and variety of fresh produce. The rapid growth of demand in the cold chain market has prompted a more complete system of fresh cold chain logistics in Wuhan. As the demand and quality of fresh food is getting higher and higher among Wuhan residents, this has put forward high requirements for the distribution of fresh food in Wuhan, prompting the development of the cold chain market in Wuhan.
Logistics of Fresh Cold Chain Analysis of Joint Distribution Paths
699
3.1.2 Wuhan Cold Chain Logistics Infrastructure The Baishazhou Cold Chain Project is the largest cold storage project in Central China, covering an area of 65,000 tonnes and covering an area of approximately 1,000 kms, which is no less than that of the Xudong Cold Storage. Its capacity has reached the point where it can drive the storage and distribution of fresh food products in various regions such as South China and North China. The development of the Baishazhou cold chain project has greatly eased the existing storage capacity of Wuhan from the source and promoted the rapid development of the fresh food cold chain logistics industry in Wuhan and even in various regions of Hubei. However, the perishable nature of fresh food has led to difficulties in the distribution process, high distribution costs and difficulties in ensuring the quality of goods. Therefore, this aspect of cold chain logistics and distribution in Wuhan is not yet developed enough [11]. 3.2 Wuhan Cold Chain Logistics Infrastructure Improvement 3.2.1 Low Level of Distribution Information Cold chain products are complex and varied, with each product having different characteristics. Each customer is independent and distributed in different locations. The demand of customers fluctuates greatly and is not stable enough, so there is no way to achieve centralised distribution without pulling through and sharing information with each other. The fresh food cold chain in Wuhan still adopts the traditional distribution method, where the distribution center is in individual contact with the customer and rarely collects the customer’s information for map route analysis. Therefore, the level of information technology in the whole distribution process is low, and a mechanism or system needs to be set up to pull through information and then carry out comprehensive processing [12, 13]. 3.2.2 Unreasonable Capacity Arrangement Due to the large number of customer types, different customer demands and low level of information technology, the distribution center is unable to obtain effective and accurate distribution data and can only make deliveries based on past experience. Some need to be delivered in time because of the time, then arrange for special vehicles to deliver, while others have a low loading rate of delivery vehicles due to the different time, often resulting in vehicles being idle but not used, and low loading rates leading to wasted costs. 3.2.3 Poorly Arranged Distribution Routes Wuhan City Cold Chain Center has not formed a complete distribution logistics system, relying on past guesses to develop distribution routes, no mechanism to map out the number and location of customer needs, how to effectively use vehicles to make deliveries, and maximize the use of the vehicle’s capacity as much as possible. It is still stuck in customer-specific routes, planning routes at will without optimizing them.
700
W. Du et al.
3.2.4 High Distribution Cost Distribution costs are too high mainly because of the following three aspects: first, the arrangement of vehicles is unreasonable, resulting in low personnel utilization, personnel efficiency is not up to standard, wasting great manpower costs; second, the distribution route is unreasonable, resulting in the vehicle driving distance repeated or multiple trips, the vehicle return journey becomes longer; third, the vehicle cost is high, various unreasonable applications of vehicles, resulting in the vehicle fuel costs, maintenance costs are too high.
4 Application of the Mileage Saving Method 4.1 The Principle of the Mileage Saving Method The mileage saving method is one of the most efficient heuristic algorithms designed for the problem of uncertainty in the number of distribution vehicles. Its basic principle is: according to the two sides of the triangle and must be larger than the third side, if the distribution center distributes goods to two distributors or customers, the distance assigned to one car for every two customers is greater than the distance assigned by one car in order of precedence. As shown in Fig. 1 to distribute goods from distribution center P to customers A and B, two cars are needed and the car transport distance is 2PA + 2PB, but if the goods of customers A and B are loaded on the same car in accordance with the order of departure from distribution center P, arrival at customer A, unloading the goods of customer A, and then going to customer B, unloading the goods of customer B and returning, only one car is needed and the car transport distance is PA + PB + AB, then according to what we said earlier the sum of the two sides of the triangle is greater than the third side, get PA + PB > AB, saving mileage for PA + PB-AB [14–17]. BB
AA
PP
AA
BB
PP
Fig. 1. The principle of the mileage saving
The basic idea of the mileage saving method is that, subject to some qualitative constraints, such as the weight and volume of the vehicle, the first vehicle of the distribution unit delivers the goods of the specified maximum weight and volume in the order of the customer’s location under the optimal route, while the second vehicle is loaded in the same way as before. In this way, the customer’s needs are all met and the route of distribution is optimized thereby achieving cost savings.
Logistics of Fresh Cold Chain Analysis of Joint Distribution Paths
701
4.2 Conditions of Application of the Mileage Saving Method On this basis, combined with the basic idea of the mileage saving method, several basic conditions for distribution route optimization using the mileage saving method have been summarized [18–21]. 1) 2) 3) 4) 5) 6) 7)
The location of the distribution center is known. The location and demand of customers are known. The distribution center has sufficient vehicles for distribution. The delivery time can meet the customer’s demand. The distance between the distribution center and the customer is known. One customer can only be delivered by one vehicle. One vehicle can deliver to more than one customer.
4.3 Model Construction for the Mileage Saving Method According to the data in the above table, the results are shown in Table 4. When the consistency ratio Cr is less than 0.1, the consistency test is passed, and the judgment matrix does not need to be adjusted again. 4.3.1 Model Description When a distribution center carries out logistics distribution to two or more customers, it is important to establish logistics distribution routes according to the specific needs of customers, so as to achieve cost savings and reduce losses. Based on the Baishazhou logistics center in Wuhan, a customer-oriented logistics and distribution route model is constructed. Assume that the Baishazhou fresh food distribution center is P, which simultaneously delivers and transports to various fresh food customers, each with a demand gi (i = 1, 2, 3 n), and that the Baishazhou logistics center has m existing vehicles of 4 tons and 2.5 tons respectively. We must design the shortest route for distribution while meeting the requirements of different customers to achieve cost savings, improve efficiency and enhance customer satisfaction. 4.3.2 Related Variables Several relevant parameters are explained as follows. n: indicates the number of customers. m: indicates the number of vehicles carrying out distribution transport. gi: the quantity of fresh food demanded by the customer. d ij : indicates the distance between two customers (i, j = 1,2……n,where i or j is equal to 0, the point is the Baishazhou fresh produce distribution center, such as dpj, which indicates the distance from distribution center P to customer j). Q: indicates the maximum carrying capacity of the distribution vehicle. D: indicates the maximum number of miles that the distribution vehicle can travel. ns : indicates the number of fruit shops to be reached by the distribution vehicle for distribution (s = 1, 2……n, when n = 0, it means that the vehicle is not involved in the distribution).
702
W. Du et al.
x ijs : the value of x is 0 or 1, 1 means that the delivery vehicle s is travelling from customer i to customer j, 0 means that the delivery vehicle s is travelling from customer j to customer i. yis : the value of y is either 0 or 1, with 1 indicating that the goods of customer i are delivered by s and 0 indicating that the goods of customer i are not delivered by s. 4.3.3 Model Building If the number of miles (S) of distribution completed by the distribution center to each customer is taken as the minimum objective function, a model is developed based on the relevant variables as follows [22–24]. n n m minS = dij xijs (1) i=0
m s=1
n i=0
j=0
yis =
1, i = 1, 2...n m, i = 0
i=0
i=1
(2)
gi yis ≤ Q, s = 1, 2 . . . m
n n n
x=1
j=0
dij xijs ≤ D
xijs = yis , j = 0, 1, 2 . . . n
(3) (4) (5)
In the model established above, formula (1) indicates that the transport vehicles required to complete the entire distribution task are m vehicles. Equation (2) indicates that the demand of any one vehicle to deliver customers cannot exceed the rated weight of the truck. The formula (3) indicates that the distance travelled by any one vehicle cannot exceed its maximum mileage. Equation (4) means that there is one and only one delivery vehicle reaching any one customer during a delivery. Equation (5) means that one and only one vehicle leaves the fresh food shop during a delivery.
5 Design of a Common Distribution Route Optimization Scheme 5.1 Data Selection When using the mileage saving method for route optimization, the first step is to know the quantity of goods demanded by the fresh produce shop. Since the goods sold by the fresh food shops are all fresh and have high requirements for freshness, the frequency of stocking for the fresh food shops is daily. From Table 1 and Table 2, we can see that this paper selects 10 fresh food shops to divide their annual demand evenly as a day’s demand for data processing. 10 tons of goods are delivered each day, and according to the mode of direct delivery of each fresh food shop, a total of 10 refrigerated trucks are needed, and a total of 206 km is needed for the round trip.
Logistics of Fresh Cold Chain Analysis of Joint Distribution Paths
703
5.2 Draw a Simplified Road Map In order to provide a more visual representation of the role of the mileage saving method in path optimization, and also for better calculation, a geographical map of the fresh food shops and the distribution center was created based on the geographical location of each fresh food shop. Table 1. Table of Demand for Fresh Food Stores Order number
Fresh food shop
volume of demand(t)
1
A1
0.8
2
A2
1.4
3
A3
0.9
4
A4
0.7
5
A5
1.4
6
A6
1.1
7
A7
0.9
8
A8
0.7
9
A9
0.6
10
A10
1.5
11
sum
10
Table 2. Table of Distance between Fresh Food Store and Distribution Center Order number Fresh food shop Distance from distribution center (km) 1
A1
12
2
A2
11
3
A3
9
4
A4
12
5
A5
10
6
A6
16
7
A7
6
8
A8
6 (continued)
704
W. Du et al. Table 2. (continued) Order number Fresh food shop Distance from distribution center (km) 9
A9
12
10
A10
9
11
sum
103
As can be seen from Fig. 2, P denotes the Baishazhou fresh produce distribution center, A1 , A2 , A3 , A4 , A5 , A6 , A7 , A8 , A9 and A10 denote the locations of the individual fresh produce shops, PA1, PA2, PA3 , PA4 , PA5 , PA6 , PA7 , PA8, PA9 and PA10 denote the distances between the distribution center and the fresh produce shops, and then based on their locations to produce A simplified route distribution diagram, from which the distances between the units can be clearly understood.
66
4
99
A A33
88
A A55
77
11
99
77
A A22
88
8
99 9
88
10 10
10 10 7
66 99
66
14 14
10 100
66 55
A A77
A A1010
A A11
12 12
66
A A66
12
A A44
A A99 12 12 A A88
Fig. 2. Road map
5.3 Optimal Design of Paths Based on the data we collected we started to calculate the mileage savings between customers using the mileage savings method where the sum of the two sides of a triangle is greater than the third side, using the principle that the mileage savings equals (PA + PB − AB), to create a mileage savings table, as shown in Table 3. The mileage saving method is used for route optimization calculations, and a mileage saving ranking table is produced in descending order of mileage saved, as shown in Table 4.
Logistics of Fresh Cold Chain Analysis of Joint Distribution Paths
705
Table 3. Save Odometer A1 A2
16
A2
A3
6
12
A3
A4
1
7
13
A4
A5
0
0
2
13
A5
A6
0
0
0
11
16
A6
A7
0
0
0
0
0
13
A7
A8
0
0
0
0
0
8
7
A8
A9
10
2
0
0
0
2
1
6
A9
A10
15
7
0
0
0
0
0
0
11
From Table 4, it can be seen that A1A2 has the largest amount of mileage savings, based on the principle of the mileage saving method so the optimal distribution route can be designed, as shown in Table 5: the first distribution route is P-A1 -A2 -A10 -P, saving 31 km. The second distribution route is P-A4 -A5 -A6 -A8 -P, saving 37 km, the third distribution route is P-A7 -A9 -P, saving 1 km, and the fourth distribution route is P-A3-P. Table 4. Saving Sorting Table Order number
Connection point
Saving mileage
Order number
Connection point
Saving mileage
1
A1 A2
16
12
A2 A10
7
2
A5 A6
16
13
A2 A4
7
3
A1 A10
15
14
A7 A8
7
4
A3 A4
13
15
A1 A3
6
5
A4 A5
13
16
A8 A9
6
6
A6 A7
13
17
A3 A5
2
7
A2 A3
12
18
A2 A9
2
8
A4 A6
11
19
A6 A9
2
9
A9 A10
11
20
A1 A4
1
10
A1 A9
10
21
A7 A9
1
11
A6 A8
8
22
A1 A5
0
706
W. Du et al. Table 5. Path comparison table Optimal path
Saving mileage
Tonnage
P-A1 -A2 -A10 -P
31
3.7
P-A4 -A5 -A6 -A8 -P
32
3.9
P-A7 -A9 -P
1
1.5
P-A3 -P
0
1.5
6 Conclusion In recent years, China’s logistics industry has developed rapidly, but due to the small market share and low profits of various logistics companies, many enterprises have focused on initiatives to improve efficiency and save costs, including Deppon Express in 2021, which has started various actions to increase efficiency and reduce costs. This paper examines the path optimization of cold chain fresh produce common distribution in Wuhan by reviewing relevant research theories at home and abroad to understand the methods of common distribution path optimization in logistics at home and abroad. The paper adopts the mileage saving method, and through the principle of mileage saving method, ten fresh food shops are selected to study the path optimisation of the distribution route of Baishazhou distribution center in Wuhan city, which achieves cost saving, loss reduction and efficiency improvement. Acknowledgment. This project is supported by Humanities and Social Sciences Research Planning Project of Ministry of Education. Research on the "double-edged sword" effect of leadership empowerment on employees and team creativity from the perspective of digital transformation. (22YJA630097).
References 1. Yao, Y.: Research on Optimization of Regional Agricultural Products Cold Chain Logistics Distribution, p. 08. China Agricultural Press, Beijing (2020) 2. Ting, P.-H.: An efficient and guaranteed cold-chain logistics for temperature-sensitive foods: applications of RFID and sensor networks. MECS Press IJIEEB 10(05) (2013) 3. Gu, W., Lu, C.: Research on path planning based on comparative particle swarm and mileage saving algorithm. China Automotive 08, 36–39 (2019) 4. Wang, H., Wang, X.: Optimization of the distribution path of Zhongbai supermarket based on the mileage saving method. Logist. Technol. 36(03), 84–87+157 (2017) 5. Yan, Y., Lai, S.: Distribution route optimization based on mileage saving method. Hunan: J. Hunan Ind. Vocat. Technol. Coll. 17(01), 29–32 (2017) 6. Li, B., Yang, L.Y., Mo, L.: Optimization analysis of express logistics distribution network in Yizhou city. Coastal Enterprises Technol. 01, 16–19 (2018) 7. Sun, S., Sun, J., Ji, S.: Research on the distribution scheme of Jilin Chunguang Dairy based on the mileage saving method. Logist. Sci. Technol. 41(06), 43–45 (2018)
Logistics of Fresh Cold Chain Analysis of Joint Distribution Paths
707
8. Guo, Y., Li, J.: Research on the optimization of the same city distribution path of fruit products considering customer classification–take Shandong XX Company as an example. Rural Econ. Technol. 31(02), 69–72 (2020) 9. Fan, H.: Study on the optimization of “fruit and vegetable butler” distribution system in Shijiazhuang. Hebei University of Science and Technology, Wuhan (2018) 10. Lu, X.H.: Research on optimization of drug distribution path of GK Guilin Pharmaceutical Distribution Company. Guilin University of Technology, Guilin (2021) 11. Jing, Y.: Research on the improvement strategy of logistics capacity of Lanzhou Best Express. Lanzhou Jiaotong University, Lanzhou (2019) 12. Youssef, B., Belkora, M.-J.: Distribution strategies toward nanostores in emerging markets: the Valencia case. Interfaces 47(6), 505–517 (2017) 13. Lin, S., Yang, J., et al.: Low carbon path optimization of logistics distribution based on mileage saving method. Logist. Eng. Manag. 41(04), 80–82 (2019) 14. Kong, L., Heng, L., Luo, H., et al.: Sustainable performance of just-in-time (JIT) management in time-dependent batch delivery scheduling of precast construction. J. Clean. Product. 193684–193701 (2018) 15. Chen, J., Shi, J.: A multi-compartment vehicle routing problem with time windows for urban distribution-a comparison study on particle swarm optimization algorithms. Comput. Ind. Eng. 13395–133106 (2019) 16. Liu, B.: Research on dynamic vehicle routing optimization method for cold chain logistics distribution with time window. Beijing Jiaotong University, Beijing (2018) 17. Raut, R.-D., Gardas, B.-B., Narwane, V.-S., et al.: Improvement in the food losses in fruits and vegetable supply chain - a perspective of cold third-party logistics approach. Oper. Res. Perspect. (2019) 18. Wan, Y.: Research on the optimization of fresh agricultural products distribution path from the perspective of low carbon. Kunming University of Technology, Kunming (2021) 19. Rezaei, N., Ebrahimnejad, S., Moosavi, A., et al.: A green vehicle routing problem with time windows considering the heterogeneous fleet of vehicles: two metaheuristic algorithms. Eur. J. Ind. Eng. 13(4), 507–510 (2019) 20. Li, J., Liu, M., Liu, P.: Route optimization of cold chain logistics vehicles with multiple models of fresh agricultural products. J. China Agric. Univ. (07) (2021) 21. Liu, G., Hu, J., Yang, Y., et al.: Vehicle routing problem in cold chain logistics: a joint distribution model with carbon trading mechanisms. Resourc. Conserv. Recycling (2020) 22. Grigorios, D., Konstantakopoulos, Sotiris, P., et al.: Vehicle routing problem and related algorithms for logistics distribution: a literature review and classification. Oper. Res. (2020) 23. Sheng, H.: Research on the vehicle routing problem of fresh e-commerce logistics distribution. University of Electronic Science and Technology, Chengdu (2019) 24. Christian, F.: A decision support system to investigate food losses in e-grocery deliveries. Comput. Ind. Eng. (2018)
Optimization of Logistics Distribution Route Based on Ant Colony Algorithm – Taking Nantian Logistics as an Example Zhong Zheng1,2(B) , Shan Liu1 , and Xiaoying Zhou3 1 School of Transportation, Nanning University, Guangxi 530200, China
[email protected]
2 King Mongkut’s Institute of Technology Ladkrabang, Bangkok 10520, Thailand 3 Guangxi Wuzhou Communications Co., Ltd., Nanning 530200, China
Abstract. The continuous development of social economy and the people’s growing yearning for a better life have promoted the rapid development of e-commerce, and urban distribution business has become increasingly important. Urban distribution is also faced with relatively complex task requirements, such as short time window, strict requirements, complex road conditions, high frequency of tasks, etc. Choosing the right path has become the key issue to improve distribution efficiency. Aiming at the problems encountered in the distribution of urban distribution routes, the paper uses ant colony algorithm to analyze and optimize the distribution routes, and uses MATLAB programming to solve the optimized distribution scheme. By comparing the data before and after optimization, it is found that the logistics transportation cost is effectively reduced, and the logistics transportation resources are saved, which proves the feasibility of the algorithm, and provides a certain reference for improving the efficiency of urban distribution and reducing the distribution cost. Keywords: Ant colony algorithm · Path optimization · Intra-city distribution
1 Introduction The logistics distribution in cities is faced with unique and complex distribution environment and requirements. The distribution volume is small and the frequency is high. The order distribution demand is personalized, there are many freight points, and the distribution of transportation networks in the service area is uneven. The distribution road conditions are usually complex. How to take into account the time, economic cost and quality of urban distribution has become a concern of more and more theoretical research and enterprise practice. The improvement of transportation path requires reasonable planning [1, 2]. The problem of route distribution has attracted the attention of many scholars. Dai Tingting and others proposed an improved ant colony optimization algorithm where the ant colony algorithm is calculated at a “20 × 20” grid environment, and the search efficiency has been greatly improved [3]; Luyuan et al. proposed an improved ant colony © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 708–718, 2023. https://doi.org/10.1007/978-3-031-36115-9_63
Optimization of Logistics Distribution Route
709
algorithm. The improved ant colony algorithm has better search ability and improved convergence speed [7]; Liu Ziyu, Zhao Lixia and others improved the ant colony algorithm by using the piecewise function to adjust the convergence speed of the algorithm, which has the advantage of effectively reducing the path length, jumping out of the local optimum and accelerating the convergence speed [5]; Ge Dayun started from the multimodal transport network under a single task, combined with ant colony algorithm to design the model solution method, and completed his goal with the minimum cost by integrating various advantages of multimodal transport [6]; Li Shuangshuang et al. proposed a new pheromone update algorithm and a global tendency pheromone distribution model, used the grid method to build a two-dimensional plane space with obstacles and carried out simulation experiments, and the results verified the effectiveness of the model [7]; Li, Sidi et al. proposed a knowledge-based hybrid ant colony algorithm to improve the overall satisfaction of all tourism groups, improve the performance of the algorithm, and avoid the generation of local optimal solutions [8]. Aiming at the problems of the above ant colony algorithm in path planning, this paper takes Nantian Logistics Company as an example, combines the specific situation of the enterprise, and proposes an ant colony algorithm based on practical applicability. By going to the enterprise to investigate the needs of customers, obtain actual data, and build a model based on data processing. Finally, apply the model and algorithm to the actual situation of the enterprise, and then obtain the optimized distribution plan through MATLAB programming.
2 Problem Description 2.1 Description of Problems Faced by Enterprises As a city distribution company, Nantian Logistics is under great pressure every day, and often due to the untimely vehicle arrangement, some urgent goods cannot be delivered to customers in time, which affects the work process and satisfaction of customers. The business volume is small and the frequency is high. The current distribution path and sequence mainly depend on manual judgment; It is unable to maximize the role of vehicles. Some lines pass repeatedly, but the goods are not arranged on the same vehicle, resulting in the waste of vehicle resources, which also reflects the low level of management. 2.2 Modeling and Optimization of Nantian Logistics Distribution Path The logistics center in this study is used for this project with a total of 4 transport vehicles; The maximum transport capacity of each vehicle is 2500 kg; The goods distribution center number is 0, and the customer number is 1–16; The earliest departure time of the transport vehicle is 9:00 in the morning; Each customer has the same service time. The quantity requirements and geographical coordinates of each customer’s goods are shown in Table 1.
710
Z. Zheng et al. Table 1. Customer parameters
Customer No
North latitude
East longitude
0
22°55 01.46
108°21 35.77
Number 0
1
22°51 44.75
108°18 02.95
21
2
22°51 39.20
108°15 37.33
3
22°49 06.46
108°18 24.67
4
22°52 24.49
108°24 09.54
5
22°52 33.75
Weight
Volume
0
0
367.5
0.84
3
296
1.52
30
240
2.638
10
200
0.545
108°24 04.76
39
235.58
0.897
6
22°52 35.76
108°25 05.98
10
400
2.4
7
22°45 30.74
108°26 46.36
6
125.4
0.24
8
22°47 58.38
108°16 45.43
3
44
0.066
9
22°47 39.73
108°18 35.03
1
11
0.0165
10
22°51 02.41
108°18 13.21
42
735
1.68
11
22°52 26.46
108°16 26.32
19
325
0.546
12
22°47 57.05
108°22 03.50
5
50
0.38
13
22°49 05.98
108°26 51.23
23
460
0.69
14
22°43 57.06
108°22 03.52
41
818
1.23
15
22°47 57.08
108°31 48.04
36
718
1.08
16
22°50 40.13
108°18 21.08
53
1058
1.59
Data source: based on the relevant data of the arrival of Nantian Logistics Group in January.
3 Model Establishment Taking 16 customer enterprises served by Nantian Logistics as the object of this study, the content of distribution work mainly belongs to the intra-city distribution in Nanning. The steps to build the model are as follows. 3.1 Determine Coordinate Points First of all, according to the data obtained in this practice, take multiple transportation destinations as an example, obtain the longitude and latitude according to the specific location of the address, and then convert them into coordinate points through the longitude and latitude and XY conversion software. After obtaining the coordinates of 12 transportation points, because the algorithm mainly calculates the distance between points, the origin is not set. The coordinate points use the map to query the distance between them (unit: km), and the distance matrix shown in Fig. 1 is generated. In the figure, 1–16 corresponds to the coordinate points of 16 transportation points, of which 0 is the headquarters of Nantian Logistics Group.
Optimization of Logistics Distribution Route
711
Fig. 1. Transportation point distance matrix
The logistics delivery vehicle needs to start from point A and deliver the goods to customers 1–12. For distribution companies, they also need to consider the requirements of distribution vehicle routes and costs. On the premise of meeting the constraints of loading weight and route, it is necessary to arrange the optimal transportation order and select the optimal transportation path, so as to achieve good results in distribution efficiency and cost control [9, 10]. According to the specific situation of Nantian Logistics, this paper uses ant colony algorithm to analyze this problem. 3.2 Assumptions The construction of the model determines the following six assumptions: (1) The distance between two points is the shortest. Set it as the shortest actual distance; (2) The traffic condition is good and there is no congestion; (3) The geographical coordinates of each distribution center are known, and there will be no shortage of vehicles [11]; (4) Knowing the needs of each customer, the goods can be mixed under the condition of meeting the on-board conditions; (5) The truck is limited in weight, starting from the distribution center and finally returning to the starting point; (6) The total distance required by each distribution plan will not exceed the maximum distance traveled by the distribution vehicles per tank of oil [12–15]. 3.3 Parameter Setting Set the number of ants participating in the algorithm model to m, Based on Ma Ning’s research on the optimal path of ant colony algorithm [5], the update and transfer probability of information elements are set, and the calculation formula is shown in (1) and (2). τij (t + 1) = ρτij (t) + τij
(1)
712
Z. Zheng et al.
Pijk (t) =
⎧ ⎪ ⎪ ⎨ ⎪ ⎪ ⎩
β
τijα ηij
β ταη s∈tabu / k is sj
, j∈ / tabuk (2)
0, otherwise
Explanation of symbol expression: τ ij represents the difference of information elements; τij refers to the value of path (i, j) information element at time t; ρ Indicates the retention of information elements on the path; η Ij is the transfer probability and expectation from location i to location j; α Indicates the information elements accumulated by ants; β Expectations for transfer; Tabuk is all the nodes that the ants walk through [13, 15]. At the beginning of the model, the starting point of each ant is a randomly selected city. By maintaining the memory vector of a path, record the city points that the ant passes by in turn. When constructing each path, the ant perceives the pheromone concentration on a path and continues to walk along that path according to the pheromone concentration [14, 15]. Ant colony algorithm has been proved to achieve good results in various path selection calculations [16–19].
4 Model Solution and Result Analysis 4.1 Model Construction 4.1.1 Model and Parameters It is assumed that the delivery address and the customer’s requirements are known. In the whole model, the number of ants, that is, the transportation vehicles in the freight center is m, and dij represents the distance between customer i and customer j; The pheromone concentration of each path is the same at the initial time point and between the points; Set pheromone factor α The value of is 1; Set heuristics factor β The value of is 3, set the pheromone volatilization factor ρ The value of is 0.85; The number of customers is n (set is N), the vertex set in the network is D (including distribution center and customer set), the distribution center can be assigned a value of 1, and the distribution vehicle set is K. N, D and K are the contents of (3) respectively. N = {2, 3, . . . , n} D = {1} ∪ N K = {1, 2, . . . , m}
(3)
The departure time of the distribution vehicles in the cargo distribution center is T0, the quantity of goods transported is Q, and the customer set served by the transport vehicle k is expressed as Sk. The decision of variables is shown in (4) and (5). ⎧ ⎪ ⎨ 1, k(i → j) (4) Xijk = ⎪ ⎩ 0, otherwise
Optimization of Logistics Distribution Route
Yik =
1, k(i) 0, otherwise
The mathematical representation of the model is shown in (6) to (12). dij Xijk min k∈K
i∈D
i∈D
qi Yik ≤ Qk
k∈K
i=N
Xojk =
i∈D
j∈D
i∈D
i∈N
i,j∈Sk
(5)
(6) (7)
Xijk = 1
(8)
Xiok = 1, ∀j ∈ k
(9)
Xijk = Yjk , ∀j ∈ N, ∀k ∈ K
713
Xijk ≤ |Sk | − 1, ∀k ∈ K X = Xik ∈ D
(10) (11) (12)
Among then, Eq. (6) is the path constraint (the shortest distance); Eq. (7) is the constraint on the carrying capacity of the distribution vehicle (less than the maximum carrying capacity of the vehicle); (8) Is the constraint on the number of delivery and service times (1 time); Eq. (9) is the constraint of the distribution vehicle path (cannot take the circular transportation route, and must return to the distribution center after distribution); Eq. (10) is the constraint of the delivery order (first delivery i and then customer j); Eqs. (11) and (12) are path constraints (each distribution point and path are interconnected). 4.1.2 Update Pheromone The expression of Pijk for the probability of selecting the next customer after completing the delivery of goods from the previous customer is shown below. ⎧ ⎪ [τij (t)]α [ηij ]β ⎪ ⎨ α β s=allowk [τij (t)] [ηij ] ,j=allow k (13) Pijk = ⎪ ⎪ ⎩ 0, otherwise τij (t + 1) = ρτij (t) + τij (t, t + 1) τij (t, t + 1) =
τijk =
m k=1
τijk (t, t + 1)
⎧ ⎪ ⎨ q , k(t, t + 1) → (i, j) lk ⎪ ⎩
0, otherwise
(14) (15)
(16)
714
Z. Zheng et al.
In formula (13), allow is the set of the next service object selected by vehicle k after completing the service of the previous customer at time t; τij (t) represents the information elements from customer object i to customer object j at time t (the initial value is set as a constant); ηij (t) represents the visibility value from customer object i to customer object j at time t; α and β. It represents the degree to which pheromones and heuristic factors affect path selection. If the collection of the next customer is empty, it means that an iteration cycle has been completed. The next iteration cycle starts from t + 1 and dynamically updates the information elements according to the information elements remaining in the previous cycle. In formula (14) ρ (value 0–1) indicates the persistence of information elements; Eq. (15) represents the information element increment of path i to j f rom time t to time t + 1, m is the distribution vehicle; In formula (16), Q is the total amount of information elements released in an iteration cycle; 1k is the length of the path that the transport vehicle k passes in an iteration cycle. 4.1.3 Algorithm Steps The algorithm mainly has the following four steps. (1) Initialize various parameters to determine the distance between Nantian logistics distribution center and each customer, as well as the distance between customers. (2) Set the distribution vehicle to depart from the cargo center (i = 1); According to the conditions of (11), after completing the demand service for the jth customer, find the next customer who needs service, and repeat this; After many searches, it provides distribution services for all customers in the system; Finally, return to Nantian Logistics Hub. (3) Select the next customer through formula (13), find the next customer meeting the shortest path from the previous customer service path set, and redistribute the delivery vehicles according to the customer’s service needs. (4) Then update the information elements according to formula (8), and return to (2) to start again; Until the maximum number of iteration cycles is reached. 4.2 Model Solution Results After using Matlab to optimize the model, the operation results not only delivered the goods to the customers in time, but also met the requirements of the model in terms of time, path selection and minimum cost.
Optimization of Logistics Distribution Route
715
Fig. 2. Route map of optimal distribution scheme
Fig. 3. Optimal route allocation diagram
5 Result Analysis It can be seen from Fig. 2 and Fig. 3 that the transportation plan only needs three vehicles, which are divided into three routes for transportation, and all the points have been completed and the distribution task has been completed on time. Table 2. Original data Vehicle No.
1
2
3
4
Driving route
0→4→5→6→0
0 → 12 → 13 → 7 → 14 → 0
0 → 2 → 1 → 11 → 16 → 0
0 → 10 → 15 → 3 → 9 → 8 → 0
Transport volume/t
835.58
1453.4
2046.5
1748
Mileage/km
16.5
62.9
28.6
34.7
Data source: compiled according to relevant data of daily work of Nantian Logistics Group
The vehicle allocation of Nantian Logistics is mainly manual. When receiving the transportation request from the customer, it will first print out the list to be delivered, and then put it neatly on the table, while the list with closer location will be arranged for delivery together. If there is an order that is far away, the delivery time will be delayed,
716
Z. Zheng et al.
and it will be arranged only when there is an order that can be mixed, so the original data assumed in Table 2 is available. It can be seen from the data in the table that the logistics center used four vehicles to deliver goods according to four routes before optimization, and completed all goods transportation tasks under the condition of meeting the customer’s time requirements. The completed transportation volume and mileage were 142.7 km, and the cost was shown in (17). Cost = 1000 ∗ number of vehicles used + total distance traveled by vehicles = 1000 ∗ 4 + 142.7 = 4142.7(yuan)
(17)
Table 3. Optimized data Vehicle No.
1
2
3
Driving route
0 → 16 → 1 → 10 → 8 → 9 → 3 → 0
0 → 4 → 5 → 6 → 13 → 14 → 7 → 12 → 0
0 → 15 → 2 → 11 → 0
Transport volume/t
2455.5
2288.98
1339
Mileage/km
42.2
60.6
27.8
It can be seen from the analysis results in Table 3 that the number of vehicles in some optimized schemes has been reduced from 4 to 3, and the corresponding routes have also been changed to 3. All customers’ freight tasks have been completed within the time required by customers. Comparing the two algorithms, we can see that although the traditional transportation routes also deliver goods, this is not the optimal solution in practice, and the improved arrangement is also more labor-saving. Cost = 1000 ∗ number of vehicles used + total distance traveled by vehicles = 1000 ∗ 3 + 130.6 = 3130.6(yuan)
(18)
From the comparison between Table 2 and Table 3, it can be seen that the improved algorithm saves 12.1 km of total vehicle mileage; At the same time, 1 vehicle was built for use; Compared with the original data, it saved 1012.1 yuan, as shown in (17) and (18). To sum up, an optimization model is established for the distribution of Nantian Logistics in the same city; Based on ant colony algorithm and combined with the actual situation of Nantian Logistics, it is more suitable for practical application. Through this case study, it can be seen that ant colony algorithm has effectively saved transportation resources, reduced logistics and transportation costs, and made great progress towards low cost and high efficiency in intra-city distribution.
Optimization of Logistics Distribution Route
717
6 Conclusion In the context of the continuous development of e-commerce, customers’ demand for online and offline single offline distribution is increasing, and urban distribution has become a key link to improve consumer satisfaction. At the same time, due to the constraints of urban congestion, large orders, small distribution orders and high frequency, the selection of appropriate routes has become an important issue of urban distribution. Aiming at the increasingly important urban distribution, the paper selects the typical distribution enterprise Nantian Logistics Group as the research object, and proposes a path planning scheme based on the data collection and analysis of urban distribution routes, and uses ant colony algorithm to analyze the path optimization. The practice of enterprises shows that the optimized scheme has effectively reduced the transportation distance, saved the logistics cost and significantly improved the logistics efficiency. Acknowledgments. This paper is supported by: Young and middle-aged teachers basic ability improvement project of Guangxi University “Research on Key Technologies of Urban and Rural Logistics Integration Operation Based on Big Data and Cloud Computing” (2023KY1862).
References 1. Tang, P., Tian, H.: Research on the layout of national economic mobilization logistics centers. Int. J. Mod. Educ. Comput. Sci. 11, 44–50 (2010) 2. Liu, Z.: Application and research of two-population ant colony algorithm based on heuristic reinforcement learning. Shanghai University of Engineering and Technology, Shanghai (2020). (in Chinese) 3. Dai, T., Liu, X., Hu, Y.: Application of improved ant colony algorithm in path planning. J. Jiamusi Univ. (Nat. Sci. Ed.) 40(01), 123–125 (2022). (in Chinese) 4. Land, M., Gao, H., Cui, Y.: Application of improved ant colony algorithm in express delivery routing. Comput. Technol. Dev. 31(11), 15–20 (2021). (in Chinese) 5. Liu, Z., Zhao, L., Xue, J., et al.: Research on improved ant colony algorithm for vehicle routing problem. J. Hebei Univ. Sci. Technol. 43(01), 80–89 (2022). (in Chinese) 6. Ge, D.: Optimal path selection of multimodal transport based on Ant Colony Algorithm. J. Phys. Conf. Ser. 2083(3) (2021) 7. Li, S., Zhao, G., Yue, W.: Research on path planning for mobile robot based on improved ant colony algorithm. J. Phys. Conf. Ser. 2278, P012005 (2021) 8. Li, S., Luo, T., Wang, L., Xing, L., Ren, T.: Tourism route optimization based on improved knowledge ant colony algorithm. Complex Intell. Syst. (2022). (prepublish) 9. Wang, Y.: Research on optimization method of air logistics distribution path based on ant colony algorithm. Inf. Technol. 11, 76–80 (2021). (in Chinese) 10. Zheng, X., Qi, Q.: Optimization and application of low carbon cold chain logistics distribution model based on improved ant colony algorithm. Preserv. Process. 22(03), 83–90 (2022). (in Chinese) 11. Lei, J., Sun, Y., Zhu, H.: Application of improved ant colony algorithm in vehicle path planning with time window. Comput. Integr. Manuf. Syst. 1–15, 15 April 2022. http://kns. cnki.net/kcms/detail/11.5946.TP.20211228.1340.008.html 12. Wang, S., Fan, S., Wang, J., Shen, R.: Research on logistics distribution route optimization based on ant colony algorithm – taking Yuzhong District of Chongqing as an example. Econ. Trade Pract. 13, 31–32 (2017). (in Chinese)
718
Z. Zheng et al.
13. Chen, J., Liu, G.: Research on cold chain logistics path optimization based on ant colony algorithm. Transp. Technol. Econ. 23(05), 38–44 (2021). (in Chinese) 14. Zhu, Y., You, X., Liu, S.: Improved ant colony algorithm based on heuristic mechanism. Inf. Control 48(03), 265–271 (2019). (in Chinese) 15. Ma, N.: Logistics optimal path based on clustering ant colony algorithm. Sci. Technol. Eng. 20(31), 12911–12915 (2020). (in Chinese) 16. Zhang, W.: Application of an improved ant colony algorithm in coastal tourism route optimization. J. Coastal Res. 98(Special), 84–87 (2019) 17. Du, P., Hu, H.: Optimization of tourism route planning algorithm for forest wetland based on GIS. J. Discret. Math. Sci. Cryptogr. 21(2), 283–288 (2018) 18. Qian, X., Zhong, X.: Optimal individualized multimedia tourism route planning based on ant colony algorithms and large data hidden mining. Multimedia Tools Appl. 78(15), 22099– 22108 (2019) 19. Zhou, X.: Research on the optimization of tourism traffic routes based on ant colony algorithm. Revista de la Facultad de Ingenieria 32(3), 819–827 (2017)
Optimization Research of Port Yard Overturning Operation Based on Simulation Technology Qian Lin1(B) , Yang Yan1 , Ximei Luo1 , Lingxue Yang1 , Qingfeng Chen1 , Wenhui Li1 , and Jiawei Sun2 1 School of Logistics, Wuhan Technology and Business University, Wuhan 430065, China
[email protected] 2 Technical University of Munich Asia Campus, Technical University of Munich (TUM),
Singapore, Singapore
Abstract. The operation of the world transport supply chain is affected by many factors, including the COVID-19 epidemic. China’s production is in a relatively stable state, which to some extent alleviates the contradiction between supply and demand in the international market. With the increase of container transportation in China, the utilization rate of container transportation is getting higher and higher. The overturning rate will have a significant impact on the utilization rate and storage efficiency of containers. Based on the theory and practical experience of container port operation, this paper studies Huangshi New Port based on rail and water transport simulation technology. By optimizing the container turnover operation of the port yard and improving the container turnover technology, the operation cost of the port can be reduced, the operation efficiency of the port can be improved to a certain extent, and the competitiveness of the port can be enhanced. Keywords: Simulation platform of Rail-water Intermodal · Simulation technology · Box-turning operation
1 Introduction According to the 2021 Global Port Development Report released by the Shanghai International Shipping Scientific Research Center, the container throughput of the world’s top 100 seaports shows a good trend of growth on the whole, among which 84 ports have a positive growth year-on-year in 2020, and 41 ports have a growth rate of 10% higher than that in 2020 (see Table 1). At present, container transport plays an important role in the world freight business [1]. Port is an important hub of cargo transportation, and its operation efficiency is directly related to the competitiveness of cargo transportation in the port, even related to the overall development of the region. Container yard is a buffer zone for port containers [2–4]. The efficiency of a container yard not only determines the port throughput, but also determines the speed of cargo operation [5].
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 719–733, 2023. https://doi.org/10.1007/978-3-031-36115-9_64
720
Q. Lin et al.
Table 1. TOP20 domestic ports in terms of cargo throughput in the world in 2021 (unit: 10, 000 TEU) Range
Country
Name of port
In 2021,
In 2020,
1 (1)
China
Zhoushan, Ningbo
122405
117240
Year-on-year growth 4.4
2 (2)
China
Shanghai
76970
71104
8.2
3 (3)
China
Tangshan
72240
70260
2.8
4 (4)
China
Qingdao
63029
60459
4.3
5 (5)
China
Guangzhou
62367
61239
1.8
7 (7)
China
Suzhou
56590
55408
2.1
9 (10)
China
Rizhao
54117
49615
9.1
10 (9)
China
Tianjin
52954
50290
5.3
13 (13)
China
Yantai
42337
39935
6.0
14 (18)
China
Taizhou
35291
30111
17.2
15 (24)
China
Jiangyin
33757
24705
36.6
16 (15)
China
Dalian
31553
33401
−5.5
17 (17)
China
Huanghua
31134
30125
3.3
18 (16)
China
Nantong
30851
31015
−0.5
20 (20)
China
Shenzhen
27838
26506
5.0
Source: Shanghai International Shipping Center Prospective Industry Research Institute (Figures outside brackets are 2021 ranges, figures inside brackets are 2020 ranks)
2 Theoretical Review 2.1 Box-Turning Operation and Its Impact Port container overturning is an important part of container ship turnover, including container loading, lifting and moving. Port container turnover affects the loading and unloading efficiency, especially in loading and unloading operations [6–8]. The overturning problem refers to that an overturning sequence can be found under the given shellfish position structure, initial container position state and container loading sequence, and the minimum number of overturning is required [9]. Reducing the number of overturning is helpful to improve the efficiency of overturning and thus reduce port handling costs [10]. 2.2 Simulation Platform for Container Rail-Water Intermodal Based on the needs of the Yangtze River Economic Belt and regional economic development, the National Multimodal Transportation Project of Huangshi New Port uses 3D virtual technology to establish the real scene and complex decision-making environment, and makes the planning scheme of container Rail-water Intermodal [11–13]. On this
Optimization Research of Port Yard Overturning Operation
721
basis, this paper focuses on how to arrange cargo space, operation equipment scheduling and transportation path optimization under different transportation modes [14–17]. The 3D virtual simulation experiment platform was designed with B/S architecture and adopted 3D simulation modeling technology. According to the real situation of the storage yard, Maya, 3DMax and other simulation software were used to model the whole storage yard and simulate the incoming and outgoing storage environment [18, 19].
3 Case Analysis 3.1 Introduction to Huangshi Newport Huangshi New Port in Yangxin County, Huangshi City, Hubei Province, supported by the Ministry of Transport, is the core port area of “East Hubei Combined Port” planned and constructed by Hubei Province. It is a national first-class water transport port among 28 major inland river ports in China, and there are only 2 ports of this class in Hubei Province. Table 2. Port cargo and container throughput of Hubei Province in 2021 Unit of measurement: ten thousand tons, ten thousand TEU Port
Hubei Province Total
Throughput of cargo
Foreign trade cargo throughput
Throughput of container
Since the beginning of the year The cumulative
This month,
Year-on-year growth rate (%)
Since the beginning of the year The cumulative
This month,
Year-on-year growth rate (%)
Since the beginning of the year The cumulative
This month,
Year-on-year growth rate (%)
48, 831
5, 111
28.6
1, 787
146
−2.4
284
23
24.2
964
73
−7.7
248
20
26.1
731
67
6.2
4
…
−28.0
Jiayu
1, 225
118
18.8
Wuhan
11, 679
1, 503
10.8
Huangzhou
1, 007
238
15.6
Ezhou
2, 176
198
49.7
Huangshi
4, 992
495
5.9
Xiangyang
4
…
−48.3
Jingzhou
4, 375
410
23.0
42
2
9.5
15
1
21.5
Yichang
11, 470
981
41.3
50
4
−16.6
15
1
20.6
Qianjiang
104
18
−0.7
Hanchuan
9
1
31.8
Other Ports in Hubei Province
11, 790
1, 150
55.7
2
…
44.3
Tianmen
It is located in the source Wei mouth town board, belongs to the key area of the Yangtze river economic belt, due to the superior g geographical position, is rare in the
722
Q. Lin et al.
middle reach of Yangtze river of deep water port, even during the dry season, the port in front of the lowest water level also can continue to achieve 6 m. In 2021, the cargo throughput of Huangshi New Port increased by 1.8% compared to this time of year, among which the throughput of scattered and miscellaneous items reached 1932.70, 000t, an increase of 4% compared to this time of year, and the container throughput reached 44, 060TEU, a decrease of 28% compared with the same period in previous years. The transshipment efficiency of container throughput at Huangshi New Port is lower than that at other ports along the Yangtze River (See Table 2). Before the construction of the port, a broad reserve site has been planned. The overall planning construction land is nearly 20 km2 . The surrounding area has a good industrial base. The planned construction includes the container dock area, the large bulk cargo dock area and the bulk cargo dock area. It plans to build a total of 69 production berths, 46 berths will be designed in the near term and 23 berths in the long term. More than 500 million yuan will be invested in the first phase of the project. The target is to build nine 5, 000-ton class berths, which can also rely on 10, 000-ton ships, and the annual throughput can reach 10 million tons. After the port is fully completed, the annual throughput of the storage yard can reach “a large port of 100 million tons and one million Tank box”. In the future, it will be built into the largest foreign trade transfer hub port in the middle reaches of the Yangtze River and an excellent site selection base for the multi-modal transport of public iron-water goods and distribution. The final planning target of this port is to be the modern comprehensive port area (integrated transport hub of water and land) closest to the sea in Hubei Province, which will become a powerful engine for the transformation and leap-forward development of Huangshi area and play an important connection role along the riverbank It can also radiate the coastal areas to the east. It is a very rare comprehensive logistics base. Table 3. Main device configurations Type of equipment
Device Name
Bank bridge
Container handling 12 bridge
The number of
The rated lifting weight under the hook is 45t and the rated lifting weight under the spreader is 35t Lifting height: 20 m on the rail surface
Dragon door crane
Container gantry crane
Fixed lifting weight: 35t under the spreader Lifting speed: full load 20 m/min, no load 30 m/min
10
Main technical parameters
(continued)
Optimization Research of Port Yard Overturning Operation
723
Table 3. (continued) Type of equipment
Device Name
The number of
Front side crane
Front container crane
High stacking machine
Empty container stacking machine
12
Maximum lifting height: > 15000 mm Minimum height clearance: > 350 mm
Inner set card
Container truck
20
Maximum load: 9000 kg Loading container size: 40 ft.
2
Main technical parameters Wheelbase: 6000–6500 mm Maximum torque/RPM: ≥1770 N * m/2100 rpm
Through investigation, it is found that Huangshi New Port has built 6 professional container berths at the present stage, with the whole coastline of the berths being 606 m long. Each berth Each berth has two 2 shore bridges, 12 container unloading bridges, 10 gantry cranes, 2 front cranes, 12 empty container strollers and 20 container trucks. The port yard has been divided into 6 yard areas with a total area of 79, 200 m2 , among which there are 4 heavy container yard areas. Each yard area is equipped with 404TEU of 10 800 m2 , 2 yard bridges and 2 empty container yard areas. Each yard area is equipped with 138TEU of 9 000 m2 and 2 flow machines (see Table 3). 3.2 There Are Problems in the Box-Turning Operation of Huangshi Newport 3.2.1 Current Status of Storage Space of Huangshi Nerport Yard The current situation of Huangshigang storage yard can be analyzed from two aspects. (1) Storage yard space resource allocation method: According to the values of “number, area, space, row and layer” of the storage yard, the theoretical number of the storage yard is calculated through corresponding values. Storage yard space resources are generally allocated in two steps, that is, the amount of working containers in each container area of the storage yard is allocated reasonably and equally, and then the containers on site are assigned to the container area. (2) Container stacking arrangement: 1) Empty containers and heavy containers are stacked in a reasonable area; 2) Empty containers are stacked according to intact containers, damaged containers, dirty containers, self-owned containers and rented containers; 3) Empty and intact boxes are stacked according to the size, type and bracket of the boxes; 4) Heavy boxes are divided into the arrival boxes, delivery boxes, transit boxes and auxiliary boxes; 5) The arrival boxes should be stacked separately by different owners; 6) TThe sending boxes should be stacked separately considering the name of the ship, the number of the ship and the bill of lading number; 7) According to the destination, according to the ship, ticket piled up, according to the carrier transit plan transit; 8)The same planned task container stack.
724
Q. Lin et al.
3.2.2 Reasons for Overturning Operation in Huangshi Newport The overturning operation will increase the operating cost of the dock. First of all, import containers should follow the “first in, first out” mode. That is, the first unloaded import containers should be transferred out first. The later unloaded containers cannot be stacked on top of the first unloaded import containers. However, in fact, when unloaded import containers have not been shipped out, in order to improve the utilization rate of the port terminal, the newly unloaded containers will be placed on top of the original containers, resulting in the export of imported containers turning over.And in the process of stacking import containers, because cannot know in advance the owner of the container pick up the order, so cannot be stacked in a certain order. In addition, there are some objective factors, such as special containers (dangerous goods, etc.) with special places, if the special places are occupied, there will be a turning operation; Secondly, improper port organization and scheduling will also lead to container overturning operation. For example, following the principle of “card collection priority”, in the process of card collection and container pick-up, the order of cargo owner is inconsistent with the order of container stacking in the storage yard, resulting in the operation of turning over the container. In this way, the running distance and running time of the field bridge vehicles in different stacks at the same bay position can be shortened. However, it will lead to unnecessary repeated operations, which will not only increase the cost of turning over the box, but also prolong the working time, resulting in a large amount of waste of waiting time for the collection card and increase the time cost. At the same time, it will also lead to the decrease of the utilization rate of the yard bridge, but also produce problems such as yard congestion.
4 Simulation and Optimization of Simulation Platform for Container Rail-Water Intermodal 4.1 Theoretical Analysis The root cause of container overturning problem is that the order of container pickup is inconsistent with the stacking order of port yard. The main reasons are: first, the import container storage management is not good; Second, the lack of collection card to the station related data; Thirdly, the container flipping strategy is unreasonable. Therefore, it is necessary to find appropriate ways to reduce container overturning rate from the following aspects. (1) Centralized supervision of containers at the port. Through centralized management of stacking containers, the port terminal adopts a more scientific stacking management mode to maximize the utilization rate of the port terminal and reduce the rate of overturning. Port container management has positive influence on container quality control and overall planning. (2) Centralized stacking of the same container owner import box set. This method requires that all ports, dock and shipping companies cooperate closely during transportation, find out the owners with a large number of imported containers in the information center, and arrange unified stacking and reasonable configuration for them, so that no overturning operation occurs when the same owner takes up the containers, that is, zero overturning of containers is realized.
Optimization Research of Port Yard Overturning Operation
725
(3) Under the same delivery list, import containers can be exchanged accordingly. Multiple import containers in the same bill of lading, pick up in no particular order. As a result, containers can be sorted at the time of delivery, thus reducing the effect of turning over. (4) Import containers by appointment. This method requires the consignee or the driver to make an appointment with the port terminal before picking up the goods at the terminal. Within a certain period of time, the port can limit the dock collection card to avoid the congestion caused by the collection card entering the wharf. At the same time, it can also obtain some collection card arrival situation in advance, so as to make reasonable arrangements for the gantry crane scheduling and unloading. In order to ensure that the operators arrive at the scheduled time, the port has also developed a set of reward and punishment mechanism to ensure that the operators can take out the containers at the scheduled time. In addition, in order to effectively improve the operation of the port terminal, we need to adopt modern information technology means to improve the port operation mode. Container turning operation is the most common problem encountered when containers are picked up, which is affected by three factors, namely, the storage status of containers in the stacking area, the order of container picking by the dispatcher and the storage location when unloading. The number of container overturning can be reduced by making a reasonable overturning strategy, optimizing the unloading location and adjusting properly according to the cargo position order. 4.2 Simulation Experiment of Container Rail-Water Intermodal Platform Container yard operation (see Fig. 1) has obvious randomness and dynamics, it belongs to a multi-level queuing complex system. Therefore, in this paper, the container port yard operating system model is constructed by simulation of the hot metal combined transport platform, reasonable analysis is made of the established simulation model, and feasible optimization suggestions are put forward.
Fig. 1. Plane diagram of container yard
The box-turning operation process was optimized through several operation experiments, and the optimal box-turning effect was finally achieved according to the
726
Q. Lin et al.
experimental results. On the premise of reducing the loss, the following optimization conclusions were drawn: (1) Box turning strategy. At present, the extensive use of the collection card reservation port mode and GPS technology enables the port to obtain the data of the collection card arrival, thus it saves a certain amount of time for the container turning rate and the best way of unloading. Proper placement of the inlet box can greatly reduce the second and more than two turnover rate. After determining the initial status of the stacking area, if the order is not changed, it is inevitable to turn over the box operation, but the number of turning over the box can be reduced as far as possible. (2) Adjust the container picking sequence. By adjusting the order of containers to match the storage location of containers as far as possible, it can reduce the turnover cost of containers and save the time of picking up containers. At the same time, it can also offset the waiting time of early arrival collection cards, but it cannot completely avoid the waiting time of collection cards at the port. Therefore, it is necessary to make a reasonable balance between the two to solve the problem of turning over boxes.
5 Countermeasures and Suggestions 5.1 Basic Idea of Yard Operation Optimization The scientific outlook on development should be thoroughly implemented, the concept of resource conservation should be combined with the basic national policy of responding to environmental protection, the policy of “two-oriented” port construction of the Ministry of Transport should be implemented into daily work, the interests of shipping companies, shippers and cargo terminals should be taken into account, and the breakthrough should be to reduce the frequency and rate of overturning containers. By integrating modern information technology, supported by scientific management methods, the container stacking strategy and turning strategy are constantly optimized, and measures such as differential service, coordination mechanism and performance appraisal are strengthened to improve economic benefits and service quality, reduce the operating cost of port containers, and gradually improve the efficiency of port operations. 5.2 Basic Principles for Optimization of Yard Operation There are four main principles for optimizing yard operation. (1) Global optimization. The yard operation will be carried out according to the shipping requirements of the water transport company and the requirements of the customer to pick up the container. Therefore, it is necessary to make overall planning for the interests of ports, water transport companies and customers. With import and export as the core, use the optimal method, adhere to the overall optimal concept and seek the lowest overall turnover rate under the constraint of the requirements of ports, water transport companies and customers.
Optimization Research of Port Yard Overturning Operation
727
(2) Standardized management. From the entry to the exit of the container, the process sequence chain is complex, involving many positions, and requiring high degree of cooperation between different positions. Therefore, the post requirements, operation contents, operation procedures and assessment indicators of field crane drivers, yard planners and central control dispatchers should be clearly defined, so as to efficiently guide each post to reduce the number and rate of overturning under the premise of ensuring efficiency. (3) Timely information. Obtaining accurate, comprehensive and detailed container information is the guarantee of fundamentally reducing the overturning rate. Therefore, coordination between relevant departments should be strengthened to ensure that necessary information can be obtained, updated and shared in a timely manner, avoid repeated information processing, and provide a reliable basis for scientific and reasonable management of port and wharf stacking. (4) Minimize the movement. When the overturning operation cannot be avoided, the reasonable overturning operation can be reduced and the overturning speed can be improved. The minimization of action can be achieved through the inner turning box, the nearest turning box, the lowest falling box and other methods. To sum up, this paper draws the port operation process, as shown in Fig. 2.
Fig. 2. Port operation flow
5.3 Reducing the Turning Rate There are six main ways to reduce the overturning speed. (1) Improve the system management rules. The shipping agent should consider the booked capacity and the actual carrying capacity of each ship in a balanced way
728
(2)
(3)
(4)
(5)
Q. Lin et al.
to ensure the balanced allocation of the space of the ships attached to the wharf, avoid the phenomenon of cabin bursting caused by blind acceptance of booking, and reduce the resulting overturning of the container; Secondly, the shipping agent should check the authenticity of shipper, cargo and other information when accepting booking, prevent false booking, strengthen the performance control after booking, and provide the wharf company with detailed freight information. Make stacking plans scientifically. The shipping company should strengthen the information communication with the affiliated port agent and the affiliated dock, improve the reliability of liner operation, try to follow the published shipping schedule for scheduled operation, minimize the disruption of the storage yard plan of the affiliated dock due to liner delay, and avoid unnecessary overturning. At the same time, the shipping company should also provide detailed information such as ship, container, arrival time, export date and subsequent docking port to the docking dock in a timely manner, and ensure the reliability and accuracy of the information, so that the dock can make reasonable yard plans with reliable and accurate information. Strengthen the connection with the front water transport chain. In shipping business, the booked capacity and the actual carrying capacity of each ship should be considered in a balanced manner to ensure the balanced allocation of the space of the ships attached to each port and dock, so as to avoid the phenomenon of “cabin bursting” caused by too much booking, reduce the resulting box turning. Strengthen the arrangement of container areas. After the container gathering port is finished and the stowing is completed, the yard controller shall use the idle time of the yard crane to timely issue the pre-turning order of the same container position, requiring the yard crane driver to arrange the export boxes to be loaded on the ship in the same container position in accordance with the order of loading, to ensure that the same container position will not be crushed, and in principle realize the zero turning of the container on the ship. Improve the comprehensive quality of relevant personnel. Firstly, improve the efficiency of yard control personnel communication and exchange, and constantly improve and enhance the function and performance of the port terminal operating system; The second is to fully consider the specific situation of the yard of the port terminal, as well as the loading efficiency and operating cost of the port terminal. With the cabin board as the boundary, reasonable stowing plan is prepared by region, and the operation route and loading sequence are optimized to ensure rationality and flexibility. Carry out stowing according to the order of discharging containers, try to match the same row of containers on the site together; Under the condition of confirming the shipping date and schedule, as far as possible, start stowing when the clearance condition is good; Only by understanding the true intentions of the stow crew can ship controllers better implement stow plans. At the beginning of loading, container collection, delivery and loading are carried out according to serial number. In fact, due to the different management modes of each port and wharf, accidents often occur in bridge crane, yard crane and collection card, etc., so ship control personnel should timely adjust their working ideas. Therefore, ship control personnel need to strengthen the communication with stowage personnel, flexible use of container interchange rules, to reduce unnecessary overturning operations.
Optimization Research of Port Yard Overturning Operation
729
(6) Reasonable implementation of box-turning operations. It is difficult for any port terminal to do zero turn over. When it is really necessary to turn over containers, yard controllers should rely on the function module of turning over optimization of port terminal operating system based on the information of export container ship name, loading date, unloading port, weight and so on, and aim at the minimum amount of turning over containers Through scientific calculation to determine the best order and stacking position of turning over boxes, to avoid the unreasonable operation caused by the double or serial turning over boxes. 5.4 Specific Methods to Reduce Overturning Speed The specific measures to reduce the overturning speed can be embodied in six aspects. (1) The Tabu algorithm is used to optimize the integrated scheme of storage position-set card scheduling. There are many operation targets and a large number of uncertain factors in container terminal yard. It is suggested to make a model plan before unloading the ship and work in order, so that the yard can classify the imported boxes within a limited time. The Tabu search algorithm can be used to pack the boxes. Under the premise that the stacking area is allowed, another Tabu search algorithm can be implemented to obtain the optimal scheduling scheme. Based on the calculation results of the model, container mixing can be reduced and overturning caused by rearrangement after unloading can be effectively reduced or avoided. (2) Design multiple preferential and penalty strategies to reduce container stop time. The dispatch should take the departure time of containers as the core to allocate the container space. First, pre-register the container departure date of the shipping company before the ship arrives at the port, so as to facilitate the dispatch to make the container space plan. Second, strengthen communication with shipping companies and carriers to obtain more reliable departure information. After unloading the imported containers, the dock shall promptly notify the consignee, freight station or carrier to pick up the containers as soon as possible to avoid long-term container stacking; The third is to formulate preferential policies to encourage off-peak pickup. It is suggested to offer convenience and preferential treatment to the pick-up customers during off-peak hours, and charge certain port clearing fees to the pickup customers during peak hours, so as to ease the traffic in the evacuation port area and avoid excessive mechanical load in the storage yard. Fourth, for the customers of free stacking period timeout, develop a step charging strategy, the longer the period, the higher the fee. Design multiple strategies to guide customers to pick up containers as early as possible, flexible compression container yard parking time, better improve the yard turnover efficiency. Practice has proved that after the Port of Los Angeles raised the stacking fees for overdue containers, the stacking time of containers in the yard was reduced by 20% on average, and the stacking capacity of the yard was significantly improved. (3) Use differential evolution algorithm to plan the differential stacking of containers. Depot planning directly affects assembly and shipment. Differential evolution algorithm can be used to guide the classification of customers. One is to open up exclusive container area for big customers of the wharf, and sign cooperation agreements with shipping companies or carriers; The second is to set up container groups for small
730
Q. Lin et al.
and medium-sized customers, and stack them in a centralized manner according to the number of the container groups. To guide customers to pick up the boxes without the designated container number and in accordance with the recommended operation order, on the one hand, it can alleviate the problems of time and cost, on the other hand, it can shorten the waiting time of picking up the boxes, so that customers have a better experience of picking up the boxes; Third, to prevent the overturning caused by mixing, it is recommended to separate the heavy containers stacked for a long time from the heavy containers stacked for a short time. This kind of containers should also be distinguished from other heavy containers and boxes to be checked after the inspection. For the cases that must be turned over, it is suggested to use differential evolution algorithm to determine the reasonable blocking position of the case from a macro perspective. (4) Promote the appointment of suitcases. Information technology makes it possible to make an appointment to pick up a suitcase. At the beginning, a series of preferential policies can be adopted to encourage large customers to use the appointment to pick up a suitcase service. When the customer gradually adapts, and then carry out heavy container appointment, which requires yard machinery to constantly improve the operating capacity. Therefore, the port should change its thinking, provide customers with suggestions on the pick-up time and order, constantly improve the booking pickup mode, and finally form the pick-up mode according to the suggested time period, to minimize or avoid “be on hand” and other uncertain factors. (5) The container management right shall be transferred from the shipping company to the unified operation of the terminal. At present, most container management rights belong to shipping companies. Dalian Port container Terminal obtains container management rights through reform and innovation, and the turnover rate can be basic Keep it at around 5%. The container management right of the terminal is conducive to the balance between the utilization rate of the discharging yard and the overturning rate. It can also improve the quality of the container and the flexibility of the unified deployment of the inbound container emphasis, so as to rationally allocate the yard resources and reduce the overturning volume. (6) Building a modern smart port by using the new generation of information technology. At present, there are some uncertain factors in the actual operation of Huangshi New Port, such as the arrival time of ships cannot be accurately predicted. With the continuous development of modern smart port equipment, more intelligent systems should be considered in the later stage to make up for this problem. The modern smart port uses the new generation of information technology to gradually integrate the port related business and management innovation deeply, make the port more intensive, efficient, convenient, safe and green, innovate the port development mode, and realize the scientific and sustainable development of the port.
Optimization Research of Port Yard Overturning Operation
731
6 Conclusion In this paper, the overturning operation of container yard is simulated and analyzed by the simulation technology of container Rail-water Intermodal, and a number of optimization methods are proposed: (1) Compared with the traditional method, this method analyzes the problems and causes of yard returning with the help of the simulation optimization model. The integrated model of storage position-set card scheduling is helpful to solve the optimization task allocation problem of ship unloading. (2) Design multiple preferential and penalty strategies to solve the problem of the time of compressed containers at the port. This method has good reusability and interoperability, which is conducive to promoting customer interaction and reducing the time of compressed containers, and can well meet the requirements of port managers. (3) Experiments show that the differential evolution algorithm can improve the solving accuracy of the simulation optimization model by planning the differential stacking of containers. At the same time, the classification of customers can be guided to further optimize the task allocation of each node, better improve the system management rules, facilitate the scientific development of stacking plans, strengthen the connection with the front-end water transport chain. (4) The container management right is transferred from the shipping company to the unified operation of the terminal. By strengthening the arrangement of container areas, improving the comprehensive quality of relevant personnel and reasonably implementing the overturning operation, the overturning rate of the container terminal yard can be effectively reduced, the utilization rate of port resources can be improved, the loading and unloading costs can be reduced, and the competitiveness of the port can be enhanced. The above suggestions put forward corresponding improvement suggestions for the current problem of the port turnover efficiency of Huangshi Newport. The next research should be based on the concept of smart port proposed by the Institute of Water Transport Science of the Ministry of Transport in 2022, and gradually use more intelligent systems and Internet of things technology to support the port information modernization. At present, the commercialization of 5G, the Internet of Things, big data and the development of artificial intelligence provide more and more underlying technical support for the automation and intelligence of ports. By September 2022, Jiangsu Province of China currently has the largest number of smart port patent applications in China, with a cumulative number of smart port patent applications reaching 930. Shanghai, Shandong and Guangdong have all applied for more than 500 smart port patents. Figure 3 shows the top ten provinces in the number of smart port patent applications in China, where Hubei is in the NO.8. On the basis of the underlying technology, AI identification technology identifies port gate and container code, etc., and relies on unmanned transportation technology and network layer to realize low-delay port communication and high-bandwidth video transmission. These technologies enable the linkage development of various sectors of smart ports. It is suggested that Huangshi Newport Dry Port should set up intelligent production management system, equipment control system and intelligent monitoring/remote
732
Q. Lin et al. 0
100
200
300
400
500
600
700
800
900
1000
Jiangshu Shanghai Shandong Guangdong Beijing Tianjing Zhejiang Hubei Liaoning Hebei
Fig. 3. TOP10 applications for smart port patents in provinces (municipalities and autonomous regions) in China by September 2022 (Unit: item) Data source: Intelligent Bud Foresight Industry Research Institute
control system of large equipment according to the actual development status and product and cargo characteristics, combine the frontier intelligent loading and unloading equipment, storage yard intelligent loading and unloading equipment and horizontal intelligent loading and unloading equipment of each port inside the port, and coordinate the subdivided shore bridge system, storage yard system and fleet management system. It will better realize the intelligent operation and operation of the whole process from the overall management, monitoring and overall control of the port, so as to improve the transportation safety of the terminal and improve the cargo efficiency of the terminal. Acknowledgment. This research is supported by Wuhan University of Business and Technology’s scientific research project: Modern Logistics Park Planning and Design Research (No. A2017005).
References 1. Kizilay, D., Eliiyi, D.T.: A comprehensive review of quay crane scheduling, yard operations and integrations thereof in container terminals. Flex. Serv. Manuf. J. 33(1), 1–42 (2021) 2. Yan, Y.: Development status and trend of container terminal operating system. Containerization 25(1), 19–23 (2014). (in Chinese) 3. Cai, P.: Review of the development of container transport in China’s ports in 2019 and outlook for 2020[J]. Containerization 31(2), 16–19 (2020). (in Chinese) 4. Huangshi New Port officially opened Phase I with an annual throughput of 10 million tons [EB/OL]. http://news.cnhubei.com/xw/hb/hs/201509/t3402605.shtml. Accessed 30 Sept 2015. (in Chinese) 5. Stojakovi´c, M., Twrdy, E.: Determining the optimal number of yard trucks in smaller container terminals. Eur. Transp. Res. Rev. 13(1), 1–12 (2021) 6. Ma, S.: Research on the problem of container turnover in container terminals. East China Normal University, Shanghai (2021). (in Chinese) 7. Zhang, Y., Han, H.: Reducing the turnover rate of container terminals. Containerization 9, 107–111 (2013). (in Chinese)
Optimization Research of Port Yard Overturning Operation
733
8. Yu, X., Ding, L.: Research on optimization of container turnover operation at the wharf. J. Logist. Sci. Technol. 12, 19 (2013). (in Chinese) 9. Zheng, S.: Research on the model and algorithm of container picking operation in container yard based on the optimization of container turnover path. South China University of Technology, Guangzhou (2018). (in Chinese) 10. Li, H.: Collaborative optimization model of import container picking order and turnover strategy based on partial container information. Dalian Maritime University, Dalian (2017). (in Chinese) 11. Otti, E.E., Okorie, E.C., Bulus, S.M.: Analysis and numerical simulation of deterministic mathematical model of pediculosis capitis. Int. J. Eng. Manuf. (IJEM) 13(1), 1–13 (2023) 12. Mei, S., Theo, N., Tao, Z., Xin, Z., Tianbao, Q.: Simulation model to determine ratios between quay, yard and intra-terminal transfer equipment in an integrated container handling system. J. Int. Logist. Trade 19(1), 2–19 (2021) 13. Luo, S., Zhan, Y., Guo, Y.: Research on the teaching and training methods of matrix theory in engineering practice. Inf. Commun. (12), 137–139 (2020). (in Chinese) 14. Zhang, Q., Yang, S., Zeng, Q., Yu, T.: Storage pricing model of container yards under fluctuating demand. Appl. Econ. 52(39), 4223–4235 (2020) 15. He, J., Xiao, X., Yu, H., et al.: Dynamic yard allocation for automated container terminal. Ann. Oper. Res. 1–22 (2022) 16. Yu, M., Zhang, Y.: Multi-agent-based fuzzy dispatching for trucks at container terminal. Int. J. Intell. Syst. Appl. (IJISA) 2(2), 41–47 (2010) 17. Zhang, Y., Tang, G., Yu, X., et al.: The impact of the length of container yard on the efficiency of terminal operation. Water Transp. Eng. 11, 94–98 (2016). (in Chinese) 18. Aveshgar, N., Huynh, N.: Integrated quay crane and yard truck scheduling for unloading inbound containers. Int. J. Prod. Econ. 159, 168–177 (2015) 19. Cahyono, R.T., Kenaka, S.P., Jayawardhana, B.: Simultaneous allocation and scheduling of quay cranes, yard cranes, and trucks in dynamical integrated container terminal operations. IEEE Trans. Intell. Transp. Syst. 23(7), 8564–8578 (2022)
A Review of Epidemic Prediction and Control from a POM Perspective Jing Wang1 , Yanbing Xiong1 , Qi Cai1 , Ying Wang2 , Lijing Du1,3,4(B) , and Kevin Xiong5 1 School of Safety Science and Emergency Management, Wuhan University of Technology,
Wuhan 430070, China [email protected] 2 School of Business, Wuchang University of Technology, Wuhan 430223, China 3 School of Management, Wuhan University of Technology, Wuhan 430072, China 4 Research Institute of Digital Governance and Management Decision Innovation, Wuhan University of Technology, Wuhan 430072, China 5 Information Technology Consulting Services, Ontario Limited, Markham, ON 1750351, Canada
Abstract. Infectious disease outbreaks have occurred many times in the past decades. They have had a tremendous impact on global health and the economy. Based on the learnings from these outbreaks, this study reviews the POM literature related to the epidemic and discusses what POM can help to address the challenges of the pandemic. This research divides the epidemic forecasting model into three types, namely compartmental model, statistical model and hybrid model. These models can be used to study the role of measures such as lockdowns and medical supplies in epidemic control. In addition, the pandemic has caused supply chain disruptions and led to a decline in the production level of manufacturers. It is important to ensure the supply of medical materials. Therefore, the relevant literature on the production of medical materials should be analyzed. Based on the comprehensive analysis of these articles, the future research directions are proposed. The results of this study are expected to be helpful for epidemic control research. Keywords: Epidemic simulation and prediction · Epidemiological model · Control measures · Medical materials production · POM model
1 Introduction Infectious disease outbreaks have been an immense challenge for humanity. The COVID19 broke out in December 2019 and spread rapidly across the globe within a few dozen days. It has become a crucial public health problem worldwide, disrupting the lives of tens of millions of people in many countries. Although the outbreak is now under preliminary control, it is still spreading globally and continues to produce new mutant strains with increasing transmission. Moreover, some people are infected more than once. The world is still facing a severe challenge. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 734–744, 2023. https://doi.org/10.1007/978-3-031-36115-9_65
A Review of Epidemic Prediction and Control
735
Almost all nations have imposed some kind of restrictions to some extent to control the spread of the COVID-19 [1]. Since the outbreak of the epidemic, many scientists have tried to make contributions to the prevention of the epidemic in various disciplines such as medicine, economics, sociology, etc. This paper analyzes the relevant literature on epidemic prevention and control from the perspective of Production and Operations Management (POM) to help mitigate the epidemic. The contributions of this review are as follows. First, in order to predict and control the development of the epidemic, scholars have established various models to simulate the evolution of the epidemic. In view of this, the research investigates the epidemiological models and classify them. Second, non-pharmacological interventions taken to mitigate the epidemic are discussed in this study. Third, the production of medical materials is an important factor in controlling the epidemic, and how to secure their production is also the concern of this paper. Finally, possible future research directions are suggested in this research. 1.1 Selection of Journals The search engine used for this study was the Web of Science. In the stage of article selection, articles using the following keywords in the topic of the articles: “pandemic”, “infection”, “epidemic”, “COVID-19”, “outbreak”, “infectious”, “influenza” or “coronavirus” were selected. The scope of the search was limited to major POM journals. In the initial screening process, 476 research papers were found. The title and abstract of each article were read carefully and irrelevant articles were excluded. Finally, 37 articles were obtained.
Fig. 1. Distribution of articles across journals
Figure 1 shows the distribution of published articles across 11 journals. It is observed that European Journal of Operational Research has published the highest number of research papers. Apart from European Journal of Operational Research, Production and Operations Management and Annals of Operations Research also represents most publications in this domain. These 23 papers of 37 papers, published in three journals,
736
J. Wang et al.
represent about 60% of all papers. Although many journals have published articles related to epidemics, their focus is completely different from epidemic control. They focus on finance, supply chain and some other issues. Therefore, only articles related to epidemic control are considered in this paper. 1.2 Research Trend The time period of articles selected for this research was from 2014 to 2023. Figure 2 shows the number of these articles based on the year of the publication. It can be seen from the figure that the curve of the number of papers from 2014 to 2019 is flat, with a sudden increase in 2020. The reason is the outbreak of the COVID-19 in 2020, which brought a huge shock to the world and posed a challenge to the research on epidemic control. The number of related articles continue to increase in 2021 and 2022. By January 8, 2023, the number of papers published has reached 8. It clearly indicates a growing interest of POM researchers in the prevention and control of the epidemic.
Fig. 2. Number of articles published between 2014 and 2023
2 Epidemic Simulation and Prediction Epidemic outbreak is a complex diffusion process occurring among people. Modeling this process provides a way to understand why and how infections spread and how they might be prevented. Researchers have developed a variety of models over many years to study the evolution of epidemic. In this study, these models are divided into three types, which are the compartmental model, the statistical model and the hybrid model. The application of these models in epidemic prediction is described below. 2.1 Compartmental Model The compartmental model is currently the most common infectious disease dynamics model. This model divides the population into compartments of susceptible (S), exposed
A Review of Epidemic Prediction and Control
737
(E), infected (I), treated (T), recovered (R), deceased (D), etc. according to epidemiological status, and describes the dynamics of continuity between compartments using ordinary differential equations. The compartmental model is more convenient compared to other models and can better fit the epidemiological trend of infectious diseases. Researchers use mathematical methods to study differential equations to simulate the epidemiological process of infectious diseases in different situations and predict the epidemic spread rate, mortality, cases of infection, etc. The classic SEIR model is mostly used to simulate the epidemic, and some scholars expand this model by considering hospitalized people, asymptomatic infected people and other groups to increase the accuracy of model simulation [1–5]. In addition, some scholars have considered the factors that affect the spread of the epidemic from multiple perspectives. For example, Liu and Zhang [6] predict the number of infections during the spread of an outbreak based on a time-discrete SEIR model to forecast the demand for healthcare resources. Kumar et al. [7] propose an extended SEIR model to study the measures to prevent the spread of infection and predict possible pandemic dynamics. Table 1 lists the articles mentioned above. Table 1. Characteristics of compartmental model Article
Model
Goal
Country
Lu and Borgonovo (2023)
Extended SEIR model
Modeling COVID-19 epidemic, sensitivity analysis of the model
Italy, USA
Perakis et al. (2022)
Multiwave SEIRD model
Forecasting COVID-19 cases
USA
Liu et al. (2021)
SEIDR model
Simulating the spread of COVID-19
USA, UK
Kumar et al. (2021)
Extended SEIR model
Forecasting possible pandemic dynamics
USA, seven European countries
Büyüktahtakn et al. (2018)
SITR-FB model
Simulating the Ebola disease transmission
West Africa
Liu and Zhang (2016)
time-discrete SEIR model
Predicting the trajectory of an epidemic diffusion
He and Liu (2015)
Modified SEIR model
Forecasting medical demand
China
2.2 Statistical Model Statistical model is generally driven by historical data. It grasps the key factors of epidemic transmission. Compared with the compartmental model, statistical model has fewer parameters and more accurately shows the changing patterns of the epidemic.
738
J. Wang et al.
Table 2 lists the statistical models used to forecast the epidemic spread. It can be seen from the table that the prediction models used by scholars include renewal equation approach, Stochastic Fractal Search algorithm, machine learning, etc., which are mainly used to forecast hospital admissions, ICU admission, mortality, newly infected people and other data [8–15]. These models take into account transmission networks and stochastic factors and allow for more detailed simulation of epidemic process. Table 2. Characteristics of statistical model Article
Model
Goal
Country
Chang et al. (2023)
Renewal equation approach
Modeling local coronavirus outbreaks
USA
Bekker et al. (2023)
Statistical model
Modeling COVID-19 hospital admissions and occupancy
Netherlands
Lotfi et al. (2022)
Regression-based robust optimization approach
Predicting COVID-19 epidemic
Iran
Saadatmand et al. (2022)
Machine learning
Predicting ICU admission, Iran mortality, and length of stay
Sbrana (2022)
Novel state-space approach
Predicting the number of newly infected people
USA
Khalilpourazari and Doulabi (2022)
Novel hybrid reinforcement learning-based algorithm
Predicting COVID-19 outbreak
Canada
Khalilpourazari and Doulabi (2021)
Stochastic Fractal Search algorithm
Predicting the number of Canada symptomatic, asymptomatic, life-threatening, recovered, and death cases
Liu et al. (2020)
Mixed-integer non-linear programming model
Controlling the H1N1 outbreak
China
2.3 Hybrid Model With the deepening of research, scholars use more complex models to simulate the epidemic. Nikolopoulos et al. [16] present statistical, epidemiological, machine- and deep-learning models, and a new hybrid forecasting method based on nearest neighbors and clustering to forecast COVID-19 growth rates in different countries. Evgeniou et al. [17] propose a hybrid model that combines the standard SEIR model with the equivalent standard machine learning classification model to simulate isolation and exit policies in France. Taylor et al. [18] formulate a compartmental and statistical model to empirically
A Review of Epidemic Prediction and Control
739
compare the interval and distributional prediction combination methods of cumulative mortality of COVID-19. Furthermore, some scholars combine these models with logistics models to predict the material demand and use logistics models to distribute these materials to control the epidemic. According to the above research, these models have different characteristics and scope of application. Hybrid model can study more complex problems and predict the spread of the epidemic more accurately.
3 Epidemic Control Measures Epidemic prevention and control measures include pharmacological interventions and non-pharmacological interventions. In the field of POM, pharmacological interventions mainly refer to vaccination, while non-pharmacological interventions include social distance, stay-at-home orders, lockdowns, etc. [19–28]. This research has investigated and analyzed the literature related to these control measures to understand the main methods of epidemic control. Table 3 shows these control measures and lists the models used in these papers. It can be seen from the table that medical materials such as ventilators and nucleic acid tests are effective for epidemic control [19, 20, 24, 29]. But how to ensure the supply of medical materials is a problem that needs to be studied. Although intervention measures can effectively contain the spread of infectious diseases, they impose substantial direct and indirect costs on societies [30]. Therefore, some scholars study the relationship between intervention measures and economy to reduce the economic losses caused by them. Birge et al. [31] propose a spatial epidemic model illustrating population mobility to restrict economic activity in different neighborhoods of a city at different levels. Their study indicates the potential to limit the economic costs of unemployment while containing the spread of a pandemic. Eryarsoy et al. [30] formulate a mathematical model to manage the lost lives during an epidemic through controlling intervention levels. Their findings demonstrate that when the projected economic costs of the epidemic are large and the illness severity is low, a no-intervention strategy may be preferable. Chen et al. [32] provide an efficient social distancing policy that minimizes the total risks of disease transmission and economic volatility. Table 3. Epidemic control measures Article
Control Measures
Model
Country
Abdin et al. (2023)
Testing capacities, control strategies
Epidemiological compartmental model
France
Rezapour et al. (2023)
Movement restrictions, social distancing, proactive testing
Multi-scale reaction-diffusion model
USA
(continued)
740
J. Wang et al. Table 3. (continued)
Article
Control Measures
Model
Country
Hosseini et al. (2023)
Behavioral changes in Statistical model the population, vaccination
Iran
Wang (2022)
Stay-at-home orders
Statistical model
USA
Li et al. (2022)
Restriction on mass gatherings, closure of schools, stay-at-home
DELPHI model
USA, UK, Russia
Chen et al. (2022)
Medical resources, social distancing policy
Modified SEIR model
China
Ertem et al. (2022)
Early social distancing Age-structured measures compartmental simulation model
USA
Biswas et al. (2022)
Lockdowns, curfews
Mixed Integer Non-Linear Programming epidemic model
France
Baveja et al. (2020)
Travel restrictions
Pandemic-management service value chain
Kumar et al. (2021)
Social media
SEIR-V model
Mehrotra et al. (2020)
Ventilators
stochastic optimization model
USA
4 Medical Materials Production in an Epidemic Medical materials play an important role in controlling the epidemic. However, due to the rapid outbreak of the epidemic, the demand for medical supplies has skyrocketed. It is critical to produce sufficient medical supplies to meet the demand in a timely manner. Many scholars have studied the production of medical materials in this special situation. Li et al. [33] formulate a risk-averse two-stage stochastic programming model to study the comprehensive production planning problem under uncertain demand and explore the impact of different risk preferences on the production of three types of masks. Sun et al. [34] analyze the construction of vaccine at-risk production capacity under the integrated and outsourcing mode and make recommendations for improving at-risk production capacity under both modes. Angelus et al. [35] discuss the large-scale production of new vaccines under two yield improvement strategies. Table 4 lists the articles on the production of medical materials under the epidemic. Current research mainly focuses on vaccine production. Some scholars have conducted the incentive program for vaccine manufacturers, some have studied the at-risk production capacity building of vaccines, and some have investigated the large-scale production of new vaccines [34–38]. However, there are few studies on the production of medical
A Review of Epidemic Prediction and Control
741
materials other than vaccines. Only a few scholars have studied the production of masks and health equipment [33, 39]. Table 4. Literature reviews of medical materials production Article
Type of Material
Model
Outbreak
Li et al. (2023)
Mask
Risk-averse two-stage stochastic programming model
COVID-19
Sun et al. (2022)
Vaccine
Signaling game model, incomplete contract model
COVID-19
Angelus et al. (2022)
Vaccine
Stochastic, multiperiod, sequential-decision model
COVID-19
Soltanisehat et al. (2022)
health equipment
Mixed-integer linear programming model, Monte Carlo simulation
Epidemic
Arifoglu & Tang (2022)
Vaccine
Backward induction
Influenza
Jansen & Ozaltin (2017)
Vaccine
Cournot competition model
Influenza
Chick et al. (2017)
Vaccine
Public procurement model
Influenza
5 Conclusions and Future Research Direction Based on the review of articles related to epidemic control in the past decade, this study mainly considers the review of epidemic simulation and prediction, epidemic prevention and control measures and medical material production. The models used for epidemic simulation and prediction are divided into three types, namely compartmental model, statistical model and hybrid model. Compartmental model is the main model for exploring the epidemic evolution at present. With the deepening of research, the use of statistical model and hybrid model is increasing. The epidemic prevention and control measures mainly include government measures such as social distancing policy, lockdown and timely supply of medical materials. As for the production of medical materials, the current research mainly focuses on vaccine production. There is little research related to medical materials such as protective clothing and ventilators. Based on our findings, the following recommendations are made for future research. First, the epidemic is now frequent, and predicting of the epidemic wave is a major challenge. Researchers should explore how to accurately predict the next possible pandemic and propose countermeasures. Second, control measures can be effective in reducing the spread of epidemic, but they can also lead to problems such as economic stagnation and social unrest. Researchers could explore how to balance the relationship between the two. Third, the supply of medical materials plays a key role in epidemic prevention and control, but currently relevant research mainly focuses on vaccine production. Future
742
J. Wang et al.
research can focus more on the production of other medical materials. Future research is considered to be combined with other disciplines to investigate epidemic control from multiple perspectives compared to just the POM perspective. Acknowledgment. This work was supported by the National Natural Science Foundation of China (72104190); and the Humanities and Social Sciences Fund of Ministry of Education (20YJC630018).
References 1. Liu, Y., Srivastava, S., Huang, Z., et al.: Pandemic model with data-driven phase detection, a study using COVID-19 data. J. Oper. Res. Soc. (2021). https://doi.org/10.1080/01605682. 2021.1982652 2. Lu, X., Borgonovo, E.: Global sensitivity analysis in epidemiological modeling global sensitivity analysis in epidemiological modeling. Eur. J. Oper. Res. 304(1), 9–24 (2023) 3. Perakis, G., Singhvi, D., Lami, O.S., et al.: COVID-19: a multipeak SIR-based model for learning waves. Product. Oper. Manag. 32, 13681 (2022). https://doi.org/10.1111/poms.13681 4. Büyüktahtakın, I.E., des-Bordes, E., Kıbı¸s, E.Y.: A new epidemics–logistics model: insights into controlling the Ebola virus disease in West Africa. Eur. J. Oper. Res. 265(3), 1046–1063 (2018) 5. He, Y., Liu, N.: Methodology of emergency medical logistics for public health emergencies. Transport. Res. E-Log 79, 178–200 (2015) 6. Liu, M., Zhang, Ding.: A dynamic logistics model for medical resources allocation in an epidemic control with demand forecast updating. J. Oper. Res. Soc. 67(6), 841–852 (2016) 7. Kumar, A., Choi, T.-M., Wamba, S.F., Gupta, S., Tan, K.H.: Infection vulnerability stratification risk modelling of COVID-19 data: a deterministic SEIR epidemic model analysis. Ann. Oper. Res. (2021). https://doi.org/10.1007/s10479-021-04091-3 8. Chang J.T., Kaplan, E.H.: Modeling local coronavirus outbreaks. Eur. J. Oper. Res. 304(1), 57–68 (2023) 9. Bekker, R., Broek, M., Koole, G.: Modeling COVID-19 hospital admissions and occupancy in the Netherlands. Eur. J. Oper. Res. 304(1), 207–218 (2023) 10. Lotfi, R., Kheiri, K., Sadeghi, A., et al.: An extended robust mathematical model to project the course of COVID-19 epidemic in Iran. Ann. Oper. Res. (2022). https://doi.org/10.1007/ s10479-021-04490-6 11. Saadatmand, S., Salimifard, K., Mohammadi, R., et al.: Using machine learning in prediction of ICU admission, mortality, and length of stay in the early stage of admission of COVID-19 patients. Ann. Oper. Res. (2022). https://doi.org/10.1007/s10479-022-04984-x 12. Sbrana, G.: Modelling intermittent time series and forecasting COVID-19 spread in the USA. J. Oper. Res. Soc. (2022). https://doi.org/10.1080/01605682.2022.2055499 13. Khalilpourazari, S., Doulabi, H.H.: Designing a hybrid reinforcement learning based algorithm with application in prediction of the COVID-19 pandemic in Quebec. Ann. Oper. Res. 312(2), 1261–1305 (2022) 14. Khalilpourazari, S., Doulabi, H.H.: Robust modelling and prediction of the COVID-19 pandemic in Canada. Int. J. Prod. Res. (2021). https://doi.org/10.1080/00207543.2021.193 6261 15. Liu, M., Xu, X., Cao, J., et al.: Integrated planning for public health emergencies: a modified model for controlling H1N1 pandemic. J. Oper. Res. Soc. 71(5), 748–761 (2020)
A Review of Epidemic Prediction and Control
743
16. Nikolopoulos, K., Punia, S., Schafers, A., et al.: Forecasting and planning during a pandemic: COVID-19 growth rates, supply chain disruptions, and governmental decisions. Eur. J. Oper. Res. 290(1), 99–115 (2021) 17. Evgeniou, T., Fekom, M., Ovchinnikov, A., et al.: Pandemic lockdown, isolation, and exit policies based on machine learning predictions. Prod. Oper. Manag. (2022). https://doi.org/ 10.1111/poms.13726 18. Taylor, J.W., Taylor, K.S.: Combining probabilistic forecasts of COVID-19 mortality in the United States. Eur. J. Oper. Res. 304(1), 25–41 (2023) 19. Abdin, A.F., Fang, Y.-P., Caunhye, A., et al.: An optimization model for planning testing and control strategies to limit the spread of a pandemic-The case of COVID-19. Eur. J. Oper. Res. 304(1), 308–324 (2023) 20. Rezapour, S., Baghaian, A., Naderi, N., et al.: Infection transmission and prevention in metropolises with heterogeneous and dynamic populations. Eur. J. Oper. Res. 304(1), 113–138 (2023) 21. Hosseini-Motlagh, S.-M., Samani, M.R.G., Homaei, S.: Design of control strategies to help prevent the spread of COVID-19 pandemic. Eur. J. Oper. Res. 304(1), 219–238 (2023) 22. Wang, G.: Stay at home to stay safe: effectiveness of stay-at-home orders in containing the COVID-19 pandemic. Prod. Oper. Manag. 31(5), 2289–2305 (2022) 23. Li, M.L., Bouardi, H.T., Lami, O.S., et al.: Forecasting COVID-19 and analyzing the effect of government interventions. Oper. Res. (2022). https://doi.org/10.1287/opre.2022.2306 24. Chen, Z., Kong, G.: Hospital admission, facility-based isolation, and social distancing: an SEIR model with constrained medical resources. Prod. Oper. Manag. (2022). https://doi.org/ 10.1111/poms.13702 25. Ertem, Z., Araz, O.M., Cruz-Aponte, M.: A decision analytic approach for social distancing policies during early stages of COVID-19 pandemic. Decis. Support Syst. 161, 113630 (2022) 26. Biswas, D., Alfandari, L.: Designing an optimal sequence of non-pharmaceutical interventions for controlling COVID-19. Eur. J. Oper. Res. 303(3), 1372–1391 (2022) 27. Baveja, A., Kapoor, A., Melamed, B.: Stopping Covid-19: a pandemic-management service value chain approach. Ann. Oper. Res. 289(2), 173–184 (2020) 28. Kumar, S., Xu, C., Ghildayal, N., et al.: Social media effectiveness as a humanitarian response to mitigate influenza epidemic and COVID-19 pandemic. Ann. Oper. Res. 319, 823–851 (2021) 29. Mehrotra, S., Rahimian, H., Barah, M., et al.: A model of supply-chain decisions for resource sharing with an application to ventilator allocation to combat COVID-19. Nav. Res. Logist. 67(5), 303–320 (2020) 30. Eryarsoy, E., Shahmanzari, M., Tanrisever, F.: Models for government intervention during a pandemic. Eur. J. Oper. Res. 304(1), 69–83 (2023) 31. Birge, J.R., Candogan, O., Feng, Y.: Controlling epidemic spread: reducing economic losses with targeted closures. Manage. Sci. 68(5), 3175–3195 (2022) 32. Chen, K., Pun, C.S., Wong, H.Y.: Efficient social distancing during the COVID-19 pandemic: integrating economic and public health considerations. Eur. J. Oper. Res. 304(1), 84–98 (2023) 33. Li, Y., Saldanha-da-Gama, F., Liu, M., et al.: A risk-averse two-stage stochastic programming model for a joint multi-item capacitated line balancing and lot-sizing problem. Eur. J. Oper. Res. 304(1), 353–365 (2023) 34. Sun, H., Toyasaki, F., Sigala, I.F.: Incentivizing at-risk production capacity building for COVID-19 vaccines. Prod. Oper. Manage. 32(5), 1550–1566 (2022). https://doi.org/10.1111/ poms.13652 35. Angelus, A., Ozer, O.: On the large-scale production of a new vaccine. Prod. Oper. Manag. 31(7), 3043–3060 (2022) 36. Arifoglu, K., Tang, C.S.: A two-sided incentive program for coordinating the influenza vaccine supply chain. M&Som Manuf. Serv. Oper. Manage. 24(1), 235–255 (2022)
744
J. Wang et al.
37. Jansen, M.C., Ozaltin, O.Y.: Note on cournot competition under yield uncertainty. M&Som Manuf. Serv. Oper. Manage. 19(2), 305–308 (2017) 38. Chick, S.E., Hasija, S., Nasiry, J.: Information elicitation and influenza vaccine production. Oper. Res. 65(1), 75–96 (2017) 39. Soltanisehat, L., Ghorbani-Renani, N., Gonzalez, A.D., et al.: Assessing production fulfillment time risk: application to pandemic-related health equipment. Int. J. Prod. Res. (2022). https://doi.org/10.1080/00207543.2022.2036381
Application of SVM and BP Neural Network Classification in Capability Evaluation of Cross-border Supply Chain Cooperative Suppliers Lei Zhang1(B) and Jintian Tian2 1 School of Logistics, Wuhan Technology and Business University, Wuhan 430065, China
[email protected] 2 School of Postgraduate Studies, PSB Academy, Singapore 039594, Singapore
Abstract. With the trade-driven expansion of the Belt and Road Initiative, overseas engineering projects are increasing. The success of overseas engineering projects is marked by the progress and benefit, and to achieve this goal, the supplier’s ability of the supply chain supporting the project is the key to the progress and benefit of the project. On the basis of literature analysis and many years of foreign engineering practice, combined with China’s reality, this paper puts forward the subcontractor competency evaluation index. That is, the initial index set of supplier competency evaluation system is constructed based on 6 main factors and 24 sub-factors including supplier credit strength, quality assurance, supply ability, service level, technical level and price level. On the basis of careful analysis of qualitative and quantitative factors, support vector machine and BP neural network method are applied to compare, analyze and evaluate the capability of engineering project suppliers, and explore the effective solution of quantitative and qualitative coordination correlation between evaluation indicators. Keywords: BP neural network · Support vector machine (SVM) · Competency
1 Introduction Since the Belt and Road Initiative was put forward in 2013, its goal is to establish a free and open economic system for countries along the Belt and Road through a reasonable mix of production factors to make them flow smoothly, and promote the common development of countries and regions along the Belt and Road through economic means. The engineering projects in the countries along the Belt and Road are the cornerstone of the complete chain of the Belt and Road, and the safe and smooth flow of the material supply chain is the guarantee of the smooth progress of the engineering projects. The competence of the cooperative supplier that undertakes the supply chain is the guarantee of the safety and smooth of the supply chain. Therefore, the competency evaluation of cross-border suppliers becomes more and more important. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 745–753, 2023. https://doi.org/10.1007/978-3-031-36115-9_66
746
L. Zhang and J. Tian
Competency theory originated from McClelland’s landmark paper Measuring Competency, not Intelligence, published in 1973 [1]. Since then, this theory has been gradually developed and improved. In 1980, McClelland proposed the concept of “competency model” and defined it as “the key abilities needed to complete the work”, which included a series of knowledge, skills and attitudes. In 1982, the book Competent Manager published by Rechard E Boyatzis regarded competency as a stable internal characteristic of a person, which is knowledge, cognition, behavior, skill, motivation, trait, etc. [2]. At present, in western countries, competency research results have been widely applied to recruitment, training, selection, performance management and other human resource management practices. Based on the strategic significance of organizational development, a strong human resource system has been established, so as to comprehensively improve the competitiveness of the organization. The manager’s competency model has been widely applied in major organizations. Chinese research on manager competency started relatively late, but some important achievements have been made in recent years. For example, Gu Qinxuan explored the effectiveness of shared leadership from the perspective of shared leadership and participation in a safe atmosphere [3]. Qi Yue developed the competency model of middle managers in banks [4]. Feng Xufang built a competency structure model based on the practical teaching of teachers in higher vocational colleges [5]. Wang Haixia and Tang Zhisong studied the competence of teachers’ core literacy education [6]. Tan Qiwei analyzed the construction method and application of the competency model [7]. Wu Qianying established a competency model for optimizing the civil servant training system [8]. Zhang Qianxue established a competency model based on human resource performance management [9]. Kang Fei and Zhang Shuibo used text analysis to analyze the research status of project manager competency at home and abroad [10]. Zhang Shuibo et al. constructed the competency evaluation index system of construction project managers and established the evaluation model [11]. He Qizong analyzed the research progress of Chinese college teachers’ competence [12]. Hou Yanhui et al. conducted a competency study on multi-project human resource allocation of engineering construction enterprises [13]. Zhao Berg analyzed the competency structure of management teachers in ethnic universities based on grounded theory [14]. Chen Zhixia and Guo Jinyuan established the structural model of graduate competency and predicted it [15]. Yang Yuekun and Lu Nan built a “trinity” evaluation system for innovative scientific and technological talents based on exploratory and confirmatory factor analysis [16]. Although domestic and foreign scholars have conducted a lot of research on the competency of managers, the research on the competency of cooperative suppliers has not received extensive attention. At the same time, there is a lack of research results on how to apply the competency model to evaluate the competency of cooperative suppliers. Based on literature analysis and expert opinions, this paper analyzes the feature vector and structural dimension of supplier capability, and establishes an evaluation system based on project cooperative supplier capability.This is a complex and multi-factor issue, and it needs to screen out the important influencing factors among the many influencing factors as the evaluation criteria. Some of these factors can be quantitatively analyzed, such as using data envelopment analysis to evaluate; Many other factors, however, are difficult to be evaluated by quantitative analysis, and can only be evaluated by qualitative
Application of SVM and BP Neural Network Classification
747
analysis, among which analytic hierarchy process (AHP) is a more commonly used method. This paper aims to establish the cross-border supplier capability evaluation system, and based on the theory of BP neural network and support vector machine, to evaluate the cross-border supplier capability evaluation.
2 Introduction to Evaluation Model 2.1 BP Neural Network Model In 1986, Rumelhard and Mc Clelland proposed BP neural network, which has one input layer, one output layer and several hidden layers. It is a typical multi-layer forward type, and the neurons in the same layer are not connected, and the mode between layers is fully connected [17]. Neural network is based on the learning and memory function of network, which enables the neural network to learn the characteristics of various categories of sample information, to treat the identification samples by comparing the characteristic information mastered with the input vector, and finally to determine the categories of samples to be identified. Therefore, it obtains decision information through the learning mechanism, which is an abstract, simulation and simplified human brain information processing model. The key is the mathematical model of neuron, the way of network connection and the way of neural network learning. BP neural network is a widely used neural network model. BP neural network has two typical learning processes of forward propagation and back propagation. In the forward propagation process, the input information is transmitted through the hidden layer to the output layer to generate the output signal, during which the connection weight of neurons is constant. By comparing the expected value with the output result, if the error between the expected value and the output result does not meet the accuracy requirements, then the error propagates backward from the output layer will constantly adjust the threshold and weight, so that the gap between the expected value and the output result is closer and closer until the accuracy requirements are met. BP artificial neural network has been widely used in pattern recognition and regression fitting, because it can select the appropriate number of hidden elements and network hierarchy, multi-precision approximation of nonlinear functions. However, BP artificial neural network has some shortcomings. One is that the network structure needs to be specified in advance or modified by heuristic algorithm in the training process, so it is difficult to ensure the optimization of the network structure. Secondly, the method of adjusting the network weight coefficient is very limited, which is easy to get the local optimal, but difficult to get the optimal solution. Thirdly, sample data is too dominant to model performance in the training process. Fourthly, it is difficult to obtain the final result due to the large amount of high quality training data and the high dimensional input in many practical problems. Lots of engineering problems are faced with similar situations [18, 19]. Engineering practice shows that the application of neural network model is an effective way.
748
L. Zhang and J. Tian
2.2 Support Vector Machines Model Support vector machine (SVM) theory is a new machine learning method based on statistical learning theory. It has many unique advantages in solving small sample, nonlinear and high-dimensional pattern recognition problems. It has strong learning ability and good generalization performance. It has unique advantages in pattern recognition, function approximation and probability density estimation [20]. The core of support vector machine (SVM) is to construct a decision surface so that the sample points in the training set are as far away from the segmentation plane as possible and the blank areas on both sides of the segmentation plane can be maximized. If the optimal classification surface with satisfactory classification effect cannot be obtained in the original space, it needs to be transformed into a linear problem in high dimensional space through nonlinear transformation, and solved in the transformation space to obtain the optimal classification surface [21]. Support vector Machine (SVM) theory is a new general learning method developed on the basis of statistical learning theory, which has been applied in many fields and achieved good results [22–24]. The function of support vector is effectively realized by the accurate fitting of high-dimensional nonlinear system with small samples, and has good generalization performance based on the principle of minimizing structural risk. The regression function of support vector machine can be expressed as: yi [(w · xi ) + b] − 1 ≥ 0 i = 1, . . . n
(1)
If Mercer can be satisfied, the objective function can be converted into a highdimensional spatial function: Q(α) =
n
αi −
i=1
n 1 αi αj yi yj K xi , xj 2
(2)
i,j=1
The corresponding function is:
∗
∗
f (x) = sgn w · x + b
= sgn
n
αi∗
yi K xi , xj + b∗
(3)
i=1
3 Establishment of Supplier Competency Evaluation System According to the principle that the selection of evaluation indexes should be scientific, representative, objective and systematic, CNKI database and Wanfang database were searched to select the indexes with high frequency, and the evaluation indexes in domestic and foreign literatures were also used for reference. Six main level factors and 24 sub-level factors, including credit and strength, quality assurance, supply ability, service level, technical level and price level, are extracted as indicators of supplier competence, and the initial index set of the competency evaluation system of cooperative suppliers is constructed, as shown in Table 1.
Application of SVM and BP Neural Network Classification
749
Table 1. The initial index set of the competency evaluation system of cooperative suppliers Target layer
Layer of criterion
Indicator layer
The serial number
Supplier competency system
Credit and strength
Contract performance capability
C1
Management system
C2
Financial position
C3
Reputation Brand
C4
Level of sales
C5
Quality assurance
Ability to supply
Quality of personnel
C6
Social responsibility
C7
Quality control system
C8
Quality certification system
C9
Production quality inspection
C10
Ex factory quality control
C11
Product performance level
C12
Timeliness of delivery
C13
Accuracy of delivery point
C14
Completeness of goods C15 Level of service
Level of technology
Level of price
Attitude towards Service
C16
Speed of response
C17
Effect of customer demand
C18
Production technology level
C19
Production technology level
C20
Level of inspection technology
C21
Level of information
C22
Rationality of price
C23
Price control
C24
750
L. Zhang and J. Tian
4 Classification Experiment Design 4.1 Experimental Design of BP Neural Network Classification To determine the transfer function, the number of hidden layer neurons, the number of hidden layers, etc., is the premise of network design. On this basis, the influence of momentum factor and random number selection in stochastic gradient descent on classification performance is studied respectively. Performance evaluation index: The algorithm performance was investigated from three aspects: classification accuracy, convergence time and curve characteristics. High accuracy is the primary goal of the algorithm, while stochastic gradient descent method and additional momentum factor are used to reduce running time and improve learning efficiency, so convergence time and curve features are selected as evaluation items. 4.2 Experimental Design of SVM Classification The parameters of SVM algorithm are less than that of BP algorithm. The influence of different kernel functions and sample number on classification performance is mainly investigated, and the performance of BP algorithm is combined with the number of training samples. There are four common kernel functions: (1) Gaussian radial basis kernel function 2 (4) K(xi , xj ) = exp(−γ xi − xj ) (2) Polynomial kernel function K(xi , xj ) =
d xi · xj + 1
(5)
(3) Linear kernel function K xi , xj = xi · xj (4) One-dimensional Fourier kernel function
2 K(xi , xj ) = 1 − q 21 − 2q cosxi − xj + q2
(6)
(7)
Radial basis kernel function has a wide convergence domain in these four kinds of kernel functions, and contains only one parameter which is easy to optimize, so it is the most widely used kernel function. 4.3 Experimental Analysis 4.3.1 Competency Evaluation Data Preparation Samples can be used as examples for learning and training. The selection of samples is crucial to the competency evaluation results, so it should be comprehensive and typical. 160 units were selected as training samples.
Application of SVM and BP Neural Network Classification
751
In order to facilitate the comparison between BP neural network and SVM model and make them comparable, it is necessary to maintain the consistency of learning samples. Here, MATLAB neural network toolkit and SVM algorithm software package are used respectively to train and evaluate the performance of the samples. In this paper, 120 samples were randomly selected as test data and 40 samples were selected as verification data in the training samples. 4.3.2 BP Neural Network Evaluation Model BP artificial neural network selection function, transfer function and transfer function of hidden layer and output layer training function, weight threshold learning function respectively use logsig, tansig, traingdx, leard to calculation. 4.3.3 SVM Evaluation Model The kernel function (RBF) commonly used in SVM has few parameters and good fitting effect. As a basic function, SVM environmental assessment model is established. After 120 test data are inputted, automatic optimal parameter selection is realized by using total error minimum programming. The model parameters were selected by cross validation method and programmed with Lib SVM library file. 4.3.4 Model Comparison The 40 verification data were substituted into the model established above, and the unit environmental assessment score was output. The predicted results are compared with those of BP artificial neural network and support vector machine, most of the absolute errors of the two models can meet the accuracy requirements of competency evaluation. In the construction of BP artificial neural network and support vector machine, although the number of hidden layer nodes determined by the former model can be adjusted, but the optimal parameters need to be tested manually repeatedly, which is a large degree of complexity. In the SVM model, the method of cross validation is adopted to realize the automatic optimization of parameter selection, which reduces the human intervention and greatly reduces the total error. Compared with the two models of BP artificial neural network and support vector machine, the absolute error curve, MSE, ME and MAPE of support vector machine are all lower than that of BP neural network, and the convergence speed of support vector machine is faster, which requires less time. Overall, therefore, support vector machine model is better than BP artificial neural network model. When the number of random samples is large enough to reflect the basic law of samples, it can effectively reduce the running time. Too much will increase the calculation amount, while too little will reduce the prediction accuracy. The accuracy of BP artificial neural network and support vector machine is not strictly correlated with the number of samples, but enough samples can improve the accuracy and generalization ability of BP artificial neural network. For support vector machine, when the number of samples is insufficient, the generalization ability of support vector machine is more powerful than that of BP artificial neural network.
752
L. Zhang and J. Tian
5 Conclusions Based on the establishment of BP neural network and SVM model respectively, and applied to the actual situation of competency, the conclusions can be drawn as follows: (1) In order to ensure the comparability of the model, 160 learning samples were selected, 24 evaluation indexes were taken as input vectors, and evaluation scores were taken as output vectors. The data results showed that the accuracy of competency evaluation could meet the requirements. (2) It can be seen from the comparison that, compared with the traditional BP neural network model, the support vector machine model can realize the automatic optimization of parameters, avoid the difficult to determine the hidden layer neurons of BP neural network, reduce the influence of human factors in the evaluation process, and convergence speed is faster and less time consuming. (3) There are many factors affecting competency. Compared with analytic hierarchy process (AHP) or fuzzy comprehensive evaluation, both BP neural network model and SVM model reduce the influence of subjective factors to a certain extent. The reasonable selection of the number of training samples determines the final accuracy of BP neural network model and SVM model. In short, compared with other evaluation methods, the BP neural network model and SVM model adopted in this paper for competency evaluation can greatly reduce the influence of human factors and avoid the factor problem of artificially determining index weights in the traditional qualitative evaluation methods, so that the data processing is more in line with the actual situation. Acknowledgment. This paper is supported by: (1) Hubei Business Service Development Research Center Fund Project: Hubei Intelligent Logistics Development Support System Research (2020Z01). (2) Doctoral Research Foundation of Wuhan Institute of Industry and Technology: Analysis and Countermeasures of Cross-border Logistics Security of Petroleum projects in China and Kazakhstan based on “Belt and Road Initiative” (D2019004).
References 1. McClelland, D.C.: Testing for competence rather than for intelligence. Am. Psychol. 28(1), 1–24 (1973) 2. Boyatzis, R.E.: The Competent Manager: A Model for Effective Performance. John Wiley & Sons Inc. (1982) 3. Gu, Q.: A Study on the effectiveness of shared leadership competency from the perspective of shared leadership and participation in a safe atmosphere. J. Manag. 12, 52–60 (2020) 4. Qi, Y.: Construction and application of financial manager competency model under the background of new regulations on capital management - a case study of P Bank. Reform Opening Up 15, 14–22+28 (2020) 5. Feng, X.: Construction of practical teaching competence structural model of higher vocational college teachers. Vocat. Educ. Forum 1, 107–114 (2021). (in Chinese) 6. Wang, H., Tang, Z.: Research on competence of teachers’ core literacy education. Curriculum. The textbook. Teaching Methods 2, 132–138 (2020)
Application of SVM and BP Neural Network Classification
753
7. Tan, Q.: Analysis on the construction method and application of competency model. Mod. Hosp. 12, 46–48 (2018) 8. Wu, Q.: Optimization of civil servant training system based on competency model. Mod. Mark. (Manage. Ed.) 1, 16–17 (2020) 9. Zhang, Q.: Construction of human resource performance management system: from the perspective of competency model. Chin. Collect. Econ. 1, 122–123 (2019). (in Chinese) 10. Kang, F., Zhang, S.: Research on project manager competency: current situation and prospect. J. Tianjin Univ. (Social Science Edition) 15(1), 35–40 (2013). (in Chinese) 11. Zhang, S., Kang, F., Li, X.: Competency evaluation of construction project manager based on support vector machine. China Soft Sci. 11, 83–90 (2013). (in Chinese) 12. He, Q.: Research on teacher competency in Chinese universities: progress and thinking. Res. High. Educ. 35(10), 39–45 (2014). (in Chinese) 13. Hou, Y., Rao, W., Hao, M.: Research on multi-project human resource allocation in engineering construction enterprises: based on competency and two-stage optimization perspective. Oper. Res. Manage. 26(4), 192–199 (2017) 14. Zhao, B.: A Study on the competence structure of management teachers in ethnic colleges and universities: based on grounded theory. Res. Ethnic Educ. 20 31(6), 152–156 15. Chen, Z., Guo, J.: The construction of graduate competency structure model and its predictive effect. Degrees Grad. Educ. 7, 55–60 (2018). (in Chinese) 16. Yang, Y., Lu, N.: Construction of evaluation model of innovative scientific and technological talents based on knowledge value. Leadersh. Sci. 1, 98–102 (2019) 17. Wang, L., Kuang, Y.: Research of BP neural network optimizing method based on ant colony algorithm. J. New Industrial. 2(4), 8–15 (2012) 18. Zhang, H., Xiaobo, Z.: Quantitative analysis of organizational behavior of container shipping in the upper and middle reaches of the Yangtze river based on hub-and-spoke network. J. Coast. Res. 73, 119–125 (2015) 19. He, Q.-M., Zhang, H., Ye, Q.: An M/PH/K Queue with constant impatient time. Math. Methods Oper. Res. 18(11), 139–168 (2018) 20. Zhang, X.: On statistical learning theory and support vector machines. Acta Automat. Sin. 26(1), 32–42 (2000) 21. Zhang, L.: Evaluation of China’s petroleum security system: based on rough set and support vector machine. Chin. Soft Sci. 11, 13–19 (2022). (in Chinese) 22. Joseph, I., Imoize, A.L., Ojo, S., Risi, I.: Optimal call failure rates modelling with joint support vector machine and discrete wavelet transform. Int. J. Image Graph. Signal Process. 14(4), 46–57 (2022) 23. Wiradinata, T.: Folding bicycle prospective buyer prediction model. Int. J. Inf. Eng. Electron. Bus. 13(5), 1–8 (2021) 24. Ahmed, N., Rabbi, S., Rahman, T., Mia, R., Rahman, M.: Traffic sign detection and recognition model using support vector machine and histogram of oriented gradient. Int. J. Inf. Technol. Comput. Sci. 13(3), 61–73 (2021)
Research on Port Logistics Demand Forecast Based on GRA-WOA-BP Neural Network Zhikang Pan1 and Ning Chen1,2(B) 1 School of Transportation and Logistics Engineering, Wuhan University of Technology,
Wuhan 430063, China [email protected] 2 Wuhan University of Technology, Sanya Science and Education Innovation Park, Sanya 572025, China
Abstract. Port logistics demand is a crucial part of China’s coastal industrial economy. In order to achieve a balanced supply and demand for port logistics and a reasonable allocation of logistics resources, the accurate prediction of port logistics demand has a vital impact on the future development of the port and the economic development of the hinterland cities. Firstly, the grey relational analysis method is used to screen out nine principal influencing factors of the logistics demand of Haikou port to realize the dimensionality reduction behavior of the input variables, the whale optimization algorithm is also introduced to optimize the weightings and thresholds of the BP neural network model, based on which the GRA-WOA-BP neural network port logistics demand forecasting model is established. The results indicate that in port logistics demand forecasting, the GRA-WOA-BP neural network to predict the results of the prediction error is all less than the BP neural network and WOA - BP neural network model. Its prediction accuracy is of significance better than that of the traditional prediction methods, with an accuracy rate of 98.82%. The research results also have certain reference significance for the strategic planning of port logistics development. Keywords: Port · Logistics demand forecasting · Grey relational analysis · Whale optimization algorithm · BP neural network
1 Introduction Along with our country economy continues to develop rapidly, at the same time also led to a rapid increase in China’s logistics needs, and the port as an important node in the transport network, to assume the main task of freight logistics distribution and transit, and at the same time, as the extent of our foreign trade exchange continues to expand, the progress of port logistics corresponding to the service of foreign trade is particularly important. At present, scholars’ research on port logistics mainly focuses on port logistics and regional economic development, sustainable evolution of port logistics, and prediction of port logistics demand, etc. Chen et al. used gray correlation analysis to screen the relevant indicators between port logistics and regional economy in Xiamen, and throw out © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 754–763, 2023. https://doi.org/10.1007/978-3-031-36115-9_67
Research on Port Logistics Demand Forecast
755
policy recommendations for this [1]. Xiao et al. studied the correlation between Yangtze River shipping logistics and inland economy from qualitative and quantitative perspectives, respectively, and the results of the study have important guiding significance for inland river ports [2]. Ji constructed a vector autoregressive VAR model and established an autoregressive model between Shanghai GDP and port cargo throughput to study the logistics development of Shanghai port [3]. Xu studied the sustainable evolution of port logistics to build a port logistics carbon emission evaluation system in the context of low carbon environment and studied the emission reduction efficiency of Dalian port to propose reasonable green optimization [4]. In order to explore the collaborative development trend between port logistics and urban economy, Zha and Tu developed a neural network prediction model on port synergistic development using port logistics indicators [5]. Yang used BP neural network to establish a forecasting model for import and export logistics demand at coastal ports and compared the analysis with the commonly used forecasting model, and the results indicated that the established forecasting model has a better prediction effect [6]. Zhuang et al. used principal component analysis to determine the factors influencing port green indicators and predicted port throughput through a gray prediction model [7]. Eskafi proposed a Bayesian statistics-based model for port cargo throughput prediction, which was verified to have a strong adaptive capability and high prediction accuracy [8]. Yan et al. used exponential smoothing and grey smoothing to forecast logistics demand, respectively, while constructing a combined forecasting model to overcome the lack of accuracy of a single forecasting model [9]. Therefore, a reasonable analysis of port logistics demand is not only a significant part of the development of the port industry or the port operations, the overall design, the target direction to play a key role. And not clear forecast of port logistics demand will affect its strategic development. In the process of forecasting, choosing the appropriate forecasting method for port logistics demand forecasting analysis is an urgent problem in the development of modern logistics.
2 Model Approach 2.1 Grey Relational Analysis Grey relational analysis is a method applicable to analyze the degree of influence of multiple variable factors on the target variable factors from a quantitative research perspective in a complex environment of variable factors, with the following main steps. Step 1: Given the target sequence y(k)(k = 1, 2, · · · , m), reference sequence x1 (k), x2 (k), · · · , xn (k). Step 2: Standardize each sequence to eliminate the influence of quantity. Get the x2 (k), · · · , xn (k). y(k) and x1 (k), Step 3: Calculate the relational coefficient. ri (k) =
dmin + dmax di (k) + λdmax
(1)
where dmin and dmax are the minimum and maximum differences between the two levels, di (k) = | y(k) − xi (k)|, λ is the resolution coefficient, generally 0 < λ < 1.
756
Z. Pan and N. Chen
Step 4: Calculate and rank the relation. 1 ri (k) m m
Ri =
(2)
k=1
where, 0 ≤ Ri ≤ 1, the closer to 1 indicates the stronger relation with the target sequence in the reference sequence. 2.2 GRA-WOA-BP Neural Network Model 2.2.1 BP Neural Network BP neural network is a kind of the input information is passed, then in the other direction, according to the error of multi-layer feedback neural network. It is often used as a function fitting, classification and prediction aspect because of its adaptive and selflearning, parallel processing of information and strong nonlinear mapping capability. The workflow of an usual three-layer BP neural network is shown in Fig. 1.
Fig. 1. BP neural network flow chart
2.2.2 Whale Optimization Algorithm Whale optimization algorithm was first proposed by the Australian scholar Mirjalili in 2016, which simulates the predatory behavior of humpback whales, mainly including the three stage process of encircling prey, bubble network attack, and searching for prey [10]. 1) Surrounding the prey During the initial stage of predation, the individual position of each whale is unknown, it is necessary to determine the location of the current prey to update the
Research on Port Logistics Demand Forecast
individual whale location and then lock the envelope as follows: → → − − → − − → D = C · X ∗ (t) − X (t) − → − → − − → → X (t + 1) = X ∗ (t) − A · D
757
(3) (4)
− → − → where X ∗ (t) is the location of the optimal individual; X (t) is the whale’s current location; − → t is the current number of iterations; D is the distance between the prey and the individual − → − → whale; C and A are the vector of coefficients to control how the whale swims, denoted as: − → → → → A = 2− a ·− r −− a (5) − → → C =2·− r
(6)
− → a = 2 − 2t/Tmax
(7)
→ → where, − a is the control parameter; − r is the random vector between [0, 1]; Tmax is the maximum number of iterations. 2) Bubble network attack (Local search) Individual whales use a bubble net feeding strategy during predation. The whale optimization algorithm is designed with two mechanisms of shrinkage envelope and spiral update position, and the model assumes a 50% probability of each for the experiment, and the specific model is shown below: − → − → − → − → p < 0.5 X ∗ (t) − A · D (8) X (t + 1) = − → bl − →∗ D · e · cos(2π l) + X (t) p ≥ 0.5 − − → − → → where, D = X ∗ (t) − X (t) is the distance between the whale and its prey; b is a constant; l is a random number between [−1, 1]. 3) Searching for prey (Global search) At this stage, whales no longer update their positions according to the prey, but randomly achieve position update by replacing the prey with other whale positions within the population, which guarantees that at this stage whales can complete the global search and avoid the defect of falling into the local optimum. → −−→ − − → − → (9) D = C · Xrand − X (t) → − − → −−→ − → X (t + 1) = Xrand − A · D
(10)
−−→ where Xrand represents the position of random individuals in the current whale population. The whale optimization algorithm has few adjustment parameters and is simple to operate, it can effectively solve the problems of BP neural network that is easy to fall into local optimum and slow convergence speed. The flow chart shown in Fig. 2.
758
Z. Pan and N. Chen
Fig. 2. WOA-BP neural network model flow chart
3 Port Logistics Demand Impact Indicator System Construction 3.1 Influencing Factor Indicator Selection With the national “One Belt One Road” strategic plan and the construction of Hainan Free Trade Port, the trend of globalization of the source of goods, trade and investment facilitation, and the trend of efficient flow of goods are obviously strengthened, and Hainan Province gradually formed the pattern of “four directions and five ports”, and As the largest port in Hainan Province, Haikou Port logistics development plays a key role in promoting the economic development of Hainan province the development of Haikou port logistics plays an important role in promoting the economic development of the whole Hainan Province. The foreign trade container route mainly connects with the international container route network of Southeast Asia region through Yangpu Port Container Port, or radiates to the national container routes of Europe and America region and East Asia region through Hong Kong, China, and the intensive domestic and foreign trade routes meet the huge market of goods import and export. Port logistics demand will be affected by a variety of factors, in order to can be more comprehensive port logistics demand forecast, combined with the existing literature research angle analysis, this paper selected three perspectives of socio-economic development, logistics industry development, port logistics level development for model index factor selection [11, 12].
Research on Port Logistics Demand Forecast
759
1) Socio-economic development. Industrial development and social and economic prosperity are inseparable, among which the three major industries of the national economy also play different degrees of influence on port logistics. Total city GDP and total retail sales of consumer goods reflect the degree of logistics demand and economic growth, while the service area of Haikou port is not only Haikou city, and the proportion of GDP of Haikou city to the province’s GDP is chosen to indicate the influence of Hainan economy on Haikou port. The total investment in fixed assets also plays a significant role in port logistics. 2) Logistics industry development. The development of the logistics industry is also as a significant influencing factor for the development of logistics in Haikou port, based on the availability of data and avoiding subjective will to lead to bias in the analysis of the results. So we choose two quantitative indicators for the analysis: the number of resident population in Haikou and the proportion of the number of people in the logistics industry to the total number of employees. 3) Port logistics level development. The comprehensive strength of Haikou port’s own logistics is a prerequisite for studying its logistics demand. In this paper, we comprehensively select four indicators, namely, the total value of foreign trade import and export, the number of berths above 10,000 tons, container throughput and freight turnover, as the reference factors affecting the development of port logistics level. 3.2 Input and Output Variables are Determined After clarifying the factors influencing the logistics demand of Haikou port, the following factors are mainly selected as input variables in this paper: Haikou GDP (X 1 ), the first industrial output value (X 2 ), the second industry output (X 3 ), the tertiary industry output value (X 4 ), total fixed asset investment (X 5 ), total retail sales of consumer goods(X 6 ) proportion of Haikou GDP to provincial GDP (X 7 ), foreign trade Import and export volume (X 8 ), berths above 10,000 tons (X 9 ), container throughput(X 10 ), freight turnover (X 11 ), resident population(X 12 ), proportion of employees in logistics industry to total employees(X 13 ), and port cargo throughput (Y ) are selected as output variables. Actual data’s are shown in Table 1. Table 1. Base data for input variables 2005–2020 Year
X1
X2
X3
X4
X5
X6
X7
2005
301.35
23.09
83.14
195.12
137.17
138.45
33.30
2006
350.12
25.94
102.12
222.06
158.23
161.35
33.30
2007
393.69
26.81
111.12
255.75
181.83
189.38
32.30
2008
443.18
31.40
113.28
298.50
221.43
234.75
30.40
2009
481.28
34.11
111.49
335.69
277.03
277.20
29.09
2010
595.14
38.19
142.83
414.13
349.65
326.94
28.83 (continued)
760
Z. Pan and N. Chen Table 1. (continued)
Year
X1
X2
X3
X4
X5
X6
X7
2011
713.30
47.71
177.91
487.68
404.59
387.18
28.28
2012
797.24
50.76
197.13
549.34
510.38
436.26
27.92
2013
904.64
58.10
217.03
629.51
649.33
505.35
28.75
2014
1091.70
57.07
217.49
817.15
821.53
558.47
31.19
2015
1161.96
57.09
223.67
881.21
1012.05
613.51
31.38
2016
1257.67
63.91
233.56
960.20
1271.73
673.30
31.10
2017
1390.58
62.51
252.22
1075.85
1415.50
742.72
31.16
2018
1510.51
63.21
276.00
1171.31
1327.74
787.25
31.30
2019
1671.92
71.18
275.99
1324.75
1123.27
823.94
31.49
2020
1791.58
79.88
269.56
1442.14
1234.47
835.89
32.40
Year
X8
X9
X 10
X 11
X 12
X13
Y
2005
12.35
2
22.03
406.60
173.73
5.34
2523.80
2006
28.50
2
24.48
443.16
176.68
4.72
3018.80
2007
35.20
3
28.6
530.97
179.45
5.65
3617.99
2008
45.40
5
34.64
555.74
183.50
5.84
3813.23
2009
38.10
7
43.52
465.16
187.85
6.25
4014.57
2010
39.48
7
61.34
564.86
204.62
6.78
4795.52
2011
39.36
7
80.8
594.91
209.73
6.70
5520.60
2012
42.15
7
100
665.66
214.13
6.93
6122.89
2013
51.40
7
116.82
839.76
217.11
7.67
7421.39
2014
34.00
7
134.67
963.54
220.07
8.36
7581.00
2015
43.40
6
127.5
708.73
222.30
9.38
8209.90
2016
39.20
17
140.18
669.68
224.36
9.33
8866.93
2017
31.10
25
163.6
460.17
227.21
10.49
10112.78
2018
50.88
25
184.67
519.42
230.23
9.90
10764.90
2019
48.12
25
197.26
1292.73
232.79
9.42
11197.56
2020
53.97
23
197.07
1497.86
288.66
9.76
10466.60
The 11 selected influencing factors were correlated with the logistics demand using grey correlation analysis, and the analysis results are shown in Table 2. In accordance with the grey correlation analysis, the influencing factor variables with a grey correlation of 0.65 or more were selected for the WOA-BP neural network model for training and testing, and X 1 , X 2 , X 3 , X 4 , X 5 , X 6 , X 10 , X 11 , X 13 were selected.
Research on Port Logistics Demand Forecast
761
Table 2. Grey relation degree between logistics demand and various factors in Haikou port Influencing Factors
X1
X2
X3
X4
X5
X6
X7
Grey relation degree
0.8491
0.8174
0.8429
0.7689
0.6618
0.8747
0.58
Influencing Factors
X7
X8
X9
X 10
X 11
X 12
X 13
Grey relation degree
0.58
0.6131
0.5848
0.7618
0.6624
0.6264
0.7433
4 Analysis of Simulation Experiments 4.1 Neural Network Parameter Setting The accuracy of GRA-WOA-BP neural network model, traditional BP neural network and WOA-BP neural network to forecast port logistics demand is studied, the neural network model parameters were set to a maximum training number of 1000, a training target error of 0.00001, and a learning rate of 0.01; the WOA model parameters were set to a population size of 30 and a maximum evolutionary generation of The research examples and model runs were conducted under the software MATLAB R2019b. 4.2 Analysis of Model Results The GRA-WOA-BP neural network model is used to train the logistics demand of Haikou port, and the prediction results are compared with the other two models, which are shown in Table 3. Table 3. Prediction results and relative errors of each model Year
True Value
BP
WOA-BP
GRA-WOA-BP
Predicted value
Relative Error
Predicted value
Relative Error
Predicted value
Relative Error
2018
10764.90
9003.9
16.4%
10073.9
6.4%
10934.9
1.6%
2019
11197.56
8838.56
21.1%
10072.56
10%
11042.56
1.4%
2020
10466.60
8845.6
15.5%
10073.6
3.8%
10407.6
0.6%
As can be seen from Table 3, the GRA-WOA-BP neural network model performs better than the traditional BP neural network model and the WOA-BP neural network model in predicting the logistics demand of Haikou port, and the relative error interval range of prediction is significantly smaller than the other two models. Meanwhile, to further study the prediction accuracy of the model, MAE, RMSE, MAPE and prediction accuracy are selected as the evaluation indexes in this research, and the smaller the error indexes representation model of prediction accuracy is higher, and the specific results are shown in Table 4.
762
Z. Pan and N. Chen Table 4. Comparison of prediction errors of each model
Model
MAE
RMSE
MAPE
BP
1913.8
Accuracy rate
1940.29
17.61%
82.36%
WOA-BP
736.50
795.42
6.74%
93.26%
GRA-WOA-BP
128.22
137.40
1.18%
98.82%
The GRA-WOA-BP model outperforms the other two models in port logistics demand prediction, and its prediction accuracy is as high as 98.82% from Table 4. In summary, the whale optimization algorithm for BP neural network is practical and feasible for weight and threshold finding, and the GRA-WOA-BP neural network model constructed in this study is applicable to the logistics demand prediction of Haikou port, and the prediction accuracy is higher than other prediction models.
5 Conclusion This research introduces the GRA-WOA-BP neural network model into the field of port logistics demand forecasting, selects the influencing factors affecting the logistics demand of Haikou port, forecasts the logistics demand of Haikou port from 2005 to 2020 by constructing the neural network model, and makes a comprehensive comparison of the forecasting results of the three models The results of the three models are compared and analyzed. Through the above study, the following main conclusions were obtained. 1) The factors affecting the development of logistics demand at Haikou port are systematically reviewed. 13 factors influencing the logistics demand of Haikou port were selected from three perspectives: socio-economic development, logistics industry development and port logistics level development, and 9 main influencing factors were clarified through grey relational analysis, among which the total retail sales of social consumer goods, GDP of Haikou city, primary industry, secondary industry and the logistics demand of Haikou port have a strong correlation relationship. 2) In port logistics demand forecasting, by introducing the whale optimization algorithm to optimize the initial weights and thresholds in the BP neural network, the relative error, mean absolute error, root mean square error and mean absolute percentage error of the forecasting model results are smaller than those of the BP neural network and WOABP neural network models, and the forecasting accuracy is higher. The GRA-WOA-BP neural network model can better clarify the nonlinear relationship in the complex network system, help to provide more valuable reference information for port decision makers, and enhance the rationality of port logistics demand planning, so as to provide a certain reference basis for the future port logistics industry. Acknowledgment. This project is supported by Major science and technology projects in Hainan (ZDKJ2020012) and Hainan open fund project (2020KF0051).
Research on Port Logistics Demand Forecast
763
References 1. Meng, C., Wanshan, Z., Jing, P.: Research on the relationship between port logistics and regional economy in Xiamen city based on gray relational analysis. In: Proceedings of 2018 5th International Conference on Education Reform and Management Innovation (ERMI 2018), pp. 92–98 (2018) 2. Xiao, R., Pan, L., Xiao, H., et al.: Research of intelligent logistics and high-quality economy development for Yangtze river cold chain shipping based on carbon neutrality. J. Mar. Sci. Eng. 10(8), 1029 (2022) 3. Ji, W.: Research on Shanghai port logistics based on VAR model. IOP Conf. Ser. Earth Env. Sci. 791(1), 012042 (2021) 4. Jinyu, X., Mo, L.: Carbon emission assessment of port integrated logistics in low-carbon environment. Int. J. Front. Sociol. 3(18), 55–60 (2021) 5. Zha, A., Tu, J.: Research on the prediction of port economic synergy development trend based on deep neural networks. J. Math. 2022, 1–9 (2022) 6. Yang, D.: Logistics demand forecast model for port import and export in coastal area. J. Coast. Res. 103(sp1), 678–681 (2020) 7. Zhuang, X., Li, W., Xu, Y.: Port planning and sustainable development based on prediction modelling of port throughput: a case study of the deep-water Dongjiakou port. Sustainability 14(7), 4276 (2022) 8. Eskafi, M., Kowsari, M., Dastgheib, A., et al.: A model for port throughput forecasting using Bayesian estimation. Marit. Econ. Logistics 23(2), 1–21 (2021) 9. Peng, Y., Lin, Z., Zhiyun, F.: Research on logistics demand forecast of port based on combined model. J. Phys. Conf. Ser. 1168, 032116 (2019) 10. Mirjalili, S., Lewis, A.: The whale optimization algorithm. Adv. Eng. Softw. 95, 51–67 (2016) 11. Ma, H., Luo, X., Varatharajan, R.: Logistics demand forecasting model based on improved neural network algorithm. J. Intell. Fuzzy Syst. 40(4), 6385–6395 (2021) 12. Ren, X., Tan, J., Qiao, Q., et al.: Demand forecast and influential factors of cold chain logistics based on a grey model. Math Biosci. Eng. 19(8), 7669–7686 (2022)
Evaluation and Optimization of the Port A Logistics Park Construction Based on Fuzzy Comprehensive Method Xin Li1 , Xiaofen Zhou1 , Meng Wang1 , Rongrong Pang2 , Hong Jiang1(B) , and Yan Li1 1 School of Logistics, Wuhan Technology and Business University, Wuhan 430065, China
[email protected], [email protected] 2 Philippine Christian University Center for International Education, 1004 Manila, Philippines
Abstract. With the rapid development of China’s cities, manufacturing and commerce, logistics, as a supporting service industry, has occupied an increasingly important position in the development of the national economy. The construction of the logistics park has become a hot spot for the promotion and revitalization of regional logistics industry and economic development. The demand for the logistics industrial park has gradually expanded, but it has also been accompanied by uneven management, services, facilities and equipment, resulting in the waste of a package of resources and not playing its due role. The logistics park plays an active role in improving economic efficiency and logistics efficiency. At the same time, it has the characteristics of large investment, long development cycle, many uncertain factors and high risk in the construction of logistics park. Taking the project of A port Logistics Park as an example, this paper analyzes various risks and related factors in the construction of the logistics park, establishes a risk evaluation index system, analyzes the overall system of the logistics park, uses the fuzzy comprehensive evaluation method to evaluate the risk level of Port A Logistics Park, and puts forward corresponding optimization strategies and plans based on the evaluation results. Keywords: Logistics Park Construction · Index System · Fuzzy Comprehensive Method · Evaluation and Optimization
1 Introduction The National Logistics Park Development Plan (2013–2020) (hereinafter referred to as the “Plan”) has been released recently, clarifying the development objectives and overall layout of the national logistics park, and drawing a “road map” for the development of the logistics park [1]. Port construction usually faces many problems and challenges, such as channel engineering, queuing phenomenon, management organization behavior, etc., which will affect port operation to varying degrees [2–4]. Port A is committed to improving the comprehensive handling capacity of containers in the port, making the container industry bigger and stronger, striving to achieve its position as the main channel in container © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 764–773, 2023. https://doi.org/10.1007/978-3-031-36115-9_68
Evaluation and Optimization of the Port A Logistics Park Construction
765
transport in central China, and playing a positive role in the construction of Wuhan Shipping Center and port A. As one of the earliest ports opened to the outside world in Wuhan, Port A has developed into a major distribution center for foreign trade import and export goods in Hubei Province, with more than 400 flights per month, accounting for about two-thirds of the total container transport in Wuhan, and is the largest professional international container terminal in the middle and upper reaches of the Yangtze River [5]. Therefore, the quality of port logistics park project construction directly affects China’s foreign trade economy. A set of index system for the construction project of Wuhan Xingang Logistics Park has been constructed, and the risk assessment method for the construction project of Port A Logistics Park has been summarized, which not only provides a reference for the development research of the construction planning of Port A Logistics Park in Wuhan in the future, but also provides a reference for the research of the construction investment project of Port A Logistics Park to a certain extent, which is conducive to the evaluation of the development level of the relevant port logistics system in theory. The logistics park has large investment scale, long construction period and many uncertain factors. The evaluation of Port A’s logistics construction project and the analysis of various regions, elements and their interactions in the port logistics system are helpful to find out some problems in the development of Port A’s logistics, and help the relevant departments and institutions in the construction of Port A to solve these problems, which has certain guiding significance for the construction and development of Port A [6].
2 Overview of Relevant Theories 2.1 Fuzzy Comprehensive Method The fuzzy evaluation method is based on fuzzy mathematics. Fuzzy mathematics was born in 1965, and its founder is American automatic control expert L.A. Zadeh. This comprehensive evaluation method transforms qualitative evaluation into quantitative evaluation according to the theory of membership degree of fuzzy mathematics, that is, to make an overall evaluation of things or objects subject to multiple factors by using fuzzy mathematics. It has the characteristics of clear results and strong systematisms, can better solve fuzzy and difficult to quantify problems, and is suitable for solving various uncertain problems [7]. Due to a series of problems such as the complexity of evaluation factors, the hierarchy of evaluation objects, the fuzziness of evaluation criteria, the fuzziness or uncertainty of evaluation influencing factors, and the difficulty in quantifying qualitative indicators, it is difficult for people to accurately describe the objective reality with the absolute “either this or that”, and there is often the fuzzy phenomenon of “this or that”, and its description is often expressed in natural language, The biggest feature of natural language is its fuzziness, which is difficult to be measured uniformly by classical mathematical models [8]. Therefore, the fuzzy comprehensive evaluation method based on the fuzzy set can comprehensively evaluate the subordination level of the evaluated object from multiple indicators. It divides the change range of the evaluated object. On the one hand, it can take into account the hierarchy of the object, so that the fuzziness of the evaluation criteria
766
X. Li et al.
and influencing factors can be reflected; On the other hand, human experience can be fully used in the evaluation to make the evaluation results more objective and consistent with the actual situation. Fuzzy comprehensive evaluation can combine qualitative and quantitative factors, expand the amount of information, improve the degree of evaluation, and make the evaluation conclusion credible [9]. 2.2 Characteristics of Logistics Park Construction Project Logistics Park refers to a place where multiple logistics facilities and different types of logistics enterprises are centrally arranged in space in areas where logistics operations are concentrated and where several modes of transportation are connected. It is also a gathering point for logistics enterprises with a certain scale and multiple service functions [10]. 1) Multiplicity of construction objectives. The objectives of general construction projects mainly focus on the use value and economic value of the project service itself. Different from general construction projects, the logistics park has a public welfare attribute, which leads to the project objectives of the logistics park not only focusing on the economic benefits of the construction of the logistics park itself, but also focusing on the social effectiveness of the development and construction and the driving effect of the construction of the logistics park on the local economy. In short, the logistics park project should pay attention to both spectator benefits and macro benefits [11]. 2) Diversity of construction contents. The logistics park project is different from the single objective logistics facility construction (such as warehouse) and transportation road construction projects. As a logistics park construction project at a certain level, for example, in domestic and international logistics services, the construction content of the logistics park not only corresponds to the basic logistics functional elements of logistics operators such as transportation, warehousing, packaging, distribution, loading and unloading, but also includes the development and construction of diversified logistics infrastructure and logistics service facilities such as customs declaration, inspection, catering, accommodation, maintenance, transportation, finance, insurance, so on and so forth [12]. 2.3 Risk Characteristics of Logistics Park Construction Projects In addition to the typical characteristics of general construction project risks, i.e., wide sources, long development cycle, dynamics, objectivity, regularity, relativity and uncertainty, the logistics park construction project risks also show the following significant characteristics: 1) Diversity and multilevel of risks. The logistics park construction project has the characteristics of large investment, long development cycle, many uncertain factors and complex types, and the cross impression between a large number of risk factors and the outside world makes the risk present multi-level. 2) One-way transitivity. In the m stages of logistics park project construction, the risk impact has a one-way transmission from front to back in time / stage, and the risk factors in the previous stage will have an impact on the subsequent stages, that is, the
Evaluation and Optimization of the Port A Logistics Park Construction
767
risk has a one-way transmission from front to back. Generally, for example, the risk factors in the m stages of logistics park construction are U1 , U2, · · · Um , and the risk impacts of each stage are R1 , R2, . . . Rm , then the one-way transmission of project construction risk can be expressed as R1 = f1 (U1 ), R2 = f2 (U2 ), · · · , Rm = fm (U1 , U2 , · · · Um ) 3) Environmental adaptability. Every logistics park construction project exists in a certain environment, and exchanges logistics, information and capital flow with the external environment. The same type of logistics park construction project risk in different project construction environment, the impact of environmental factors is also different. 4) Policy impact. Policy risks mainly include industrial development policy, land policy, investment policy, fiscal and tax policy and monetary policy, which may have an adverse impact on the investment and construction environment of the logistics park.
3 Analysis on Risk Factors of Port a Logistics Park Construction 3.1 Analysis on the Environment of Port a Logistics System 1) Shoreline resources. Port A planning area includes Hanjiang River and Yangtze River. The conditions of the Yangtze River channel are relatively superior. The Wuhan Shanghai section of the Yangtze River channel is a first-class channel with a deeper water depth than the upstream section of the Wuhan Yangtze River Bridge. 5000-ton ships can enter and leave the port all year round, ensuring the normal navigation of river sea direct ships throughout the year. The “river sea direct” route can be opened normally and the “river sea combined transport” can be realized. 2) Collecting and distributing network resources. The collection and distribution network resources of Wuhan Newport are mainly composed of three levels. The first level is the general framework composed of the planned “two rivers, four rings, four railways and nine shoots”. The second level is composed of some expressways and expressways connected with each other. The third level is the traffic structure composed of the traffic in the port. The third level constitutes the collection and distribution network system of Wuhan Newport [13]. 3.2 Analysis on Risk Factors of Logistics Park Construction The main contents of the investment, construction and operation stages of port A logistics park project are analyzed, as shown in Fig. 1.
768
X. Li et al.
Fig. 1. Phase tasks of logistics park
Many risk factors are involved in the whole implementation cycle of port A logistics park project. Through the analysis of the risks and risk nature of each stage of the project, it is summarized that the main investment risks of the port A logistics park project are as follows: Technical risks. In the two stages of port A logistics park project implementation, risks will be caused due to the uncertainty of relevant technical factors. For example, the rationality of site selection, the suitability of scale, the rationality of planning and layout, the suitability of functional design, the advanced applicability of technology, etc. Market risks. It is also the most important risk faced by port A logistics park after its operation. Port A logistics park needs to objectively investigate and analyze the logistics demand in Central China and even the entire economic hinterland, the status of market competitors and the current market situation of logistics services, and analyze whether the potential of the future logistics market matches the scale and service level of the project [14]. Project management risks. For the port A logistics park project, the factors that affect the management risk mainly include: the rationality of the management organization and division of labor, the soundness of the quality safety system, the coordination of public relations, the selection of contracting mode and the management ability of the management personnel. Economic and financial risks. For the port A logistics park project, the investment scale is huge, the financing channels are diverse, and the construction cycle is long. The land acquired by the project, the project infrastructure construction and the follow-up operation of the project all need a lot of funds. Therefore, fund raising is the biggest factor affecting the economic and financial risks of the project. Environmental risks. In the investment, construction and operation stages of port A logistics park project, many factors will affect the normal implementation of the park project, such as the port climate, the national political and economic environment, changes in the microeconomic environment [15].
Evaluation and Optimization of the Port A Logistics Park Construction
769
4 Evaluation of Port a Logistics Park Construction Based on Fuzzy Comprehensive Method 4.1 Construction of Risk Evaluation Index System for Logistics Park According to the characteristics of port A logistics park project with multi-level risk factors and many uncertain factors, the fuzzy comprehensive evaluation method is selected as the risk evaluation method of port A logistics park construction project. At the same time, the risk grade of port A logistics park project risk and the weight of each influencing factor are determined through expert investigation. Through the analysis of the implementation stage and risk identification of the port A logistics park project, it can be seen that the port A logistics park project is divided into two stages, and the risk factors are different in different stages. The specific evaluation index system is shown in Fig. 2.
Fig. 2. Schematic diagram of risk assessment index system
4.2 Establish Indicators and Evaluation System To make the students become the “protagonist” in the class, first of all, let the students look at the major and study from a different perspective, let the students clearly know what to learn and how to use it? Driven by the project, with the international logistics operation process as the core and team cooperation as the necessary form, students are guided to in-depth practice and form the project results in the course. Integrate the course achievements with the enterprise needs to achieve the “conversion agent” of learning style. 1) Define the main factor layer indicator set as U: (U 1 ,U 2 ,U 3 ,U 4 ,U 5 ), and its corresponding weight set as W: (W 1 ,W 2 ,W 3 ,W 4 ,W 5 ), where W i represents the proportion of indicator U i in U.
770
X. Li et al.
2) Define the sub factor layer indicator set as U i = (U i1 ,U i2 ,…,U ij ), and the corresponding weight is W: (W i1 ,W i2 ,…,W ij ), where W ij represents the proportion of indicator U ij in U i . The expert investigation method is used to judge the importance of each risk factor, and the weights of the primary and secondary risk factor indicators are as follows: W = (0.20, 0.20, 0.10, 0.30, 0.20)
(1)
W1 = (0.30, 0.30, 0.20, 0.20)
(2)
W2 = (0.30, 0.30, 0.20, 0.20)
(3)
W3 = (0.20, 0.30, 0.30, 0.20)
(4)
W4 = (0.30, 0.30, 0.30, 0.20)
(5)
W5 = (0.20, 0.20, 0.25, 0.20, 0.15)
(6)
Define the comment set as V: (V1,V2,…,V5), which respectively represent low risk, low risk, medium risk, high risk and high risk. 4.3 Effect Fuzzy Evaluation Matrix ⎡
⎤ 0.3 0.3 0.2 0.1 0.1 ⎢ 0.3 0.4 0.2 0.1 0.0 ⎥ ⎥ R1 = ⎢ ⎣ 0.4 0.4 0.1 0.1 0.0 ⎦ 0.4 0.3 0.2 0.1 0.0 ⎡ ⎤ 0.3 0.2 0.2 0.2 0.1 ⎢ 0.3 0.3 0.2 0.2 0.0 ⎥ ⎥ R2 = ⎢ ⎣ 0.3 0.4 0.1 0.1 0.1 ⎦
(7)
(8)
0.4 0.3 0.3 0.0 0.0 ⎡
⎤ 0.3 0.4 0.2 0.1 0.0 ⎢ 0.3 0.3 0.2 0.1 0.1 ⎥ ⎥ R3 = ⎢ ⎣ 0.4 0.4 0.2 0.0 0.0 ⎦
(9)
0.3 0.3 0.2 0.1 0.1 ⎡
⎤ 0.4 0.3 0.2 0.1 0.0 ⎢ 0.3 0.3 0.2 0.2 0.0 ⎥ ⎥ R4 = ⎢ ⎣ 0.3 0.2 0.3 0.1 0.1 ⎦ 0.4 0.3 0.2 0.1 0.0
(10)
Evaluation and Optimization of the Port A Logistics Park Construction
⎤ 0.3 0.4 0.2 0.1 0.0 ⎢ 0.2 0.4 0.3 0.1 0.0 ⎥ ⎥ ⎢ ⎥ ⎢ R5 = ⎢ 0.3 0.3 0.3 0.1 0.0 ⎥ ⎥ ⎢ ⎣ 0.3 0.2 0.3 0.1 0.1 ⎦ 0.2 0.3 0.3 0.2 0.0
771
⎡
(11)
4.4 Fuzzy Evaluation Calculation Bi =Wi × Ri = (bi1 , bi2 , . . . bi5 ) B =(W1 , W2 , W3 , W4 , W5 ) × (B1 , B2 , B3 , B4 , B5 )T =(0.321, 0.311, 0.214, 0.215, 0.119, 0.031) According to the principle of maximum risk subordination, the comment V 1 represented by 0.322 - low risk is the overall risk level of the logistics park project. As the overall risk is at a low level, the project can be implemented from the risk perspective.
5 Optimization of Port a Logistics Park Construction 5.1 Improve the Service Mode of Diversified Logistics Parks The logistics park has continuously improved its service capacity and provided basic supporting services such as office, catering, property, parking, accommodation, industry and commerce, taxation, etc. for the settled enterprises. Some logistics parks have extended the service chain and provided value-added services such as logistics consulting, logistics finance, commodity display, facility leasing, insurance agency, etc. for the settled enterprises. The service categories are increasingly rich, becoming a new growth point of the park. By improving the standardization of logistics operations, reducing the cost of logistics services and striving to improve the level of logistics services can enterprises improve their competitiveness in the logistics market with high-quality services and reasonable service prices. 5.2 The Logistics Park is Upgraded to Intelligence Increase the establishment of the information platform of Port A Logistics Park, provide basic services such as information release, goods tracking, data exchange, property management, and improve the informatization level of the park. Some parks have developed business auxiliary functions such as capacity trading, payment and settlement, financing insurance, credit management, etc. in combination with business needs to accelerate the digital development of the park. With the development of equipment technology, the integration of new generation communication technologies such as the Internet of Things, big data, cloud computing, artificial intelligence and logistics parks is the main feature of the development of logistics parks at present.
772
X. Li et al.
5.3 Actively Seek Government Support The introduction of the national logistics development plan and policy will clarify the investment orientation and support for logistics infrastructure projects and enterprises, including logistics parks, and will be more beneficial to the development of logistics parks that meet the planning requirements or are included in the government planning. For example, preferential land tax policies, supporting infrastructure construction, and some construction funds. The active support of the government can effectively reduce the construction cost, reduce the financing difficulty, facilitate the coordination between various departments, and reduce the risk of the planning and implementation of the logistics park to a certain extent [16].
6 Conclusion Logistics parks play an increasingly important role in improving the efficiency of logistics services, promoting the adjustment of industrial structure, transforming the mode of economic development, and serving the national development strategy. As a distribution center for inbound and outbound goods and a logistics platform for the procurement and distribution of large manufacturers, the logistics park occupies a core position in the entire logistics chain. The construction of logistics park is characterized by large investment, long development cycle and many uncertain factors. Therefore, the risk of logistics park construction project is high. Taking Port-A Logistics Park project as an example, this paper analyzes various risks and related factors in the construction of the logistics park, establishes a risk evaluation index system, analyzes the overall system, evaluates the risk level of Port A Logistics Park through fuzzy comprehensive evaluation method, and puts forward corresponding optimization strategies. Acknowledgment. This research was funded by Industry-education integration curriculum project of Wuhan Technology and Business University.
References 1. Nan, L., Zhao, J., Lyu, M.: Exploration of land development intensity index of port container Logistics Park based on quantitative algorithm and pent analysis method. Pol. Marit. Res. 25(3), 61–67 (2018) 2. He, Q,-M., Ke, J., Dong, S., Wang, X., Yuan, Z.: Simulation Modelling of State-dependent Queueing Network: impact of deepening on vessel traffic in Yangtze River Estuary. Adv. Mech. Eng. 11(5), 1–12 (2019) 3. He, Q.-M., Zhang, H., Ye, Q.: An M/PH/K queue with constant impatient time. Math. Methods Oper. Res. 18(11), 139–168 (2018) 4. Zhang, H., Zhao, X.: Analysis of organizational behavior of container shipping in the upper and middle reaches of the Yangtze River based on Hub-and-Spoke network. J. Coastal Res. Sp(73), 119–125 (2015) 5. Mainwaring, G., Olsen, T.O.: Long undersea tunnels: recognizing and overcoming the logistics of operation and construction. Engineering 4(2), 249–253 (2018)
Evaluation and Optimization of the Port A Logistics Park Construction
773
6. Guan, S.: Smart E-commerce logistics construction model based on big data analytics. J. Intell. Fuzzy Syst. 40(2), 1–9 (2020) 7. Wu, Y., Zhang, T.: Risk Assessment of offshore wave-wind-solar-compressed air energy storage power plant through fuzzy comprehensive evaluation model. Energy 223(5), 120057 (2021) 8. Zhang, H., He, Q.-M., Zhao, X.: Balancing herding and congestion in service systems: a queueing perspective. Inf. Syst. Oper. Res. 58(3), 511–536 (2020) 9. Qian, J., Wu, J., Yao, L., et al.: Comprehensive performance evaluation of wind-solar-cchp system based on emergy analysis and multi-objective decision method. Energy 230(190), 120779 (2021) 10. Liu, L.: Oil spill risk assessment of submarine oil pipeline based on fuzzy comprehensive evaluation and accounting (Retraction of vol 14, art no 1911, 2021). Arab. J. Geosci. 24, 14 (2021) 11. Wang, J., Liu, S., Wang, S., et al.: Multiple indicators-based health diagnostics and prognostics for energy storage technologies using fuzzy comprehensive evaluation and improved multivariate grey model. IEEE Trans. Power Electron. 36(11), 12309–12320 (2021) 12. Verbic, G., Keerthisinghe, C., Chapman, A.C.: A project-based cooperative approach to teaching sustainable energy systems. IEEE Trans. Educ. 60(3), 221–228 (2017) 13. Fatima, S., Abdullah, S.: Improving teaching methodology in system analysis and design using problem based learning for ABET. Int. J. Mod. Educ. Comput. Sci. 5(7), 60–68 (2013) 14. Kaviyarasi, R., Balasubramanian, T.: Exploring the high potential factors that affects students’ academic performance. Int. J. Educ. Manag. Eng. (IJEME) 8(6), 15–23 (2018) 15. Al-Hagery, M.A., Alzaid, M.A., Alharbi, T.S., et al.: Data mining methods for detecting the most significant factors affecting students’ performance. Int. J. Inf. Technol. Comput. Sci. 12(5), 1–13 (2020) 16. Zhiqin, L., Jianguo, F., Fang, W., Xin, D.: Study on higher education service quality based on student perception. Int. J. Educ. Manage. Eng. 2(4), 22–27 (2012)
Advances in Technological and Educational Approaches
OBE Oriented Teaching Reform and Practice of Logistics Information System Under the Background of Emerging Engineering Education Yanhui Liu1 , Jinxiang Lian2 , Xiaoguang Zhou2(B) , and Liang Fang2 1 Department of Automation, Century College, Beijing University of Posts and Telecommunications, Beijing 102101, China 2 School of Automation, Beijing University of Posts and Telecommunications, Beijing 100876, China [email protected]
Abstract. The construction of Emerging Engineering Education is a major action plan to continuously deepen the reform of engineering education to respond the challenges of the new economy, meet the needs of the industry and face the future development. It has the characteristics of reflecting the feature of the times, new and rich connotation, multi-disciplinary integration, multiplex subject participation, and wide coverage. Under the background of Emerging Engineering Education construction, guided by the OBE teaching concept and taking the logistics information system course as an example, this paper has conducted in-depth research and practical innovation from the aspects of courses’ content, teaching methods, assessment ways, and the analysis of the degree of reaching the curriculum objectives, which reflects the multidisciplinary cross integration curriculum design idea of focusing on students and facing industrial needs and capacity training. The research of this paper has a certain reference significance for the training of compound and applied talents and the reform of curriculum teaching in relevant majors of colleges and universities. Keywords: Emerging Engineering Education · OBE · Logistics information system · reform in education
1 Introduction Engineering education is an important task to promote national development [1]. For the purpose of actively responding to a new round of scientific and technological revolution and industrial reform, and promoting the reform and innovation of engineering education, the Ministry of Education of China began to actively promote the construction of Emerging Engineering Education (EEE) disciplines in 2017 [2, 3]. It is a response to China’s implementation of major national strategies such as innovation driven development, “Made in China 2025”, “Internet plus”, “Network Power”, and “the Belt and Road”, and is also a strong driver of China’s engineering education reform [4, 5]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 777–786, 2023. https://doi.org/10.1007/978-3-031-36115-9_69
778
Y. Liu et al.
One of the core concepts of China’s engineering education reform is outcome based education (OBE for short), which mainly follows the principle of reverse design [6– 8]. Starting from the needs (including internal and external needs) setting which is the first step; the second step is to determine the training objectives; the third step is to establish the graduation requirements; and then fix on the curriculum system according to the graduation requirements. Finally, decide the course content and the class hours of the teaching content according to the contribution of a specific course to achieving the graduation requirements [9–12]. This paper takes the logistics information system course as the research object, combined with the OBE education concept, aiming at the three major problems of teachers’ teaching, students’ learning and knowledge application difficulties [13]. Focusing on project teaching, our teaching practice and innovation are carried out from five aspects, namely: teaching objectives, teaching content, teaching strategies, evaluation system and ability achievement, so as to effectively improve the teaching quality of our application-oriented undergraduate professional education.
2 Reform Objectives and Ideas of Course Teaching 2.1
Ideas for Curriculum Teaching Reform
The course Logistics Information System is one of the core course of logistics specialty in colleges and universities, which has the characteristics of multiple contents, wide coverage and complex knowledge structure. The difficulties of teachers’ teaching are mainly reflected in the selection of teaching materials and the design of teaching contents; The trouble s of students’ learning are mainly reflected in the unclear learning objectives, low learning initiative, and lack of engineering thinking and system thinking; The difficulty in comprehensive application is mainly reflected in the weak ability of students in comprehensive application of single or multiple professional courses [14]. Considering the above problems and the OBE concept, combined with the actual needs of enterprises, this paper introduces project teaching, carries out the project team as the teaching object, takes a single semester as the teaching cycle, forms a project based through and full cycle teaching management, and carries out feedback and improvement through teaching practice. The basic idea of curriculum construction is shown in Fig. 1. Under the guidance of OBE concept, the training objectives for college students are formulated and graduation requirements are clarified. On this basis, carry out curriculum construction. The main problems to be solved are “why to learn”, “what to learn”, “how to learn”, “how to evaluate” and “how to improve”. The contents involved include curriculum objectives, content reconstruction, teaching organization, curriculum evaluation and degree of achievement analysis. In the meantime, the quality and effect of curriculum construction need to be constantly improved through information feedback. In the course construction, the integration between schools and enterprises is also needed to achieve the transformation of learning achievements.
OBE Oriented Teaching Reform and Practice of Logistics Information System OBE philosophy
779
Graduation requirements
Cultivation goals Curriculum building
Course objectives
Content refactoring
Teaching organization
Course evaluation
Degree of achievemen t analysis
Why study? ·Course import ·Build a team
What to learn? ·Project oriented ·Knowledge refactoring
How to learn? ·Student-oriented ·Process management ·Online + offline
Learning assessment ·Multi-index evaluation ·Phased tasks
Learning Improvement ·Achievement analysis ·Improve the program
Feedback School-enterprise integration Translation of learning outcomes
Fig. 1. Logistics information system course construction ideas
2.2
Determination of Course Teaching Objectives
Under the guidance of OBE concept, the curriculum teaching objectives are determined according to the degree of contribution of the curriculum to the graduation required ability index points [15, 16]. Those ability index points with a contribution of 50% or more are selected as the objectives of the course teaching [17, 18]. The teaching of logistics information system course has the following six specific objectives. Objective 1: Have ability to apply mathematics, computer, engineering foundation and professional knowledge to the analysis of complex engineering problems of logistics information system; Objective 2: Be capable to master certain methods of literature search and investigation and analysis, identify, analyze and express complex problems in the logistics information system by using relevant basic engineering knowledge and engineering technology methods, and have the ability to learn independently; Objective 3: Have a certain awareness and ability of project management, be able to effectively carry out project communication and team cooperation, and complete project progress management, scope management, cost management, quality management and risk management; Objective 4: Be familiar with the business and data processing process of typical logistics information systems such as order management, procurement management, warehousing management, distribution management and transportation management, and be able to use software development technology, database technology, prototype design and other auxiliary tools to design and develop typical logistics information systems;
780
Y. Liu et al.
Objective 5: Be able to master the application principles and methods of typical logistics information technologies, such as one-dimensional barcode technology, twodimensional code technology, RFID technology, EDI, GIS and tracking technology; Objective 6: Can to make rational use of industry standards or norms, and use system thinking and engineering thinking to propose overall solutions to complex engineering problems in the logistics information system, and have a strong team spirit and innovation awareness. The relationship between the curriculum objectives of logistics information system and the graduation required ability index points is shown in Fig. 2. Graduation requirements
GR2 Engine ering Basics
decompose Competency indicators
GR22
GR32
GR3 Proble m analy-s is
GR4 Design /develo p solute-i ons
GR41 GR42 GR43 GR44
GR5 resea rch
GR52
GR6 Engineeri ng applicatio ns and problem solving
GR61 GR62
GR7 Use of modern tools
GR71 GR72
GR12 Comm unicati on and team-w ork
GR121 GR122
GR13 project manag ement
GR131 GR132
GR15 Self-dir ected learn-i ng
GR151 GR152
prop up
Course objectives
Course objectives 1
Course objectives 2
Course objectives 3
Course objectives 4
Course objectives 5
Course objectives 6
Fig. 2. The relationship between curriculum objectives and graduation requirements
In which, GRi is the graduation requirement, GRij refers to the ability index points required for each graduation, which are described as follows: GR22 : Be able to apply mathematics, computer, engineering foundation and professional knowledge to the analysis and interpretation of complex engineering problems of logistics information system; GR32 : Be able to identify the main links and key factors in the logistics system, and conduct qualitative and quantitative analysis of the survey results with the help of commonly used analysis tools; GR41 : Have a certain project management awareness and ability, be able to effectively communicate, cooperate and motivate the team, and implement quantitative assessment of the project process; GR42 : Be familiar with the business and data processing process of typical logistics systems, such as procurement management, warehousing management, distribution management, transportation management, and carry out project planning and implementation according to specific objects; GR43 : Master the application principles and methods of typical logistics technologies in the logistics system, such as one-dimensional barcode technology, two-dimensional code technology, RFID technology, EDI, GIS and some tracking technologies;
OBE Oriented Teaching Reform and Practice of Logistics Information System
781
GR44 : Be able to assist the design and development of typical logistics information systems with the help of commonly used tools such as prototype design, database design and process drawing; GR52 : Propose reasonable and effective solutions based on the actual needs of the enterprise and taking into account the operability, implement ability, application value and other indicators; GR61 : Have systematic and innovative thinking, and be able to analyze and solve problems from multiple perspectives such as project management, project planning and technology development; GR62 : In combination with the actual needs of the enterprise, considering the operability, implementability and application value of the scheme, propose an integrated logistics information system solution for complex problems; GR71 : Use EXCEL, SPSS and other statistical analysis tools or some big data analysis tools to make statistics, analysis and prediction on complex engineering problems; GR72 : Make scientific and rational use of network resources, and reserve some commonly used technical tools and platform resources; GR121 : Establish communication awareness, master communication skills, and be able to communicate and report effectively with peers or the public by modern means; GR122 : Through team activities, we will improve our ability to collate data, organize words, write documents and make PPT for summary and report. GR131 : Be able to master the basic principles and methods of project management; GR132 : Be able to complete the project planning and implementation under the interdisciplinary environment as a team member or leader; GR151 : Establish the awareness of lifelong learning and explore the methods of lifelong learning; GR152 : Develop goals and plans for independent learning to adapt to long-term career development.
3 Content Reconstruction and Organization Implementation of Curriculum Teaching 3.1
Reconstruction of Course Teaching Content
In view of the characteristics of multi-disciplinary intersection and integration emphasized in the construction of EEE courses, this paper focuses on project teaching and constructs systematic theoretical teaching content and continuous practical teaching content according to the course teaching objectives. In the design of theoretical teaching content, the process management of the project is added, including the project application, scope management plan, schedule management plan, cost management plan, quality management plan and risk management plan; The design of practical teaching content focuses on comprehensive and designed experimental projects, including feasibility analysis, project planning, system analysis and system design. This combination of in class experiments and project teaching not only reflects the integration of teaching content design, but also reflects the project oriented, all cycle teaching management concept. It is helpful to cultivate students’ systematic thinking and engineering practice ability.
782
Y. Liu et al.
3.2 Organization and Implementation of Teaching Process Figure 3 shows the arrangement of key teaching links of logistics information system. The teaching process of logistics information system is carried out in the form of a project team. A total of 18 weeks from the beginning to the end of the semester are required to complete the phased tasks. The class hours include two parts: explicit class hours and implicit class hours. Teachers use explicit class hours to explain, comment, supervise and guide, and students use implicit class hours to investigate, plan, analyze and design. Through the organization and implementation of this teaching process, the participation and initiative of students majoring in logistics engineering in practical teaching have been significantly enhanced, and the quality of learning achievements has also been significantly improved, which truly reflects the student-centered education and teaching philosophy. Serial Numbe r
Week
Stage Tasks
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
0
1
2
3
4
5
6
8
9
Project 1
Initiation Review
2 3 4
System planning systems analysis System design System
5
impleme ntation
Fig. 3. Arrangement of key teaching links of logistics information system
4 Process Based Multi Indicator Comprehensive Course Evaluation and Analysis of Course Goal Achievement The logistics information system curriculum has reformed a single outcome evaluation, built a multi indicator assessment mechanism under phased tasks with process management, and formed a process assessment file. The specific curriculum evaluation indicators are shown in Fig. 4. The evaluation indicators mentioned above are classified according to the teaching links to obtain the supporting relationship and assessment proportion between the curriculum objectives and teaching links, as shown in Table 1.
OBE Oriented Teaching Reform and Practice of Logistics Information System
Systems analysis report
System prototype System design report
System report
Systems analysis
Systems design
Project application
Core assessment objectives
Project Initiation Review Survey report
Requirements analysis report Scope plan
Program management
test
System implementation reports
Requirements analysis
783
Schedule Cost planning
Systems implementation
Quality plan Risk plan
Fig. 4. Assessment indicators and phased results of logistics information system
Table 1. The supporting relationship between curriculum objectives and teaching links and the assessment ratio.
According to Table 1, the supporting weight of each teaching link to the curriculum goal and its corresponding goal score are calculated, as shown in Table 2. Assume that the sum of the supporting weights of each teaching link j for a course objective i is 1, and the calculation formula is as follows: n j=1
ωij = 1
(1)
784
Y. Liu et al.
Considering the supporting relationship between teaching links and curriculum objectives, the weight of each teaching link j supporting a certain curriculum objective i is: x j , pij = 1 xj pij ωij = (2) 0 , pij = 0 where xj is the assessment proportion of each teaching link, pij is a 0–1 variable. 1 indicates that teaching link j supports a certain curriculum goal i, and 0 refers that teaching link j does not support a certain curriculum goal i. The support weight of each teaching link is obtained according to formula (2), as shown in Table 2. Table 2. The supporting weight and target score of each teaching link on the course objectives.
According to the support weight of each teaching link j to a certain curriculum goal i in Table 2, the degree of achievement of the curriculum goal is calculated, and the calculation formula is: fj × ωij (3) di = Fj where di is the degree of achievement of the course objectives, fj is the actual score value obtained by students corresponding to teaching link j, Fj is the target score of teaching link j. If the assessment scores of each teaching link of a student’s logistics information system are respectively 4, 3, 11, 28, 5 and 35, according to formula (3), it can be got that the degree of achievement of each goal is 0.8555, 0.8555, 0.8706, 0.8555, 0.8800 and 0.8555, and then compare the degree of achievement of each goal with the set qualified standard value.
OBE Oriented Teaching Reform and Practice of Logistics Information System
785
If it is greater than the qualified standard value, it means that the course goal is achieved, if it is less than the qualified standard value, it implies that the learning objectives of the course have not been achieved. This evaluation method is used in our teaching practice, and the assessment results obtained are reasonable, pertinent, consistent with the actual situation, and generally accepted by teachers and students.
5 Conclusion The construction of EEE courses in colleges and universities, which has attracted much attention from the education and related industries, has gradually entered the implementation stage. Compared with the traditional engineering, the “EEE” emphasizes more on the practicality, intersection and comprehensiveness of the discipline. Different from the traditional engineering talents, the new economy in the future needs high-quality composite EEE talents with strong engineering practice ability, innovation ability and international competitiveness. This forces colleges and universities to adopt reverse design to continuously promote the reform of education and teaching. Based on the OBE teaching concept, the research analyzes and studies the teaching content, teaching methods, teaching evaluation and goal achievement of the logistics information system course, and puts forward a project based teaching through and full cycle teaching management idea. The teaching content reflects the intersection and integration of multiple disciplines. The teaching method adopts an effective process management mechanism. The teaching evaluation has designed a multi indicator comprehensive evaluation system based on projects. The goal achievement degree analysis attaches the corresponding weight to each teaching link to evaluate the achievement of the curriculum goals. The teaching reform practice of the course has provided new ideas and laid a solid foundation for the training of engineering and technical talents in the new era. Acknowledgment. This paper is supported by: (1) Project of Century College of Beijing University of Posts and Telecommunications: “Research and Practice of Logistics Engineering Construction Based on CDIO Engineering Education Mode” (JSKY-1604); (2) Project of the Teaching Steering Committee of Logistics Management and Engineering Specialty in Colleges and Universities of the Ministry of Education: “Integration and Teaching Practice of Applied Undergraduate Technical Courses - Taking Logistics Information System as an Example” (JZW2022038).
References 1. Varekan, K., Renukadevi, S.: A review of knowledge management based career exploration system in engineering education. Int. J. Mod. Educ. Comput. Sci. 8(1), 8–15 (2016) 2. Liu, K., Chen, T.: On the governance of new engineering education. China Univ. Teach. 64(1), 37–41 (2020). (in Chinese) 3. Hadgraft, R.G., Kolmos, A., et al.: Emerging learning environments in engineering education. Australas. J. Eng. Educ. 5(1), 3–16 (2020)
786
Y. Liu et al.
4. Xie, B., Wu, L.: New engineering education construction integrated with regional development: logic, challenge and approach. China Higher Educ. Res. 6, 51–56 (2021). (in Chinese) 5. Korkmaz, Ö.: The effect of project-based cooperative studio studies on the basic electronics skills of students’ cooperative learning and their attitudes. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 10(5), 1–8 (2018) 6. Xu, P., Tang, Q., Liu, X.: Teaching reform of OBE oriented “Mechanical System Design” under the background of new engineering. Heilongjiang Educ. (Theory and Practice) 01, 91–92 (2022). (in Chinese) 7. Niu, G., Wei, W., Zhu, D., Tang, Y., Zhu, L.: Teaching reform and practice of storage and processing of fruits and vegetables curriculum based on OBE concept. Curriculum Teach. Methodol. 5(9), 20–24 (2022) 8. Zhang, X.: The application of flipped classroom in the marketing course based on OBE concept. J. Educ. Res. Policies 4(2), 56–58 (2022) 9. Zhang, W.: Design of capstone projects under CDIO mode. Int. Conf. Comput. Sci. Comput. Intell. (CSCI) 12, 849–852 (2019) 10. Xing, H., Xu, Z.: Construction and application of project driven one body, two wings and three stages teaching mode based on OBE concept. Int. J. Educ. 10(3), 29–37 (2022) 11. Luo, J., Xiao, S., Chen, L.: Research on the evaluation system of practical teaching quality based on OBE concept. Comput. Informatization Mech. Syst. 5(1), 22–24 (2022) 12. Mallikarjuna, B., Sabharwal, M., Kumar, P.: Research on the teaching reform of Python programming curriculum based on the OBE-CDIO concept. Front. Educ. Res. 5(11), 67–70 (2022) 13. Ma, C., Zhang, H., Zhang, S.: An inquiry into the teaching reform of programmingbasic curriculum based on OBE-CDIO concept. J. Higher Educ. 01, 89–91 (2020) 14. Dong, J., Li, Q., Peng, K., Cui, J., Lu, Y.: Research on evaluation methods for achieving curriculum objectives in the certification of engineering education. Higher Sci. Educ. 4, 121– 125 (2019). (in Chinese) 15. Li, M., Hongmin, L., Wei, Y., Kun, Q., Mei, X.: Research on quantitative evaluation of course goal achievement based on engineering education professional certification. Sci. Technol. Vis. 14, 72–74 (2021). (in Chinese) 16. Zhang, J.: Teaching design and implementation of design mode in the course of Java programming in higher vocational colleges. J. Phys. Conf. Ser. 1856(1), 1–7 (2021) 17. Herala, A., Knutas, A., Vanhala, E., Kasurinen, J.: Experiences from video lectures in software engineering education. Int. J. Mod. Edu. Comput. Sci. (IJMECS) 9(5), 17–26 (2017) 18. Faizi, A., Umar, M.S.: A conceptual framework for software engineering education: project based learning approach integrated with industrial collaboration. Int. J. Educ. Manag. Eng. (IJEME) 11(5), 46–53 (2021)
Teaching Practice of “Three Integration” Based on Chaoxing Learning Software – Taking the Course of “Complex Variable Function and Integral Transformation” as an Example Huiting Lu1 and Xiaozhe Yang2(B) 1 The Faculty of General Education, Nanning University, Nanning 530299, China 2 The Faculty of Intelligent Manufacturing, Nanning University, Nanning 530299, China
[email protected]
Abstract. It is necessary to explore more and more effective ways to cultivate talents with all-round development. “Three integrations”, that is, integrating the elements of application, innovation and moral education into the teaching of the curriculum, is a method that has been gradually recognized, accepted and valued. Taking the course of “complex variable function and integral transformation” as an example, according to the characteristics of application-oriented undergraduate students, this paper uses the information-based teaching tools in the new era, that is, relying on the Chaoxing Learning Software, combining with the teaching conditions of smart classrooms, to optimize the structure of classroom teaching, implement the “three integration”, highlight the cultivation of students’ application ability and innovation ability, and help students establish correct values, To solve the “pain point” problem of low classroom activity, low ideological and political integration and low professional integration of this course. Through research, we can gain experience that is worth learning and promoting, and better serve the training of professional talents. Keywords: Complex function and integral transformation · Innovative teaching · Chaoxing learning APP
1 Introduction It is a highly demanding work involving a wide range of factors for colleges and universities to cultivate high-quality compound talents with a solid foundation [1–3]. Complex variable function and integral transformation is a public basic course (or public compulsory course) for science and engineering majors in colleges and universities. It is a prerequisite course for power electronic technology, signal and system, automatic control principle, signal analysis and image processing, intelligent control and other courses. The complex variable function has its specific object theory, function operation skills, theory and calculation method, which provides an important analytical method for mathematics and is a further extension of the real variable function. The integral © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 787–800, 2023. https://doi.org/10.1007/978-3-031-36115-9_70
788
H. Lu and X. Yang
transformation is widely used in other disciplines and engineering fields, especially in physics, automation technology, communication technology, etc. [4]. The course focuses on cultivating students’ abstract thinking ability, spatial imagination, logical reasoning ability and scientific and technological ability, and is another important mathematics course after advanced mathematics [5–7]. However, the actual teaching hours of the complex variable function and integral transformation course in colleges and universities are compressed. Most colleges and universities have 48 class hours of this course, and some colleges and universities have compressed it to 32 class hours. However, the subject itself has abstract knowledge, various concepts, and strong theoretical nature. The calculation of many knowledge is relatively complex. Therefore, many engineering students have great difficulties in learning complex variable functions and integral transformations. If they want to truly master the subject and flexibly use knowledge to solve practical problems, it is a challenging task, which puts forward reform and innovation requirements for the teaching of the course [8]. At the same time, the construction of new engineering courses and the social demand for talents also make the school education work face new challenges [9–11]. In combination with the requirements of contemporary education for students’ ideology and morality and overall quality, the continuous improvement of the curriculum is also an urgent task to be completed [12–14]. Teaching reform has therefore become a topic of increasing concern in the education sector [15–17]. In order to solve the “pain point” problem of low class activity, low ideological and political integration and low professional integration in the course of “Complex Variable Function and Integral Transformation”, this paper explores the common problems encountered in the teaching of complex variable function and integral transformation [18, 19]. Through the implementation of “three integration” teaching based on the Chaoxing learning software, it explains the way to solve the problem, and shows some teaching results obtained in teaching practice, to represents the effectiveness of the teaching reform.
2 Three Major Problems in Course Teaching In teaching practice, the problems commonly existing in classroom teaching mainly include the following three aspects. 2.1 Lack of Activity in Large Class Teaching With the increase in the number of engineering majors and classes offered, it is difficult for the construction of the teaching team to follow up with the demand of teaching in the short term. The teaching of public courses in many schools has to be carried out in the form of large classes, which refer as a centralized teaching with a large number of students. This is the case with Nanning University, where the number of classes in which the complex variable function and integral transformation courses are taught exceeds 75. Large class teaching restricts the form of classroom teaching activities and makes it difficult to carry out effective discussions with the participation of teachers and students. Most of the teaching methods and means, such as group discussion teaching method, role playing and debate teaching method, are also difficult to operate in large class
Teaching Practice of “Three Integration” Based on Chaoxing Learning Software
789
classes, which leads to the lack of innovation in teaching activities, The participation of students is not high, and the teachers give lectures in the form of full talk, which makes the classroom atmosphere dull. This situation needs to be solved urgently. Under the existing teaching resources, innovative teaching is also imminent if we want to achieve the innovation of large class teaching activities and improve the enthusiasm of students to participate in classroom activities. 2.2 Low Integration of Teaching Content and Ideological and Political Education For many years, the integration of the teaching of college mathematics public foundation and ideological and political education elements is relatively low, mainly due to the following reasons: (1) The teaching content of mathematics courses has the characteristics of strong theoretical and abstract content, and it is difficult to extract the elements that significantly include “moral education”; (2) The teaching methods used by teachers when introducing ideological and political elements are inappropriate and easy to copy mechanically, resulting in insufficient guidance of students’ values and the lack of integration of ideological and political elements; (3) The teaching team has a superficial understanding of the connotation of the ideological and political construction of the curriculum, does not combine the requirements of the new era, has few new cases, and lacks novelty and attractive highlights. Taking the complex variable function and integral transformation course of Nanning University as an example, as the school’s model course of ideological and political education, certain teaching results have been achieved in the early stage, but the integration of ideological and political education is still not enough. Some ideological and political elements are only “cut in” added to the teaching content in a relatively rigid way, and have not achieved the effect of moistening moral education. Therefore, it is possible to achieve better results only by deepening theoretical research, improving teaching methods, strengthening moral education construction and summarizing practical experience. 2.3 Mathematics Teaching and Professional Application Derailed Under the background of professional certification of engineering education, it is required that the curriculum objectives be combined with the graduation indicators of the talent training program, which requires that the university mathematics curriculum be constructed in combination with the professional needs. Taking Nanning University as an example, complex variable function and integral transformation are offered to students majoring in electrical engineering and automation, communication engineering and robotics engineering. These majors focus on the knowledge points of the course, which is integral transformation, while the demand for conformal mapping is small. Therefore, the class schedule should focus on the second part of integral transformation, and combine the following professional knowledge to carry out teaching design. For engineering students, the tedious proof and deduction process in most mathematics
790
H. Lu and X. Yang
courses can be omitted, but the teaching of basic concepts, basic theories and basic methods must be strengthened. Mathematical reasoning depends entirely on basic concepts. The basic concepts are not clear, and many contents are not understood at all, and basic methods cannot be mastered, let alone correct application. Because the LBL (Lessonbased Learning) model currently used is a knowledge-based teaching concept, which ignores the connection between mathematical knowledge and practical problems to a certain extent. In addition, most of the mathematics teachers are researchers in the direction of mathematics, and there are more theoretical studies, while the understanding of professional knowledge is not deep enough. For them, it is difficult to design the teaching content according to the professional situation, which leads to the derailment of mathematics teaching and professional application. Therefore, we need to think about how to realize the intersection and integration of public basic courses and professional basic courses through teaching reform. While consolidating students’ mathematical foundation, we should also strengthen students’ mathematical application ability and break the barriers between traditional disciplines.
3 Practice of Innovative Teaching Faced with the “pain point” problem of low class activity, low ideological and political integration and low professional integration in the course of Complex Variable Function and Integral Transformation, the teaching team carried out innovative teaching practice in the teaching process of this course. 3.1 Innovative Thinking of “Three Integration” The course teaching uses Chaoxing Learning APP, an online learning software, to complete the moral education process from three different perspectives through the innovative idea of “three integration”, i.e., integrating application elements, innovation elements and ideological and political elements. (1) The integration of application elements in the curriculum requires teaching to break the barriers between basic courses and professional courses, and organically integrate professional knowledge into the original knowledge; (2) The integration of innovative elements in the curriculum requires the innovation of teaching content, which is based on and higher than the textbook, and is in line with the cutting-edge knowledge. It also requires the innovation of teaching mode to change the boring traditional classroom and make the classroom appear a new atmosphere; (3) The integration of ideological and political elements in the curriculum requires the implementation of moral education objectives in the whole process of teaching, and the “salt” of moral education into the “soup” of the curriculum. The innovative measures of “three integration” must be built on the premise of the original teaching objectives of the course. It is not necessary to mechanically copy the knowledge points in the professional courses and turn them into professional courses because of the need to integrate the application elements. Instead, it is necessary to
Teaching Practice of “Three Integration” Based on Chaoxing Learning Software
791
explain the teaching contents according to the professional characteristics and cultivate the students’ thinking of applying knowledge on the basis of the original knowledge; The emphasis on the integration of innovative elements should not deviate from the teaching content of the basic course, making the class become a paper report, and the teaching content is too difficult to exceed the requirements of the syllabus; Nor can it become a moral education classroom and deviate from the teaching objectives because of the consideration of ideological and political elements. In short, the “three integration” is inseparable, and can be generative integration, embedded integration, or migration integration at any stage of the teaching process. They are an organic combination of mutual influence and promotion. Under the teaching idea of “three integration” based on Chaoxing Learning, the teaching team adhere to the teaching concept of “OBE”, process the teaching content, cultivate and guide students with multiple teaching methods, improve teaching evaluation, assist the implementation of teaching methods, ensure teaching effects, and create efficient classroom. Figure 1 shows the basic idea of the project plan.
Fig. 1. Basic ideas of teaching practice
3.2 Processing Teaching Content Complex variable function and integral transformation is a highly theoretical and abstract subject for engineering students in colleges and universities, and it is also a common mathematical tool to solve engineering problems. This requires that the course needs to integrate teaching content. Math teachers should communicate with professional teachers to learn, and put knowledge points into application elements to meet the requirements of engineering construction. In other words, the teaching content is based on the “service professional curriculum”, adding the relevant professional application content in the corresponding knowledge points, expanding the breadth of content and digging the depth of knowledge, infiltrating professional ideas, not rigidly bound to textbook knowledge, solving the “pain point” problem of students’ mathematical knowledge and professional technology application derailment. At the same time, through the integration with professional knowledge, the teaching content is reconstructed, and the ideological and political elements of the curriculum are added. Let’s take a teaching section to show
792
H. Lu and X. Yang
how to integrate the “three elements” into teaching through the treatment of teaching content. 3.2.1 Add Innovative Elements Take the Fourier change in the course as an example to see the integration of innovation elements. First, it is needed to explore the physical meaning of Fourier transform: Example 1: Find the Fourier transform of rectangular pulse function f (t) = 1, |t| ≤ 0.5 and its amplitude spectrum and phase spectrum. 0, |t| > 0.5 On the platform of Chaoxing Learning, teachers and students work together to solve the answer, and then compare with an example in the previous lesson: the function 1, |t| ≤ 0.5 expression f (t) = of a function with a period of 2 in a period is to 0, |t| > 0.5 find its Fourier series, as well as its amplitude spectrum and phase spectrum. Figure 2 is drawn accordingly.
Fig. 2. Comparison between the spectrum of periodic function and that of non-periodic function
Integrate the original knowledge and methods with the current situation to form a new understanding and cognition. From concrete to special, help students better understand and master the physical meaning of Fourier transform, cultivate students’ limit thinking in mathematics, and develop innovative thinking ability. Then the continuous spectrum generation of a simple rectangular pulse signal function is shown in a dynamic diagram, as shown in the Fig. 3, so as to obtain the physical meaning of Fourier transform - the aperiodic function is composed of infinitely continuous harmonic waves of multiple frequencies.
Teaching Practice of “Three Integration” Based on Chaoxing Learning Software
793
Fig. 3. Time domain and frequency domain of rectangular pulse signal function
The characteristics of innovation are reflected through the integration, expansion and extension of knowledge through the above methods. 3.2.2 Add Application Elements According to the physical meaning of Fourier transform, the signal in the time domain can be converted into the frequency domain through this integral transform, so as to obtain the picture of amplitude and frequency correlation, as well as the picture of phase and frequency correlation. Although the signal is relatively intuitive in the time domain, it is not perfect enough, and it is difficult to study and process some problems, so the frequency domain is needed, which makes Fourier transform have a very broad role in signal processing. Here, we only study the Fourier transform of continuous aperiodic signals, while in engineering, Fourier transform has several forms shown in Table 1. Table 1. Form of Fourier transform in engineering Time function
Frequency function
Continuous and aperiodic Aperiodic and continuous Continuous and periodic
Aperiodic and discrete
Discrete and aperiodic
Periodic and continuous
Discrete and periodic
Period and Discrete
Expand knowledge, point out topics, arouse students’ attention to learning, and pave the way for subsequent learning of relevant professional knowledge. 3.2.3 Add Ideological Elements Add 5G-related news to the lecture. For example, China achieved full coverage of 5G at the 2022 Winter Olympics, reflecting China’s leading position in 5G technology, and building students’ national pride.
794
H. Lu and X. Yang
Through the processing of teaching content, it is a dominant way to integrate innovation elements, application elements and ideological and political elements into the teaching process, so we can directly experience the integration of these three elements. In addition, there are two implicit ways, that is, through teaching methods and means, and teaching evaluation to reflect the “three integration” discussed in this article. 3.3 Improve Teaching Methods and Means Because the teaching content of complex variable function and integral transformation is difficult, the concept is abstract and the calculation is large, the efficiency of the classroom is very low before the teaching reform. In this teaching practice, under the teaching conditions of the smart classroom, the teaching mode shown in Fig. 4 is adopted by using digital teaching equipment and combining with Chaoxing Learning Communication software.
Fig. 4. Teaching mode of the course
SPOC teaching relies on the Chaoxing Learning Connect platform to effectively integrate the existing curriculum resources and teaching resources, put learning videos, learning materials, courseware documents and other resources in the Learning Connect resource library, provide students with an online platform to acquire knowledge, and guide students how to learn and what to learn. For example, teachers can carry out “task driven” teaching by using the “group task” function of Learning Pass to cultivate students’ collaborative spirit, and use the “person selection” function of Learning Pass to carry out “discussion method” teaching, improve students’ courage, improve students’ communication and communication ability, and achieve the integration of “ideological and political elements”; Use the function of “classroom activities” to carry out “practice method” and “discussion method” teaching. The teacher throws questions on the platform, displays students’ learning results, and encourages students to actively apply knowledge to solve problems, that is, “integrate application elements”. At the same time, in the ideological collision with others, cultivate innovation ability, that is, integrate “innovation elements”; This “three integration” teaching method and means based on Chaoxing learning has greatly improved the participation and efficiency of classroom activities, and truly let students have a sense of gain.
Teaching Practice of “Three Integration” Based on Chaoxing Learning Software
795
The reform of teaching mode makes teaching activities dynamic and innovative. Large-class teaching with a large number of students can also have a highly participatory classroom, so that every student, even the students sitting in the last few rows, can obtain better learning results and make the classroom active. At the same time, the relevant data of the study conducted on the Chaoxing platform can be effectively saved, providing teachers with the original materials of the teaching process, and helping teachers better obtain the teaching feedback information. 3.4 Improve Teaching Evaluation Methods Teaching evaluation is mainly about the evaluation of students’ learning effect. The teaching evaluation will be diversified, the objective quantitative scoring and the subjective effect evaluation will be organically combined, and the students’ cognition, emotion, values and other contents will be included in it. The traditional evaluation method will be combined with the novel evaluation method, the result evaluation and the process evaluation will be combined, and the traditional evaluation method of 30% of the usual performance plus 70% of the final performance will be retained. The innovation will be made in the evaluation of the ordinary performance, making the evaluation system more inclusive It is feasible and easy for students and teachers to accept, help students complete learning tasks, and solve the problems of low teaching effect and low participation in classroom activities encountered by teachers in teaching. Among them, the result evaluation is mainly based on the students’ final paper performance, while the process evaluation is reflected in the evaluation criteria of their usual performance. Relying on the learning pass platform, independent pre-class preparation, participation in in-class activities, testing and after-class review will be recorded in the Chaoxing learning pass platform, and all will be given a certain score. The class score quantifies the participation and learning status of students in each activity, and determines the class performance, See Fig. 5 for details. The process evaluation criteria are divided into homework scores, attendance scores and classroom performance scores, in which the classroom performance scores are further subdivided. The calculation methods of class points are as follows: (1) Classroom practice: score (2–5 points) is set according to the difficulty of the exercise, and you can get points no matter whether you participate in it right or wrong; (2) Ask questions: students who can ask constructive questions will be given 4 points; (3) Answer: 3 points correct, 2 points partially correct, 0 point wrong; (4) Select the person to answer the question: no matter right or wrong, you can get points as long as you participate; (5) Group tasks: team leader 2–5 points, other members 1–4 points (score according to actual completion); (6) Awards in the competition: according to the situation of students’ mathematics awards, the first prize is 10 points, the second prize is 8 points, the third prize is 5 points, the first prize is 4 points, the second prize is 3 points, and the third prize is 2 points;
796
H. Lu and X. Yang
Fig. 5. Teaching evaluation structure
(7) Learning situation of general micro-class: self-study micro-class and score (1–5 points) according to the completion. The results of the teaching evaluation reflect the growth of students, and systematically reflect the degree of combination of knowledge transfer and value guidance, realizing the improvement of teaching effect through scientific evaluation. The results of teaching evaluation reflect the growth of students, and systematically reflect the combination of knowledge transfer and value guidance, which is conducive to the realization of the goal of improving teaching effectiveness through scientific evaluation.
4 Effect of Innovative Teaching 4.1 Improvement of Students’ Learning Autonomy Through the implementation of teaching reform and innovation in two demonstration classes (also known as experimental classes, or key class), students’ learning autonomy and interest in learning mathematics have been greatly improved, and the “pain point” problem of low participation of students in large class teaching activities has been solved, which is mainly reflected in the active classroom atmosphere. The number of students with class score above 76 points is up to 86%. The numerical distribution automatically generated by Superstar Learning is shown in Fig. 6.
Teaching Practice of “Three Integration” Based on Chaoxing Learning Software
797
Fig. 6. Distribution of class points in key Class 1 and key Class 2 of Electrical Engineering and Automation of grade 2019
Two months after the end of the course in December 2020, the background of the course shows that there are still students who continue to learn the content of the course by watching the micro-class, which also reflects the goal of teaching innovation to cultivate students’ lifelong learning ability. As shown in Fig. 7, there are still good learning records in February and March.
Fig. 7. The learning situation of the demonstration class two months after the end of the class
798
H. Lu and X. Yang
4.2 The Number of Students Winning Awards Has Increased Significantly The four classes of electrical engineering and its automation in 2019, as the demonstration class of “one solid and three integrated” innovative teaching, have the same overall number of students compared with the students of electrical engineering and its automation classes in 2017 and 2018. The final examination results of advanced mathematics are not significantly different. The three classes of students have the same math foundation, but the number of students participating in the national college students’ math competition has increased significantly, It shows that students’ enthusiasm for mathematics has increased after the curriculum reform, and the number of winners has increased significantly, from one winner to two winners and then to eleven winners, reflecting the success of curriculum innovation and construction of complex function and integral transformation. 4.3 The Students’ Learning Feedback is Constantly Praised The questionnaire survey shows that the vast majority of students approve of innovative teaching, and more than 80% of students are well accepted (as shown in Fig. 5). The hybrid teaching launched by Chaoxing Learning Communication embodies the teaching concept of “student-centered”. Students really learn knowledge from teaching, and feel the love and care of teachers . Finally, students should shape correct values from knowledge learning and realize the unity of value shaping and ability training in knowledge teaching (Fig. 8).
Fig. 8. Student acceptance survey of two experimental classes (light blue indicates acceptance and approval, dark blue refers to not accustomed)
5 Conclusion According to the characteristics of application-oriented undergraduate students, by processing teaching content, improving teaching methods and improving teaching evaluation, promote the “three integration”, i.e., integrating application elements, innovation elements and ideological and political elements into teaching, which can achieve the goal of cultivating more talents with all-round development. The teaching team, based on the Chaoxing Learning Communication platform, combined with the OBE (Outcomebased Education) teaching concept and, designed a set of pre-class, in-class and postclass teaching activities suitable for large class teaching, which effectively promoted the
Teaching Practice of “Three Integration” Based on Chaoxing Learning Software
799
implementation of “three integration” teaching. On the whole, the teaching reform is relatively successful, and the research of the project has achieved more results. Teaching and learning are mutually beneficial. While students progress, teachers also get a lot of gains. Some problems have also been encountered in the research. Because the teaching goal of “three integration” is a higher level goal, it is difficult for a few students with poor mathematical foundation to fully complete the two requirements of “application” and “innovation”. Therefore, it is necessary to further consider how to help these students better master knowledge. In this teaching practice, it is a pity that due to the limited team energy and the lack of communication between the teachers of the school’s interdisciplinary and inter-school, we have not further tracked the students’ learning of professional knowledge after class. In the next new teaching reform practice, it is hopeful to introduce teachers of different professional courses, so that we can observe the students’ learning situation and learning results on the whole front of talent cultivation. Acknowledgment. This project is supported by the project of 2022 Nanning University Educational reform (2022XJJG22) References.
References 1. Beltadze, G.N.: Game theory-basis of higher education and teaching organization. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 8(6), 41–49 (2016) 2. Singh, V., Dwivedi, S.K.: Two way question classification in higher education domain. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 7(9), 59–65 (2015) 3. Hamada, H.: Action research to enhance quality teaching. Arab World English J. Conf. Proceed. 1, 4–12 (2019) 4. Liu, C.: Curriculum reform of complex function and integral transformation in the context of engineering education. Xueyuan 11(18), 65–66 (2018). (in Chinese) 5. Zhang, Q., Li, H., Shi, K.: Exploration of teaching reform of complex variable function and integral transformation course for engineering majors. J. Higher Educ. 23, 120–123+126 (2018) (in Chinese) 6. Yu, R., Chen, M., Fei, X.: Exploration of teaching reform of complex variable function and integral transformation in the context of new engineering. Educ. Modernization 7(49), 34–36 (2020). (in Chinese) 7. Zhang, J.: On the importance of higher mathematics foundation in engineering basic education. Math. Learn. Res. 22, 8 (2018). (in Chinese) 8. Meng, G., Zhao, H., Li, X.: The exploration of complex variable function and integral transformation in engineering teaching. For. Teach. 11, 87–88 (2012). (in Chinese) 9. Ni, X.: Research on the integration path of general education and professional education under the OBE concept. Educ. Rev. 01, 48–55 (2020). (in Chinese) 10. Yan, D., Xue, J., Zhang, X.: Exploring the construction path of gold courses in higher vocational colleges based on OBE concept. Educ. Teach. Forum 604(01), 157–160 (2023). (in Chinese) 11. Lin, J.: China’s new engineering construction facing the future. Tsinghua Univ. Educ. Res. 38(02), 26–35 (2017). (in Chinese) 12. Lili, Y., Zhang, L., Fanbo, M.: Research on the cultivation of innovative talents in universities under the background of new engineering 2 Sci. Technol. Entrepreneurship Mon. 1, 87–89 (2018). (in Chinese)
800
H. Lu and X. Yang
13. Zhan, J.: Learn the “melting” formula of ideological and political education in higher vocational courses. China Education News, 2022–11–29 (in Chinese) 14. Lin, X., Xu, S., Liu, Z., et al.: Analysis of curriculum teaching reform under the background of engineering education certification. J. Higher Educ. 8(6), 136–138+142 (2022) (in Chinese) 15. Weiss, K.A., McDermott, M.A., Hand, B.: Characterising immersive argument-based inquiry learning environments in school-based education: a systematic literature review. Stud. Sci. Educ. 58(1), 15–47 (2022) 16. Liu, Z., Fei, J., Wang, F., Deng, X.: Study on higher education service quality based on student perception. Int. J. Educ. Manag. Eng. 2(4), 22–27 (2012) 17. Hui, C., Li, Z., Li, W., Mao, H.: Discussion on teaching pattern of cultivating engineering application talent of automation specialty. Int. J. Educ. Manag. Eng. 2(11), 30–34 (2012) 18. Robles, A.C.M.O.: The use of educational web tools: an innovative technique in teacher education courses. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 5(2), 34–40 (2013) 19. Al-Hagery, M.A., Alzaid, M.A., Alharbi, T.S., Alhanaya, M.A.: Data mining methods for detecting the most significant factors affecting students’ performance. Int. J. Inf. Technol. Comput. Sci. 12(5), 1–13 (2020)
Transformation and Innovation of E-Commerce Talent Training in the Era of Artificial Intelligence Lifang Su1 and Ke Liu2(B) 1 School of Economics and Management, Wuhan Railway Vocational College of Technology,
Wuhan 430205, China 2 School of Marxism, Wuhan Railway Vocational College of Technology, Wuhan 430205, China
[email protected]
Abstract. Artificial intelligence technology has changed the form and structure of social business services, and the relationship between technology and labor tends to be diversified. E-commerce in the era of artificial intelligence has transformed into the integration of Internet-based digital algorithms and professional knowledge, which requires workers engaged in e-commerce to have diversified knowledge of cross-border integration. In order to meet the needs of economic and social development in the era of artificial intelligence, the cultivation of ecommerce talents should be closely combined with the market. At present, the cultivation mode of business talents in colleges and universities mainly adopts school-enterprise cooperation, combined with the national 1+X certificate system. Through the investigation of vocational education and training evaluation organization, we found that the current Chinese commercial talent training still stays in the shallow mode of single and directed cooperation, which is not conducive to the change and transformation of commercial talent in the era of artificial intelligence. Therefore, colleges and universities should change the concept of enterprise talent training, innovate and develop the school-enterprise cooperation system, improve the commercial talent training system, and realize the deep integration of production and education. Keywords: Artificial intelligence · E-commerce · University-enterprise cooperation
1 Introduction As a prominent form of technology after “Internet+”, artificial intelligence is accompanied by “big data”, “informatization” and “digital”, “Economy”, “platform economy”, “intelligent manufacturing”, “virtual world”, “automation” and other keywords entered the academic and public vision. The emergence of new technologies has brought a new business model of information technology, big data technology, cloud technology, Internet of things technology and mining technology on the Internet as a platform [1]. In the intelligent era, traditional business must undergo digital transformation if © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 801–810, 2023. https://doi.org/10.1007/978-3-031-36115-9_71
802
L. Su and K. Liu
it wants to develop, which requires workers to have cross-boundary integration and multi-compound knowledge structure and skill level, so as to promote the improvement of system innovation capability. In the era of intelligence, the commercial form has reshaped the new job demand of modern business, and the cultivation of commercial talents has attracted much attention [2, 3]. To cultivate talents for the country and society, colleges and universities need to adapt to the different needs of social development in different periods. At present, the training of business talents should conform to the upgrading of skills and diversified transformation in the intelligent era, clarify the goal of talent training, and constantly enrich China’s high-level technical talents.
2 The Requirements for Business Talents in the Era of Artificial Intelligence and the Orientation of Their Training Transformation 2.1 Requirements for Business Talent in the Era of Artificial Intelligence The cultivation of business talents in the intelligent era faces various reform missions. On the one hand, the integration of complex knowledge structure. The “fusion” characteristics of business talents in the era of artificial intelligence are shown in the intersection of disciplines and the fusion of knowledge, that is, traditional business knowledge is integrated into intelligent technology and new methods of skills, such as collaboration and negotiation skills, data mining and analysis technology, intelligent optimization technology, collaborative decision-making, new business model operation method, Internet of things knowledge, etc. [4]. On the other hand, mathematical literacy has become a new requirement for workers and consumers. Under the condition of digital economy, digital literacy becomes the ability that workers and consumers should possess. With the continuous improvement of digital technology, artificial intelligence technology has penetrated into various fields, and “digitalization” has been upgraded to “digital intelligence” [5]. As a result, workers increasingly need to have “dual” skills: digital and professional. Therefore, having higher intellectual literacy has become an important factor for workers to win in the job market [6]. 2.2 Path Selection of Business Talent Training in the Era of Artificial Intelligence In the cultivation of vocational ability, teaching, learning and practice are inseparable, especially related to the cultivation of numerically intelligent business professionals [7]. In the era of artificial intelligence, technology is developing rapidly. University teachers have high business theory knowledge, but they are not as good as enterprises closely connected with the market for new technology knowledge such as artificial intelligence and big data [8]. Based on this, it is an effective way to cultivate digital intelligent business talents by connecting with industrial reality and deepening school-enterprise cooperation. The university-enterprise cooperation and production-education cooperative education have undergone three modes of evolution in China. The first model is a single, directed cooperation based on students’ going out. This model starts from the early
Transformation and Innovation of E-Commerce Talent Training
803
vocational education, which is characterized by student employment-oriented on-thejob practice [9]. Schools are active partners in cooperation, while enterprises usually do not have strong enthusiasm and initiative for cooperation because they have no initiative in the choice of internship time and interns. This model is not suitable for business majors in the intelligent era, because business majors cover loose and numerous enterprises and a certain enterprise cannot provide a large number of internship positions, resulting in low student management efficiency [10].The second mode is the platform mutual assistance and production-education docking introduced by enterprises, which is mainly reflected in the construction of training base “school in factory”. However, the business in the era of artificial intelligence there is no need for large-scale production lines and hardware facilities. This mode of school-enterprise cooperation is only carried out through entrepreneurship incubation projects, which cannot realize the cultivate on of talents with multiple intelligence. Third, the mode of co-education, sharing and strategic alliance with deep integration of industry and education. This is reflected in the deep involvement of the enterprise in the talent training of the university, the integration of artificial intelligence, big data and other skills into the setting, curriculum standards and learning practices of business majors, and the absorption of opinions from enterprises and industry experts [11–13]. Schools and enterprises began to focus on the construction of “virtual full simulation, full process” business work scenes.
3 Influencing Factors of School-Enterprise Collaboration in Cultivating E-Commerce Talents in the Era of Artificial Intelligence The orientation of application-oriented talents training in colleges and universities, combined with the cross-border knowledge accomplishment of e-commerce talents, requires that the training of business talents must rely on the market, industry and enterprise, and break through the single knowledge imparting of traditional education. The introduction of school-enterprise cooperative education mode is widely accepted by the commercial talents training of colleges and universities at the Age of Artificial Intelligence [14]. As new technologies such as artificial intelligence, automation and digital platforms change the skills of the service industry, skill-oriented scientific and technological growth and industrial transformation and upgrading in the intelligent era rely more on the optimization and upgrading of the labor skill structure [15]. These demands are driving the country’s transformation of higher education [16]. Currently, China is actively promoting the 1+X certificate system, and the pilot vocational skills areas are oriented to 20 skills shortage areas such as modern agriculture, advanced manufacturing industry, modern service industry and strategic emerging industries, so as to promote school-enterprise cooperation and train high-quality workers and technical skills personnel [17]. The certificates related to modern business talents include: e-commerce data analysis, financial intelligence, financial robotics, logistics management, etc. These certificates are closely related to artificial intelligence technology. In order to accurately understand the influencing factors of the current school-enterprise collaborative education of e-commerce students, so as to construct an exact and effective education model of new business
804
L. Su and K. Liu
students, we conducted surveys on enterprises, higher vocational teachers and students through online and offline interviews and questionnaires [18, 19]. 3.1 The Basic Situation of E-commerce Education in Higher Vocational Colleges Enterprises are an indispensable part of the new business education in higher vocational colleges. The participation and initiative of enterprises in the school-enterprise cooperation are related to the effect of the new business education. In this research, the enterprise as an important theme research. The specific data are as follows: 3.1.1 Establishment Year of the Enterprise Carrying Out School-Enterprise Cooperation and Education in New Business
Fig. 1. Carrying out school-enterprise cooperation
As shown in Fig. 1, Research is carried out from the years of school-enterprise cooperative education, there are two distinct types of school-enterprise cooperation experience, one is with more than 15 years of school-enterprise cooperation experience, the other is with less than 15 years of school-enterprise cooperation experience. Specific data are as follows: Enterprises with less than five years of school-enterprise cooperation experience account for 9.5% of the respondents; Enterprises with less than ten years of experience in school-enterprise cooperation account for 11.2% of the respondents; Enterprises with less than 15 years of experience in school-enterprise cooperation account for 10.06% of the respondents; Enterprises with more than 15 years of school-enterprise cooperation experience account for 69.24% of the respondents。 According to the above data, the foundation of school-enterprise cooperation in Chinese colleges and universities is well. A large number of enterprises have established the school-enterprise cooperation model of business talent training with universities. However, we also find that most of the school-enterprise cooperation has more than 15 years of experience, and most of the school-enterprise cooperation carried out is a single superficial school-enterprise cooperation. How to use Internet technology to develop the school-enterprise cooperation mode of business talent training in the age of digital intelligence is the direction that needs to be explored.
Transformation and Innovation of E-Commerce Talent Training
805
3.1.2 Enterprises Participate in the Goal Work Integrated Education Enterprise participation is the goal of cooperative education between New Business School and enterprises, among them, “provide skills standard services for the enterprise” accounted for up to 72.63%. The proportion of “reserving the skilled talents of the unit” and “expanding the influence of the unit in the industry” was 12.29% and 13.41% respectively. The proportion of profit-seeking was the smallest, only 1.68%. (As shown in Fig. 2).It can be seen that the purpose of most enterprises participating in schoolenterprise collaborative education is to achieve their own sustainable development by improving their service capabilities and influence.
Fig. 2. The Goal of school-enterprise collaborative education
3.2 The Development of Cooperative Education between New Business Schools and Higher Vocational Colleges 3.2.1 Number of Teachers Who Have Received Corporate Training through School-Enterprise Cooperative Education in Recent Three Years According to the survey data, in the school-enterprise collaborative education, in the past three years, 34.64% of teachers in higher vocational colleges have participated in enterprise training for five or more times, 6.7% for four times, and 13.41% for three times, as shown in Fig. 3.
806
L. Su and K. Liu
Fig. 3. Received corporate training
It can be seen that through school-enterprise collaborative education, teachers have significant benefits in acquiring cutting-edge technological knowledge of enterprises. 3.2.2 New Business Students Obtain Vocational Skills Certificates through SchoolEnterprise Cooperative Education Among the students in the higher vocational colleges surveyed, the certificates they have obtained now account for 44.69% of the total, while enterprise certificates and other types of certificates account for 21.23% and 24.58% respectively.( As shown in Fig. 4)What is noteworthy is that 9.5% of students obtained skills certificates through off-campus training institutions. This reflects the need for further cooperation between higher vocational schools and enterprises in the collaborative education of new business.
Fig. 4. Students obtain vocational skills certificates
Transformation and Innovation of E-Commerce Talent Training
807
3.2.3 Future Investment Plan of New Business School and Enterprise Collaborative Education As shown in Fig. 5, among the surveyed schools and enterprises, 70.39% said they would increase the investment, 21.79% said they would large investment, 6.15% said they would maintain the current situation, and 1.68% said they would reduce the investment. It can be seen that the school-enterprise collaborative education model is recognized by most enterprises and vocational colleges. But at the same time, it can also be found that the enthusiasm of both sides of the new business school and enterprise collaborative education is not particularly high, which is the direction of the new business school and enterprise collaborative education needs to be improved.
Fig. 5. Future investment plan
4 School-Enterprise Collaborative Education Strategy for E-Commerce Talents in the Era of Artificial Intelligence 4.1 Cultivate E-commerce Talents’ Philosophical Thinking Under the Background of Artificial Intelligence The innovation and development of colleges and universities cannot be separated from the support of enterprises in the industry and the shallow cooperation between colleges and enterprises can no longer meet the new requirements for the development of digitalized business. Under the background of the era of artificial intelligence, colleges and universities should establish the educational concept of student-centered development and cultivate business talents from various angles when training business talents. In the process of training, students’ compound knowledge structure is built, including enterprise digital production, management and marketing, etc. Colleges and universities should provide students with a good learning environment, necessary network conditions, the construction of information wisdom classroom, the establishment of commercial virtual simulation training room, to provide a number of intelligent commercial training base [20]. At the same time, colleges and universities should establish the integration mode of
808
L. Su and K. Liu
production, education and research, build an educational ecosystem for the cultivation of e-commerce talents, promote the deep integration of colleges and industries, and make education supply and enterprise demand seamless. 4.2 The Systematic Construction of “Central-Local” Legislative Hierarchy of School-Enterprise Collaborative Education To realize the integration of industry and education in vocational education, we need to overcome the disadvantages of management system and deal with the relationship between government, school and society, so as to promote the development of modern vocational education. In university-enterprise cooperation to promote vocational school, accelerate development to promote vocational education cooperation between colleges ordinance, clear the specific rights and obligations on both sides, make laws to education cooperation between colleges and more maneuverability. To be specific, first of all, we need to speed up the transformation of government functions. At present, the vocational education responsibilities of the Chinese government are quite numerous and the government needs to strengthen the macro-management, overall coordination and classification guidance of vocational education through overall planning, policy guidance and other means, as well as tax finance, financial payment and other levers. Secondly, we should promote the establishment of modern school system in vocational colleges [20, 21]. Vocational colleges need to formulate regulations that conform to the characteristics of running a school and can integrate various forces to actively participate in the management, so as to absorb all interested parties into the scientific management of the school. Finally, promote cross-border integration, introduce policies to encourage more “education-oriented enterprises” to participate in the development of vocational education, from tax incentives, direct financial subsidies, purchase services and other aspects, give qualified enterprises the legal status of “education-oriented” enterprises, vocational education organized by industrial enterprises into the vocational education system. 4.3 Improve the Social Training System and Realize Education and Social Equity With the deep application and integration of the Internet in the economic field, a large number of skilled workers pour into the platform economy model. A large number of workers who rely on the platform economy, such as take-out workers, express delivery workers and ride-hailing drivers, are facing job replacement and skill crisis. Their jobs are highly mobile, highly substitutable, and repeatable. The big data algorithm logic applied by artificial intelligence technology requires certain theoretical knowledge and operation skills, and a grasp of the integrity of the production process, so as to improve the autonomy and freedom of platform economy work [22, 23]. In the context of the development of artificial intelligence, e-commerce talent training is not only a transformation within the scope of universities, but also should pay attention to the skill education and labor protection of labor under the platform economy, and establish a training system for skilled labor that is adapted to the development of intelligence and digitalization. China’s 1+X certificate system needs to continue to explore and improve the skill training and vocational support system for low-end skilled labor, incorporate
Transformation and Innovation of E-Commerce Talent Training
809
the platform economy labor skill education into the vocational education system, play its role as a social fairness regulator, which is conducive to ensuring educational fairness and social justice, and forming a learning society.
5 Conclusion The transformation of economic structure and mode of production driven by skilloriented scientific and technological progress needs to continuously improve the skills of workers and pay attention to the transformation and upgrading of skills of workers with alternative skills. To cope with the rapid change of industrial transformation and labor market skill demand in the smart era, the construction and transformation of higher education system are of great importance. Under the diversified differentiation pattern formed by skills in the intelligent era, higher education, as the main body of skill supply, should adhere to the training goal of compound talents with “thick foundation and wide caliber”, and integrate new technology and traditional business theory. In the framework of macro system, the society and the university should form the comparative institutional advantage of the skilled society through system coordination and reform and innovation. Under the technological change with artificial intelligence as the core, higher education should be further integrated into the main economic battlefield of “artificial intelligence and industrial transformation”, transform empowerment into internal motivation, serve the society, cultivate the intellectual skills of workers, and build a lifelong learning society.
References 1. Abbasov, I.M.: Digital economy and digital education. Digit. Econ. Azerbaijan Int.Sci.-Pract. Conf. New Stage Econ. Develop. 6(11), 5–9 (2020) 2. Nye, J.: The power we must not squander. New York Times 1(1), 19 (2000) 3. UK Commission for Employment and Skills. Ambition 2020: world class skills and jobs for the UK (2009) 4. UNESCO Education strategy 2014–2021. UNESCO (2014) 5. Better Skills, Better Jobs, Better Lives: A strategic approach to skills policies. OECD (2012) 6. Department for Education, Department for business, innovation and skills. Rigor and responsiveness in skills 2013(4) (in Chinese) 7. Carnevale, A., Rose, S.J., Cheah, B.: The college payoff: education, occupations, lifetime earnings, p. 36. Georgetown University Center on Education and the Workforce, Washington DC (2011) 8. OECD: The Knowledge-Based Economy. Paris: Head of Publications (4) (1996) 9. Ouahbi, I., Darhmaoui, H., Kaddari, F.: Visual block-based programming for ICT training of prospective teachers in morocco. Int. J. Mod. Educ. Comput. Sci. 5(10), 56–64 (2022) 10. Karim, M.A., Masnad, M., Ara, Y., Nandi, A.D.: A comprehensive study to investigate student performance in online education during COVID19. Int. J. Mod. Educ. Comput. Sci. 2(3), 1–25 (2022) 11. Roy, S., Kabir, H., Ahmed, T.: Design and implementation of web-based smart class routine management system for educational institutes. Int. J. Educ. Manage. Eng. 12(2), 38–48 (2022) 12. Rigopoulos, G.: Student satisfaction, student assessment, student feedback. Int. J. Mod. Educ. Comput. Sci. 3(5), 1–9 (2022)
810
L. Su and K. Liu
13. Karim, M.A., Masnad, M., Ara, Y., et al.: A comprehensive study to investigate student performance in online education during COVID-19. Int. J. Mod. Educ. Comput. Sci. 14(3), 1–25 (2022) 14. Kyrpychenko, O., Pushchyna, I., Kichuk, Y., Shevchenko, N., Luchaninova, O., Koval, V.: Communicative competence development in teaching professional discourse in educational establishments. Int. J. Mod. Educ. Comput. Sci. 13(4), 16–27 (2021) 15. Aliyev, A.G.: Technologies ensuring the sustainability of information security of the formation of the digital economy and their perspective development directions. Int. J. Inf. Eng. Electron. Bus. 5(14), 1–14 (2022) 16. Aliyev, A.G.: Problems of regulation and prospective development of E-commerce systems in the post-coronavirus era. Int. J. Inf. Eng. Electron. Bus. 14(6), 14–26 (2022) 17. Mohdhar, A., Shaalan, K.: The Future of e-commerce systems: 2030 and Beyond. In: AlEmran, M., Shaalan, K. (eds.) Recent Advances in Technology Acceptance Models and Theories, pp. 311–330. Springer, Cham (2021). https://doi.org/10.1007/978-3-030-649876_18 18. Joma, H.N., Abdelazez, M.J., Mohammed, H.E., Ali, A.A.: Data-driven e-commerce techniques and challenges in the era of the fourth industrial revolution. Sci. J. Inform. 7(2), 291–302 (2020) 19. Ministry of Education. Ministry of education: National adult education strategies and implementation, Addis Ababa, Ethiopia, unpublished. http://bit.ly/2JO5Job (in Chinese) 20. Faizi, J., Umar, M.S.: A conceptual framework for software engineering education: project based learning approach integrated with industrial collaboration. Int. J. Educ. Manage. Eng. 11(5), 46–53 (2021) 21. Martin, J.G., López, C.L., Martínez, J.E.P.: Supporting the design and development of project based learning courses. In: Proceedings of IEEE Frontiers in Education Conference Proceedings, pp. 1–6. IEEE (2014) 22. Harms, S., Hastings, J.: A cross-curricular approach to fostering innovation such as virtual reality development through student-led projects. In: Proceedings of IEEE Frontiers in Education Conference, pp. 1–9. IEEE (2016) 23. Bender, W.N.: Project-based learning: differentiating instruction for the 21st Century. Corwin. 2(3), 26–29 (2012)
Talent Training Mode Based on the Combination of Industry-Learning-Research Under the Background of Credit System Reform Shanyong Qin and Minwei Liu(B) Shandong Women’s University, Jinan 250300, Shandong, China [email protected]
Abstract. The reform of credit system is the further deepening of educational reform. The flexible school system is conducive to the training of comprehensive talents, and promotes the reform of the talent training mode of industry-universityresearch cooperation education. This paper is based on combing the flexible training mode and characteristics of applied talents in the United States, Britain and Germany, and finding out the advantages of the flexible training mode of applied talents in foreign countries. Through the questionnaire survey, it is concluded that the talent cultivation mode of industry-university-research cooperation education under the credit system background is divided into two types: the mode based on the main body of industry-university-research cooperation and the mode based on the form of industry-university-research cooperation. This paper puts forward the problems that should be paid attention to in the context of the credit system reform from two dimensions: the flexible training of talents in the credit system reform and the training of talents in the industry-university-research cooperation education. This research is conducive to the innovation of talent training mode, promoting the teaching reform of credit system, and promoting the flexible training of application-oriented talents, which has strong theoretical guiding significance and practical reference value. Keywords: Credit system · Industry-learning-research cooperation education · Talent training · Flexible training curriculum system
1 Introduction In China, the credit system originated from Cai Yuanpei’s elective system at Peking University in 1918. By 1996, nearly one-third of colleges and universities in China had implemented and experimented with the talent training mode of the credit system. With the further deepening of educational reform, a good internal and external environment was provided for the credit system reform. With the implementation of the credit system, the curriculum has broken the boundaries of disciplines, and has been set as compulsory courses and interdisciplinary elective courses. When students choose courses independently, they realize the diversification and © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 811–821, 2023. https://doi.org/10.1007/978-3-031-36115-9_72
812
S. Qin and M. Liu
personalized development of training. The credit system is conducive to the talent training mode of industry-learning-research cooperation education dominated by cultivating students’ strong comprehensive theoretical knowledge and solving practical problems. Under the credit system, the cultivation of talents in industry-learning-research cooperation education should not only focus on the improvement of students’ learning ability and the mastery of basic knowledge of system theory, but also cultivate students’ strong technological innovation ability and adaptability to the future employment market and entrepreneurship market [1–4]. Based on the research on talent cultivation of industry-learning-research cooperation education under the background of credit system reform, the paper aims to clarify the connotation of industry-learning-research cooperation education under the background of credit system reform, identify its advantages and disadvantages, and propose corresponding strategies to deepen the reform of the credit system, so as to lay a foundation for the subsequent training of comprehensive and application-oriented talents.
2 Literature Review 2.1 Research on Credit System Brubac and John Seiler summarized the academic progress of American students and believed that American students could make their own choice of courses according to the price indicated in the credit. They could not only control the cost, but also learn the courses they liked. Mestenhuser and Breeder summarized the three credit system forms and characteristics of the United States, and believed that the American credit system includes semester, quarter and tri meter. The research of domestic scholars can be roughly divided into the following three stages [5–7]. In the first stage (1994–2000), there are many research literatures, but the research content is relatively shallow. In 1994, Chang Feng of Dalian University of Technology published his first article on the reform of the credit system in Higher Education Research. The title of the article is “On Student Work under the Credit System”. In the second stage (2001–2005), there were 308 articles on the credit system, and the research content developed in depth. Compared with the first stage, the research content in this stage is more extensive, and the discussion on model, system and system is gradually increasing. In the third stage (since 2006), the number of research articles on the credit system is increasing, researchers are also increasing, and the research content is more extensive and in-depth. After several years of practice in the credit system, all colleges and universities are more experienced in dealing with the credit system, and the research literature is more rational and mature, and can explore problems from a deeper level [8, 9]. 2.2 Research on the Cooperation Between Industry, University and Research Azaroff L V and Boyle K (1982) applied case analysis, taking universities as research objects, to study the characteristics of industry-learning-research. Becker and Peter (2000) found that the innovative scientific research model of school enterprise cooperation has improved the innovation vitality of enterprises and is conducive to the incubation
Talent Training Mode Based on the Combination of Industry
813
of patent achievements. Schmiedeber (2008) analyzed the practice of industry-learningresearch cooperation in German manufacturing industry, and found that there was a significant complementary effect between internal R&D and external R&D cooperation. Jirjah and Kraft (2006) found that independent R&D and collaborative innovation R&D of enterprises have significant complementary and substitutive effects under different control variables. Yang and Wang et al. (2008) pointed out that the contract spirit and trust relationship between the participants in the production university research cooperation would directly affect the stable development of the alliance relationship. Cheng Qiang and Shi Linna (2016) believed that industry-learning-research cooperation has its unique characteristics. Based on the perspective of partner heterogeneity, the study found that knowledge heterogeneity and relationship heterogeneity have a positive impact [10]. Wu Jun et al. (2016) showed that government R&D subsidies and enterprises’ technology absorption capacity have a steady and significant positive impact, and enterprises without the above characteristics have a steady and significant negative impact on enterprises’ innovation performance [11]. Lu Yanqiu and Ye Yingping (2017) believed that network practices were the key factors affecting the performance of industry-learning-research cooperation innovation. In the cooperation of industry, university and research, the influence of network practice on cooperative innovation performance presents an inverted U-shape, and inter organizational learning plays a significant intermediary role between network practice and cooperative innovation performance [12]. 2.3 Research Review at Home and Abroad Scholars at home and abroad have made fruitful achievements in the research on credit system teaching and industry-learning-research cooperation. The research on the basic concept, content, type and significance of the credit system and industry-learningresearch cooperation is relatively in-depth, in place and relevant, but there are still problems such as low research level, lack of professional theoretical support, less peopleoriented perspective and concept, and lack of systematic and comprehensive research. The research literature on the cultivation of talents in industry-learning-research cooperation education under the credit system background is few and not in-depth [13–15]. It is urgent to study the flexible educational system, course selection system, curriculum design, and practical teaching of the cultivation of talents in industry-learning-research cooperation education under the credit system background, and propose feasible plans to promote the talent cultivation plan and training process to a new level.
3 Analysis of Talent Training Mode Under the Background of Foreign Credit System 3.1 Characteristics of Talent Training Mode Under the Credit System in the United States The first is that the specialized courses of different majors in different colleges and departments are selected by students in this scope. The credits of compulsory courses in basic education require students to take courses from different disciplines. The second is that students still emphasize general education after choosing their major. The
814
S. Qin and M. Liu
third is that the university does not rigidly stipulate the specific length of schooling to complete their studies. Students can choose to postpone graduation or graduate ahead of schedule according to their own interests and life plans. Elective courses account for a large proportion, leaving more choices to students, with a large degree of freedom. Fourth, students can change their major at any time after selecting their major. Fifth, the knowledge of natural sciences, social sciences and humanities and social sciences should be equally emphasized. Sixth, tutor production is an important part of the credit system. Seventh, not only pay attention to professional courses and theoretical learning in the classroom, but also emphasize the development of practical ability and practical ability. The main features are summarized in Fig. 1.
The combination of centralization and decentralization
The combination of basic education and general education The combination of high flexibility of teaching plan and high freedom of course selection Characteristics of talent training mode under the credit system in the United States
The combination of dual major system and minor major system Pay equal attention to the knowledge of natural science, social science and humanities and social science The tutorial system was widely implemented Emphasize the cultivation of students' practical ability
Fig. 1. Characteristics of talent training mode under the credit system in the United States
3.2 Characteristics of Talent Training Mode Under the Background of British Credit System The first is that the duration of the bachelor’s degree in British universities and colleges is usually three years. The credits of the three academic years are interlinked and progressive, which is conducive to the scientific training and comprehensive development of students. The second is that the modularization of the curriculum design makes the curriculum structure independent and examinable, and the curriculum content comprehensive and comprehensive, which can reduce the students’ workload and realize the scientific and reasonable arrangement of the curriculum; The flexibility of the elective system is to provide a large number of optional courses and respect the students’ right to choose independently. The thirdly is that the duties of the tutor mainly include academic guidance and spiritual care, which not only guide students in learning and academic guidance, but also care for and care for students in life. The main features are summarized in Fig. 2.
Talent Training Mode Based on the Combination of Industry
815
The length of schooling Characteristics of talent training mode under the background of British credit system
Modular courses and course selection system The tutorial system
Fig. 2. Characteristics of talent training mode under the background of British credit system
3.3 Characteristics of Talent Training Mode Under the Background of German Credit System The first is that credits are divided into two stages. The study time of the first stage is mainly for the study of basic courses, and the corresponding credits of basic courses will be obtained after passing the course study assessment; The second stage of study is the study of specialized courses. The second is that the average student must accumulate 30 credits each semester to complete university studies within the standard study period. The third is that the length of study is usually four to six academic years. Each academic year is divided into two different semesters, winter and summer. It usually takes eight to twelve semesters to get a bachelor’s degree from a German university. Fourth, lectures, discussions and training are the three most common and basic teaching methods in German universities. In the lecture class, teachers teach and answer students’ knowledge, and students listen and record. The discussion class allows students to participate in class discussions and interact with teachers. The training course is to train students’ corresponding skills according to the content of the course. Fifthly, by recording and evaluating students’ majors, grades, teaching logs, degree application records, etc. before and after going abroad, the credits obtained before and after going abroad are converted and unified to achieve mutual evaluation and recognition of credits between different schools. The main features are summarized in Fig. 3.
The hierarchy of learning The Credit requirement Characteristics of talent training mode under the background of German credit system
The length of study
The form of teaching
The European Credit Transfer System
Fig. 3. Characteristics of talent training mode under the background of German credit system
816
S. Qin and M. Liu
4 The Classification of Talent Training Modes of IndustryLearning-Research Cooperative Education Under the Credit System and the Problems to be Concerned 4.1 The Classification of Talent Training Modes of Industry-Learning-Research Cooperative Education Under the Credit System In order to more accurately understand the talent training mode of industry-learningresearch cooperation education under the background of credit system reform, the project team used questionnaire survey to obtain data. In order to ensure the logic and preciseness of the questionnaire design process, the questionnaire design was carried out according to Gerbing and Anderson (1988)’s suggestions and in combination with the conclusions of the discussion with questionnaire design experts, including three stages: literature review, field interview and questionnaire prediction test. Then the questionnaire survey enters the sample data collection stage. The data collection scope of this paper involves universities, scientific research institutes, participating school enterprise cooperation enterprises and other subjects. See Table 1 for sample survey. Table 1. Valid Sample Questionnaire Factor
Category
Valid sample
Proportion
Gender
Male
156
54.17
Female
132
45.83
58
20.14
Industry Category
Colleges and Universities Scientific Research Institutes
35
12.15
195
67.71
Doctor
65
22.57
Master
51
17.71
Undergraduate
83
28.82
Junior College and Below
89
30.9
Under 28
28
Enterprise Education
Age
28–45 Personnel Classification
9.72
163
56.6
Over 45
97
33.68
Senior Management
26
9.03
Middle Manager
89
30.9
Grass Roots Management Personnel
65
22.57
108
37.5
Scientific Researchers
Talent Training Mode Based on the Combination of Industry
817
Through combing and summarizing the survey data, combined with the special lecture interview, the talent training mode of industry-learning-research cooperation education under the credit system background is divided into two categories: the mode based on the main body of industry-learning-research cooperation and the mode based on the form of industry-learning-research cooperation.
Rely on the model of colleges and universities
Relying on the main mode Typical model of talent training in industry university research cooperative education
University science park Discipline company National science and technology projects
Rely on project mode
Relying on Enterprise Projects Establishing a strategic alliance for technological innovation
Joint talent training mode
Co construction of postgraduate and postdoctoral workstations Order based cultivation Technology transfer mode
Traditional mode Based on the mode of industry university research cooperation
Entrusted R&D mode Joint research mode
Joint mode
Co construction mode
Cooperative development mode Co construction of scientific research base model Co construction entity mode
Fig. 4. Classification of talent training modes
The mode of relying on the main body of industry-learning-research cooperation includes the mode of relying on colleges and universities, the mode of relying on projects and the mode of jointly cultivating talents. The mode based on the form of industry-learning-research cooperation includes traditional mode, joint mode and joint construction mode. See Fig. 4 for specific classification. (1) University Science Park. University science park is a kind of joint transformation of scientific research achievements based on the technology of universities or scientific research institutions and the capital of enterprises. (2) Disciplinary companies. The main characteristics of discipline oriented companies are based on the advantages of disciplines, investing in scientific and technological strength, and then establishing a technology company with core technology, scientific research ability and development prospects.
818
S. Qin and M. Liu
(3) Relying on national science and technology projects. Rely on the scientific and technological innovation activities guided by the state and the government to carry out important key technologies, core strategies and national key projects. (4) Relying on enterprise projects. Relying on the enterprise project mode can not only improve the efficiency of scientific research achievements transformation, but also realize the training of talents. (5) Establishing the Strategic Alliance of Industry, University and Research Technology Innovation. This method not only realizes the advantage sharing of the main body of the industry-learning-research cooperation, but also improves the scientific research innovation level of enterprises. (6) Co construction of postgraduate and postdoctoral workstations. Universities and enterprises jointly send training personnel to train post doctors, and post doctors choose scientific research topics according to the actual situation of enterprises. (7) Order based cultivation. The method feature is that it can realize the supervision of enterprises on the talent training mode of colleges and universities, focusing on the needs of enterprises, and enterprises have high participation in the talent training process. (8) Technology transfer mode. The technology transfer mode is conducive to enterprises introducing world advanced technology and equipment, reducing the R&D cost of enterprises, and improving the production efficiency of enterprises. (9) Entrusted R&D mode. The entrusted R&D mode refers to the behavior that the entrusted party conducts research on specific projects according to the requirements of the entrusting party and obtains scientific research achievements at a given time. (10) Joint research mode. Joint research projects usually take research projects as a bridge and subject groups as a support. Each party of industry-learning-research cooperation sends representatives to form a temporary team to carry out research and development of scientific research projects. (11) Cooperative development mode. The cooperative development mode is generally based on scientific research projects or topics, according to market demand or government driven, universities and enterprises cooperate to complete scientific research projects or solve scientific and technological problems. (12) Co construction of scientific research base model. The mode of co building scientific research base refers to the behavior that enterprises, scientific research institutes and universities invest a certain amount of funds, equipment and other resources to jointly build scientific research bases. (13) Co construction entity mode. The co construction entity mode refers to the mode in which all parties of industry, university and research form a research and development entity by investing capital, manpower, equipment, technology to carry out technology development or technology management. 4.2 Problems that Should Be Paid Attention to in Flexible Talent Cultivation Under the Background of Credit System Reform Members of the project team conducted in-depth interviews with universities, scientific research institutions and enterprises to understand the problems in the talent training mode of industry-learning-research cooperation education under the background
Talent Training Mode Based on the Combination of Industry
819
of the credit system reform, and then put forward operable and feasible suggestions, and divided the problems that should be concerned about in the talent training mode of industry-learning-research cooperation education under the background of the credit system reform into two categories. See Fig. 5 for the specific classification. Form a student-centered training mechanism Form the motivation mechanism of students’ learning
Problems that should be paid attention to in the cooperative education of industrylearning-research under the background of credit system reform
Flexible cultivation of talents in credit system reform
Form an optimization mechanism for professional construction Form the competitive incentive mechanism for teachers' teaching Reflect the diversity of majors Improve the efficiency of talent training
Cultivation of talents in industryLearningResearch cooperation education
Focus on students' practice and experiential education Pay Attention to Cultivating Students' Innovative Ability and Spirit Cultivate talents based on the advantages of cooperative main body and resources Attach importance to strengthening the support and guidance role of the government
Fig. 5. Problems that should be paid attention to in flexible talent cultivation
1) Form a student-centered training mechanism. By implementing the credit system reform, students can choose their own courses and teachers, which fully respects the wishes of students, their interests and hobbies, and is conducive to students’ personality. 2) Form the motivation mechanism of students’ learning. The reform has been promoted to give students the right to choose. The school allows ordinary undergraduate students in the first and second grades to change their majors once in school, advocates students’ independent design and free choice, fully arouses their enthusiasm for learning, and forms a learning motivation mechanism. 3) Form an optimization mechanism for professional construction. The reform of teaching measures plays a direct role in promoting the professional construction of the school, and also provides an opportunity for the adjustment and optimization of the professional structure, promoting the overall professional construction of the school and improving the level of teaching management. 4) Form a competitive incentive mechanism for teachers’ teaching. Students choose courses and teachers independently, which has an internal competitive incentive effect on teachers’ teaching. Teachers’ teaching responsibilities have also been greatly enhanced, teaching energy has been invested more, updating teaching content and reforming teaching methods are more active and self-conscious than in the past.
820
S. Qin and M. Liu
5) Reflect the diversity of majors. The cultivation of talents in industry-learning-research cooperation education is more diversified. The cultivation of talents in industry-learningresearch cooperation education is more diversified. With the development of society, economy and technology, there is an increasing demand for applied talents and compound talents.. 6) Improve the efficiency of talent training. In order to avoid the waste of resources between students and schools, and improve the efficiency of talent training in industrylearning-research cooperation education, the implementation of the credit system must strictly limit the flexibility of learning time. 7) Focus on students’ practice and experiential education. Through the real situation mode, it lasts for a certain period of time to provide innovative training and practical experience for its participants, thereby improving their professional quality. 8) Pay Attention to Cultivating Students’ Innovative Ability and Spirit. The industrylearning-research cooperation mainly focuses on the ability of the participants to raise, analyze and solve problems. Only when these abilities are improved, can the innovation ability achieve a qualitative leap, and thus improve the high compatibility between the talent training objectives of universities and the talent needs of enterprises. 9) Cultivate talents based on the advantages of cooperative main body and resources. Enterprises create talent training bases by virtue of the industry-learning-research cooperation model. Universities and scientific research institutions use scientific research advantages to promote academic exchanges and achievements transformation. 10) Attach importance to strengthening the support and guidance role of the government. The government has the behavior effect of management, organization, coordination and promotion, and can promote the development of this cooperation mode through perfect, supporting and reasonable policies and measures.
5 Conclusion The main conclusions of this study are as follows: 1) On the basis of the analysis of the talent training mode under the credit system background of the United States, Britain and Germany, this paper expounds two typical types of industry-learning-research cooperation education modes based on the principal mode of industry-learning-research cooperation and the form mode of industry-learning-research cooperation. 2) The mode of relying on the main body of industry-learning-research cooperation includes the mode of relying on colleges and universities, the mode of relying on projects and the mode of jointly cultivating talents. 3) The mode based on the form of industry-learning-research cooperation includes traditional mode, joint mode and joint construction mode. 4) The value of the flexible training mode of applied talents under the background of the credit system reform includes six aspects: forming a student-centered training mechanism, forming a dynamic mechanism for students’ learning, forming an optimization mechanism for professional construction, forming a competitive incentive mechanism for teachers’ teaching, the diversity of professional settings and the efficiency of talent training.
Talent Training Mode Based on the Combination of Industry
821
Acknowledgment. This project is supported by 2017 Teaching Reform Research Project of Shandong Women’s University: Research on talent training mode of Industry-learning-research education mode under the background of credit reform. 2022 Jiajiayue Shandong Women’s College demonstrative off-campus practice base construction project.
References 1. Veugelers R., Cassiman B.: R&D cooperation between firms and universities. Someempirical evidence from Belgian manufacturing. Int. J. Industrial Organ. 23(5–6), 355–379 (2005) 2. Pierre, V.: Identifying collaborative innovation capabili ties within know ledge-intensive environments: insights from the ARPANET project. Eur. J. Innov. Manag. 1, 152–155 (2012) 3. Zhang,Y.: Challenges and countermeasures for college class management under the credit system. Henan Educ. High. Educ. 2, 44–47 (2018). (in Chinese) 4. He, Y., Liang, F.: Research on influencing factors and mechanism of enterprise knowledge search in industry-learning-research cooperation. Sci. Sci. Manage. Sci. Technol. 03, 12–22 (2017). (in Chinese) 5. Yu, L.: Research on the “cooperative education” model cultivating in higher vocational education. Int. J. Educ. Manage. Eng. (IJEME). 1, 35–41 (2012). (in Chinese) 6. Wang, P., Wang, H.: A comparative study on the innovation network of textile university industry-learning-research cooperation based on knowledge graph. J. Zhejiang Sci-Tech Univ. (Social Science Edition) 48(05), 513–525 (2022). (in Chinese) 7. Luo, S.: Research on the cultivation of innovative undergraduate talents under the credit system. Anhui University, Hefei (2010). (in Chinese) 8. Khine, P.T.T., Win, H.P.P., Naing, T.M.: Towards implementation of blended teaching approaches for higher education in Myanmar. Int. J. Educ. Manage. Eng. (IJEME), 1(11), 19–27 (2021) 9. Hu, N., Wan, J.: Construction of higher vocational talents training program under the full credit system. Vocat. Educ. Forum 18, 75–78 (2015). (in Chinese) 10. Cheng, Q., Shi, L.: Research on the co-evolution mechanism of industry-university-research collaborative innovation based on self-organization theory. Soft. Sci. 30(04), 22–26 (2016). (in Chinese) 11. Wu, J., Zhang, J., Huang, D.: Research on the impact of industry-university-research cooperation on the innovation performance of strategic emerging industries. Contemp. Finan. 09, 99–109 (2016). (in Chinese) 12. Lu, Y., Ye, Y.: The impact of network practices on innovation performance in industryuniversity-research cooperation. Sci. Res. Manage. 38(03), 11–17 (2017). (in Chinese) 13. Yang, X.: Research on entrepreneurship education and innovative talent training in Chinese universities. China High. Educ. Res. 01, 39–44 (2015). (in Chinese) 14. De Matas, S.S., Keegan, B.P.: A case study on adult and workplace learning. J. Educ. Manage. Eng. 1, 11–19 (2020) 15. Ye, F.: Credit system realization mode and talent cultivation of industry-learning-research cooperation education. Res. High. Eng. Educ. 6, 106–111 (2015). (in Chinese)
Analysis of the Innovation Mechanism and Implementation Effect of College Students’ Career Guidance Courses Based on Market Demand Jingjing Ge(B) School of Mechanical and Electronic Engineering, Suzhou University, Suzhou 234000, China [email protected]
Abstract. The negative impact of the post-epidemic era on economic development is gradually emerging, and the current employment situation is severe and complicated. As a course focused on college students’ career development and employment, there are many problems in the actual teaching process, such as incoherent curriculum, tedious teaching content and lack of practical links. This paper systematically analyzes the practical significance of the innovative research of the employment guidance course for college students, the problems and difficulties in the current employment guidance course, and how to innovate the lecture form, course content and evaluation methods of the employment guidance course from the market perspective and students’ needs, so as to help college students crack the employment dilemma, clarify the employment goals and achieve more satisfactory employment. And through the comparison of course effects, the significant effect of the findings of this paper on improving college students’ employment readiness is clarified. Keywords: Career Guidance · Innovation · Market
1 Introduction Many scholars have made a lot of research results on career guidance and career development, Wang conducted a study on the innovation of career guidance model in China in 1998, Sun et al. summarized the implementation of career guidance in China during 2001–2012 and gave targeted suggestions, Qin analyzed the situation of students with employment difficulties in electronics and provided guidance suggestions [1–3]. Zhang gave an effective method to optimize career planning for undergraduate students, Liu proposed to establish a perfect career guidance model [4, 5]. Watts A G et al. compared the similarities and differences of employment policies in several countries [6]. Grubb W N emphasized the mutual influence of government and market on employment information, Stukalina, Y pointed out the importance of career management for technical universities, Kuzmin A M discussed how to strengthen the training of engineers based on a project approach [7–9]. Vaidya N M developed an effective model for predicting the © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 822–832, 2023. https://doi.org/10.1007/978-3-031-36115-9_73
Analysis of the Innovation Mechanism and Implementation
823
overall performance of students, El Mrabet H proposed that schools can provide good employment programs for students with the help of technology [10, 11]. Thus, it can be seen that career guidance and career planning are important for personal growth and social development. In the post-epidemic era, the negative impact of the epidemic on the economic development has been gradually highlighted, and the employment situation of college students is unprecedentedly severe and complicated, while employment is a two-way choice based on market demand and college students’ employment values. Therefore, colleges and universities need to change their thinking mode when providing employment guidance and training to college students, actively obtain the new changes of current society and enterprises for talents, and integrate them into the whole process of employment guidance and career planning, so as to truly realize the unity of talent cultivation, scientific research and social service of colleges and universities, and accurately help college students to find high-quality employment.
2 The Practical Significance of the Innovation of Career Guidance Courses for College Students 2.1 Be Beneficial to Talent Reserve for Industrial Technological Revolution Employment is the largest and most basic livelihood, the 20th Party Congress report pointed out: strengthen the employment priority policy, improve the employment promotion mechanism, and promote high-quality full employment. This is not only a longterm plan made by the state at the strategic level, but also a realistic need of today’s rapid technological update and industrial revolution. New industrial changes and technological development have broken the boundaries of traditional industries, and more industrial forms and employment models are emerging. This is not only a new employment opportunity that contemporary college students need to seize under the severe economic situation, but also a new challenge to the employment quality and career psychology of college students. The so-called new employment mode refers to more employment positions derived from artificial intelligence, big data, Internet of Things and 5G technology, which have distinctive requirements of high knowledge and strong technical reserve, while the professional course education received by contemporary college students in the university stage focuses on basic and systematic knowledge inculcation, with strong theoretical but lack of practical ability, and innovative thinking also need to be strengthened [12]. Therefore, colleges and universities should focus on cultivating students’ industry foresight and ability, through the organic connection of career guidance courses, based on the skills education of professional courses, helping students understand the latest development trend and technological innovation of the industry, the new job demands brought by them, and the accumulation of knowledge and ability required by these jobs. Only then will students have a sense of identity and belonging in their hearts, consciously and spontaneously examine the gap between themselves and their goals, reasonably set career goals, gradually refine their career plans, strive to improve their professionalism, jointly promote the development and progress of society, and become high quality labor talents.
824
J. Ge
2.2 Be Beneficial for College Students to Achieve Higher Quality Employment According to the statistics of the Ministry of Education, the number of graduates in China is gradually climbing and the number is estimated to reach 11.58 million in 2023, breaking the 10 million mark again. In the past two years, the phenomenon of “slow employment” and “delayed employment” of college students has been gradually highlighted, and many scholars have analyzed and researched this phenomenon and put forward corresponding countermeasures. A search on China National Knowledge Infrastructure with the keyword of “slow employment” shows that there are as many as 637 relevant research articles and more than half of them were published after 2021, this fully indicates that the employment motivation of contemporary college students needs to be enhanced, and the establishment of correct employment values in the post-epidemic era can help graduates achieve high-quality employment [13]. This includes both the influence of college students’ self-selection mentality and the influence of external environment on college students’ employment behavior, which should never be ignored. The university students are limited by the barrier of information acquisition ability and the lack of social experience, so it is difficult for them to have accurate insight into the recruitment psychology of enterprises and the changes of talent demand. This requires universities to build a strong communication bridge, provide a wide range of information channels and give accurate graduation support, so that graduates have a rational and calm employment mindset based on a full understanding of the job market and prospects. 2.3 Be Beneficial for Efficiency Improvement of Enterprise Human Resources In the process of recruiting talents, enterprises need to spend a lot of manpower, time and money costs, so they hope that the comprehensive skills of the graduates recruited are better matched with the positions, which can help the efficiency of talent recruitment and thus reduce human resources costs. The reality is that the “structural difficulties” in employment have existed for a long time and have become the main reason for the current difficulties in employment of college students. On the one hand, graduates facing seemingly “massive” recruitment information is still difficult to find a satisfactory job, on the other hand, enterprises have paid a lot of human cost but cannot recruit talents that match the job demand. Solving the “structural difficulties” of employment requires the joint efforts of the government, society, enterprises, universities and individual students, colleges and universities should meet the needs of enterprises in the whole process of talent training, examine the shortcomings of the current curriculum setting, teaching methods, evaluation and feedback and carry out targeted reforms in order to really improve the quality of talent training, so that graduates have a solid reserve of professional skills and good employment competitiveness [14]. Therefore, career guidance courses for college students should be the leading place for reform and innovation, thus driving other courses to gradually integrate into the system and forming a synergistic mechanism of extensive consultation, joint construction and sharing [15].
Analysis of the Innovation Mechanism and Implementation
825
3 The Common Problems Existing in the Current Career Guidance Courses for College Students 3.1 Insufficient Attention from Colleges and the Curriculum is not Systematic Although college students’ career guidance course is included in the talent training program of most colleges and universities as a compulsory general education course, it often faces many problems in the actual teaching process, such as the course opening time is arbitrary, the teachers are not professional, and the career planning and career guidance courses are not coherent. For example, some colleges and universities put the career planning course in the first semester of freshman year, while the career guidance course is not reopened until the second semester of junior year or even the first semester of senior year, when students are already facing the key choice of choosing career and further education [16]. Therefore, colleges and universities should appropriately consider moving the career guidance course forward or in the form of short-term classes of four to six hours per semester to enhance the relevance and consistency of career guidance, especially in the sophomore year, when students have just passed the confusion stage of freshman year and have certain knowledge about the prospect and knowledge structure of their majors, but they still have many questions about how to better combine their strengths with their majors, so they need constructive guidance and advice from experienced professional teachers. Furthermore, most of the career planning and career guidance courses in colleges and universities are taught by young counselors or class teachers, many of whom are just going from college to college and are very young as career professionals themselves, their classes are more in accordance with the textbook content of theoretical knowledge, while the career guidance course is precisely a very practical and application-oriented course [17]. Therefore, colleges and universities should focus on coordinating internal and external resources, optimizing faculty allocation, and enhancing students’ sense of experience and access to the course. 3.2 The Teaching Content is Boring and not Very Practical Most of the current career guidance courses are still presented in the traditional classroom teaching style, and the course content is set up in a similar way, generally divided into basic parts such as employment policy explanation, employment psychological counseling, resume making, interview process and etiquette, labor relations establishment, etc. Students are not interested in facing the boring theoretical indoctrination, and the classroom effect is not ideal, and the applicability is even worse in the actual job search process. Graduates in the first step out of the ivory tower job search need to determine their own job search goals, followed by the need to establish a confident and generous job search mentality and a positive job search image, and then adequate interview preparation and appropriate interview skills. Therefore, colleges and universities can re-integrate and re-order the content of the traditional career guidance courses, and after students have been guided by the career planning course to have a preliminary understanding of their own personality, interests, hobbies and career selection factors, etc. in the early stage, combining with the current employment prospect of this profession
826
J. Ge
to help graduates to reasonably choose the intended enterprises and positions, so as to effectively avoid the result of “fishing in the sea” and eventually getting nothing. The career guidance department of colleges and universities can also regularly invite the personnel department of large-scale enterprises in the industry to open lectures or forums and other forms, from the perspective of employers to give students professional interview image and job search skills advice, which is far more than the lectures in the classroom away from the actual job search scene to bring students more sense of job search reality and a sense of self-acquisition, students will also have the motivation to self-improvement and progress. At the same time to link up with the professional power of the counseling department, in advance to teach students the skills and ways to relieve anxiety and overcome nervousness, to avoid students have experienced a number of unfavorable blow to the job search, and then to seek psychological counseling and help, this can very easily have a negative impact on the later job search process, and research results have shown that psychological education taken in modern college students’ career guidance can further improve their mental health [18]. 3.3 Large Class Teaching, Lack of Individual Guidance Unlike ordinary general education courses that are suitable for large classes, the purpose of the career guidance course is to enable everyone to master the skills applicable to their own job search, which requires the teacher to pay attention to the subjective motivation and receptivity of each individual student in the process of teaching, in addition to the general theoretical knowledge [19]. In contrast, colleges and universities are limited by the lack of class time and faculty, and career guidance courses are generally taught in large classes. Take electronic information major, where the author once taught, for example, the standard class size is about 60 people. No matter it is simulated one-toone interview scene or leadless group interview discussion, 60 people are obviously not suitable to carry out similar activities at the same time, including the classroom venue is also difficult to simulate the real interview environment. Therefore, colleges and universities should consider opening career guidance courses in small classes and establishing career profiles for each student, so that each student can have a clearer understanding of their own employment advantages and shortcomings through theoretical learning and practical activities of the courses, so that they can target to enlarge their advantages and make up for their shortcomings, and gain knowledge and ability accumulation that can really benefit the future job search process.
4 Analysis of the Innovation Mechanism of College Students’ Career Guidance Course Based on Market Demand 4.1 Help College Students Broaden Their Horizons and Develop a “Strong Heart” for Employment A good employment mentality is a prerequisite for college students to achieve high quality employment. The graduates of 2023 are generally the post-00 s generation. They grew up in the two decades of China’s rapid economic development. Their parents
Analysis of the Innovation Mechanism and Implementation
827
provided them with good material conditions and rich experience, which shaped them to have a strong sense of self, strong subjective initiative and strong will to realize themselves. But at the same time, most of them are the only child, and they are easy to show multiple problems such as poor teamwork, weak anti-pressure ability and weak empathy. This is bound to cause their job hunting and career selection behavior will exist double contradiction. The author once conducted a simple test on 120 students in two classes of electronic information major, which asked what are the first factors you are most concerned about in the process of determining your ideal career? The complete statistics result of factors that affects graduates showed as Fig. 1. Among them, more than 80% of the students put salary in the first priority, followed by career prospects, company scale accounts for 6.2%, 4.9% cares about working site most, which also fully shows that contemporary college students in the mentality of choosing a career is impatient, the vision is narrow. 82.0%
6.9%
Salary Package
Career Prospects
6.2% Company Scale
4.9% Site/Distance
Fig. 1. Factors that affect graduates’ choice of career
Engels said in Dialectics of Nature: “The development or negation of negation caused by contradiction – the spiral form of development. The same is true of career development. Spiraling means that no matter which career development path they choose, they will experience both twists and turns and progress. College career guidance courses can try to help graduates rationally view the spiraling progress of their career through the following three methods. Only when graduates have a clear judgment and rational cognition of employment goals can they obtain a “strong heart” in employment. Method 1: Invite outstanding student representatives who have already developed in the industry to come into the class and exchange their experiences, which can increase students’ sense of inclusion. Method 2: Make full use of opportunities such as social practice, internship and training in winter and summer vacation to encourage students to enter local enterprises to assume certain work roles and experience the real world of work. Method 3: Make good use of new media technology to provide accurate employment services for college students, promote the spirit of labor in a way that students are willing to accept [20].
828
J. Ge
4.2 Promote Graduates to Improve Themselves and Have a “Voice” in Career The ultimate goal of talent recruitment is to create economic value for the enterprise. Therefore, if graduates want to stand out in the cruel job market and get the ideal job opportunities, they must “cultivate both inside and outside” and constantly improve their comprehensive quality. At present, a common problem of career guidance courses in colleges and universities is that the teaching content is divorced from reality, and it is difficult for students to acquire practical career skills through the course learning. This is caused by many factors. First, colleges and universities generally do not pay enough attention to this kind of course. There is no corresponding teaching content setting and no assessment requirements, which results in teachers ignoring the cultivation of students’ practical skills when teaching. On the other hand, college teachers are far away from the real workplace environment, and they are not familiar with what practical skills the enterprise needs in addition to professional knowledge. According to a survey the author ever did on 120 students in two classes of electronic information major (there are exactly 60 students in each class), students desire to acquire a variety of career skills and the detailed data showed in Table 1 and Fig. 2. From Table 1, we see that the skills students care about most is career goals setting, total 44 out of 120 students go to this option, followed by office skills guidance. According to Fig. 2 we can find this survey results place resume making skills as being 16.7% in all options, with 14.2% concern interpersonal relationship most, while 7.5% choose interview experience and ability and the reminder goes to labor rights related content. Table 1. Statistics of different skills that students desire to acquire in career guidance courses Skills
Number in class1
Number in class2
Total number
Career goals setting
23
21
44
Office skills guidance
11
14
25
Resume making skills
8
12
20
Interpersonal relationship
9
8
17
Interview experience and ability
4
5
9
Labor rights related
2
3
5
7.5% 4.2% 14.2%
36.7% 16.7%
20.8%
Career goals setting Office skills guidance Resume making skills Interpersonal relationship Interview experience and ability Labor Rights Related
Fig. 2. Percentage of demand for different skills
Therefore, the career guidance courses of colleges and universities need to strengthen the in-depth cooperation with enterprises, pay attention to the recruitment points of
Analysis of the Innovation Mechanism and Implementation
829
different positions set by employers and students’ demand for different skills, then make up for it in the course teaching process [21]. Method 1: Contact long-term cooperated internship units or intern employment bases to ask for the professional human resources assistance, they can provide practical office skills guidance for students to help them improve their “hard strength”. Method 2: Invite enterprises to participate in the simulated recruitment competition held by schools or colleges, creates a real environment for recruitment and interview, sets up on-site recruitment links, so that graduates can really find their own gap and find the goal of efforts through the simulated recruitment competition. Method 3: The university coordinates with the human resources security department to hold regular lectures on issues related to the interests of graduates, such as the signing of labor contracts, the protection of labor rights and interests, and how to protect the legitimate rights and interests of graduates through legal channels in case of labor disputes [22]. These are precisely the general knowledge of the workplace that graduates must have before entering the workplace, and the knowledge reserve that universities should build up for graduates through career guidance and education. Only with solid knowledge and skills and a clear understanding of the rights and obligations of workers can graduates master the workplace “voice”. 4.3 Whole Process and Whole Staff Participation in Talent Cultivation, to Create a Sense of Professional “Atmosphere” Most of the career planning and career guidance courses currently offered in colleges and universities are taught by student work managers or administrative personnel, who often have very trivial and complicated daily work and find it difficult to devote a lot of time to course preparation and design, so the classroom teaching effect is not particularly satisfactory. Therefore, to strengthen the education effect of the career guidance course, we should take the optimization of the faculty as a breakthrough and establish a full staff and whole process education mechanism [23]. Graduates need to gain a detailed understanding of the current employment situation, the frontiers of development and job opportunities in the industry and advice available to them through career guidance courses, which are often sorely lacking for non-professional faculty. The external environment is changing rapidly and policies are being introduced very quickly, so it is essential to help graduates keep abreast of industry developments in real time. Career guidance courses in colleges and universities should give due consideration to the participation of professional teachers in the preparation and teaching process. On the one hand, it can provide students with a systematic introduction to explain the industryrelated enterprises, how to choose a suitable position from the many positions in the enterprise, to maximize their strengths and specialties. On the other hand, professional teachers have been developing in the industry for many years and have rich enterprise resources, which can help graduates achieve fast and high-quality employment. In addition, graduates can seek help from professional teachers when they encounter difficult problems in the job search process, so as to lay a good foundation for the next successful job application.
830
J. Ge
Furthermore, all teachers in the process of teaching professional courses should consciously guide students in the concept of employment and career selection, cultivate innovative and entrepreneurial ability, educate professional ethics and other comprehensive professional literacy education, create a sense of professional “atmosphere”, help students to establish a good employment concept as early as possible, give full play to the function of ideological and political education of each course, provide the whole process of accurate guidance services, and jointly promote high-quality employment of graduates to realize their career dreams.
5 Implementation Effect Evaluation In order to evaluate the effect of the research results on the promotion of the career guidance course for college students, 25 students were randomly selected from the two parallel classes of electronic information major (60 students in each standard class) that the author used to teach. A total of 50 students were evaluated and scored on their degree of employment readiness at the beginning and end of the course. The full score of the survey is 10 points, and the higher the comprehensive score, the better the preparation for employment. The course lasted for 8 weeks, and the teaching content is reasonably arranged according to the course innovation mechanism of this paper. The paired t-test method was used to sort out and analyze the scores of 50 students by SPSS software. As shown in Table 2, the results show that after continuous employment guidance and assistance for students according to the research results of this paper, the students’ preparation degree of employment has been significantly improved. t = −13.212, p < 0.001, 95%CI was (−1.69~−1.24). Table 2. Evaluation of curriculum implementation effect based on students’ survey Group
df
X ±S
t
p
95%CI of the difference
Pre-test
49
5.62 ± 1.02
−13.212
< 0.001
−1.69 ~ −1.24
Post-test
49
7.08 ± 0.85
6 Conclusion The quality of talent training in colleges and universities directly determines whether there is a qualified and high-quality workforce to undertake the important task of future social development and progress, and the fierce job market requires graduates to change their employment willingness, improve their employment literacy and enhance their jobseeking skills in response to the market and enterprise needs. Therefore, colleges and universities should strengthen the overall coordination, improve the education effect of the employment course from the curriculum design, faculty allocation, school-enterprise contact and other multi-dimensions, innovate the teaching methods, broaden the practice
Analysis of the Innovation Mechanism and Implementation
831
platform, really create a deep, high and warm employment guidance gold course for students, so that the career and employment guidance course can escort the graduates’ job search road and play the proper value of the course. Acknowledgements. This project is supported by Suzhou University 2021 Community-level Party Building Innovation Secretary Project(2021sjxm011) and Suzhou University 2022 the Second Batch of School-level Platform Open Subject Project(2022ykf29)
References 1. Lijuan, W., Muzhi, W.: Vocational guidance and career planning in China. Int. J. Adv. Couns. 20, 27–35 (1998) 2. Sun, V.J., Yuen, M.: Career guidance and counseling for university students in China. Int. J. Adv. Couns. 34, 202–210 (2012) 3. Qin, Y., Peng, J.: The Employment Services and Guidance of the Disadvantaged Groups of Electronics Professional College Graduates. In: Zhang, L., Zhang, C. (eds) Engineering Education and Management. Lecture Notes in Electrical Engineering, vol 111. Springer, Berlin, Heidelberg (2012). https://doi.org/10.1007/978-3-642-24823-8_114 4. Zhang, T., Shao, B.: The Research of Theory Education and Practice in University Career Planning. In: Zhang, L., Zhang, C. (eds) Engineering Education and Management. Lecture Notes in Electrical Engineering, vol 111. Springer, Berlin, Heidelberg (2012). https://doi.org/ 10.1007/978-3-642-24823-8_80 5. Fangfang, L.: Innovative Research on the Whole-process Model of college employment guidance in the new situation. Front. Educ. Res. 3(9), 49–51 (2020) 6. Watts, A.G., Sultana, R.G.: Career guidance policies in 37 countries: contrasts and common themes. Int. J. Educ. Vocat. Guidance 4, 105–122 (2004) 7. Grubb, W.N.: An occupation in harmony: the roles of markets and government in career information and guidance. Int. J. Educ. Vocat. Guidance 4, 123–139 (2004) 8. Stukalina, Y.: Career Management in a Technical University as an Essential Factor Influencing Its Competitiveness. In: Kabashkin, I., Yatskiv, I., Prentkovskis, O. (eds) Reliability and Statistics in Transportation and Communication. RelStat 2017. Lecture Notes in Networks and Systems, vol 36, pp. 639-648, Springer, Cham (2018). https://doi.org/10.1007/978-3-31974454-4_61 9. Kuzmin, A.M., Kunina, O.O., Fedorov, A.M., et al.: From career guidance of schoolchildren to professional training of future engineers at university of engineering and technology. Mobility for Smart Cities and Regional Development-Challenges for Higher Education. In: Proceedings of the 24th International Conference on Interactive Collaborative Learning (ICL2021), pp. 654–660 (2022) 10. Vaidya, N.M., Patel, K.K.: Learner performance and preference meter for better career guidance and holistic growth. Proc. ICT Anal. Appl. 2020, 47–54 (2019) 11. El Mrabet, H., Ait, M.A.: IoT-school guidance: a holistic approach to vocational selfawareness & career path. Educ. Inf. Technol. 26, 5439–5456 (2021) 12. Sun, Y., Aizhen. R.: Measures to improve the employability of college students under the new employment pattern. Vocat. Educ. Mech. Ind. (11), 16–20 (2022) (in Chinese) 13. Xing, Z.: Cultivation of college students employment values in the post epidemic era. J. Bengbu Univ. 11(5), 94–97 (2022) (in Chinese). 14. Junhui, L., Shujiao, C., Wuyu, Z., et,al.: Solution to “Slow Employment” of college students under Covid-19 Epidemic. J. Ningbo Univ. Technol. 34(3): 126–132 (2022) (in Chinese).
832
J. Ge
15. Sihang, C., Hui. X.: Using high quality employment guidance to help with college students employment. Chinese Social Sciences Today (2022) (in Chinese). 16. Xiaoting, L., Yanfang, Z.: Reflections on promoting precise career guidance services in colleges and universities with high quality. Youth Res. 05, 101–105 (2022). (in Chinese). 17. Hao, D., Sun, V.J., Yuen, M.: Towards a model of career guidance and counseling for university students in China. Int. J. Adv. Couns. 37, 155–167 (2015) 18. Lu, Y.: The application of positive psychology education in modern college students’ employment guidance to enhance self-confidence and improve personality. China J. Multimed. Netw. Teach. (06), 164–167 (2022) (in Chinese). 19. David, L.T., Trut, a, C., Cazan, A.M., et al.: Exploring the impact of a career guidance intervention program in schools: effects on knowledge and skills as self-assessed by students. Curr. Psychol. 41, 4545–4556 (2022) 20. Xiao, L.: Discussion on precisely college students’ employment service under the background of new media. Ability Wisdom 11, 133–135 (2022) (in Chinese). 21. Kuijpers, M.: Career guidance in collaboration between schools and work organisations. Br. J. Guid. Couns. 47(4), 487–497 (2019) 22. Yanmei, T.: Exploration of College Student Management Based on Employment Orientation. Heilongjiang Sci. 13(21), 61–63 (2022). (in Chinese) 23. Keller-Schneider, M., Zhong, H.F., Yeung, A.S.: Competence and challenge in professional development: teacher perceptions at different stages of career. J. Educ. Teach. 46(1), 36–54 (2020)
Comparative Study on the Development of Chinese and Foreign Textbooks in Nanomaterials and Technology Yao Ding1 , Jin Wen1 , Qilai Zhou1,2 , Li Liu1 , Guanchao Yin1(B) , and Liqiang Mai1(B) 1 School of Materials Science and Engineering, Wuhan University of Technology,
Wuhan 430070, China {guanchao.yin,mlq518}@whut.edu.cn 2 Faculty of Science, Shizuoka University, Shizuoka 422-8529, Japan
Abstract. Textbooks in colleges and universities are the carriers of knowledge that can reflect the teaching content and methods. It is also the media that can broaden students’ horizon and track modern scientific development, which is also the important symbol reflected the scientific research achievements. This paper compares the textbooks of “Nanomaterials and Technologies” in China with similar textbooks abroad from three aspects. First, by comparing the proportion of backgrounds, basic and expanded knowledge, the design of knowledge system is compared to find the advantages in foreign textbooks. Second, by comparing the readability, logic and self-conscious study which are provided by textbooks, the morphology of the textbooks is compared to find the weakness in Chinese textbooks. Finally, the application of information technology in textbooks is explored to find the improvements. The purpose of this study is to pave the way for the organization of Chinese textbooks in nanomaterials and technology in future. Keywords: Textbooks · Comparative study · Nanomaterials
1 Introduction Textbooks, one of the main means to help teachers and students understand the knowledge, needs to be systematic researched for its development in both Chinese and foreign ones, which can pave a way for the Chinese teachers in universities and colleges to promote the qualities of both traditional course teaching and E-learning [1–7]. Nanomaterials science and technology, a newly developed major, is an undergraduate major of regular higher education, which belongs to materials major [8, 9]. However, in China, only a few universities and colleges have offered this major at present [10–12]. Besides, less work has been done on the comparative research of the written and organization of Chinese and foreign textbooks [13, 14]. More importantly, as E-learning and online courses have become a new mode in teaching instead of traditional course teaching after COVID-19, the study of the differences in the usage of information technic between Chinese and foreign textbooks is important [15–20]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 833–843, 2023. https://doi.org/10.1007/978-3-031-36115-9_74
834
Y. Ding et al.
Thus, under the guidance of the above weakness in researches about Chinese and foreign textbooks, through a comparative study of Chinese and foreign textbooks in the major of nanomaterials science, this work explores the close relationship between cultivation objectives of students, curriculum setting and content construction, as well as the close relationship between the training mode of students’ abilities and the construction of textbooks. By statistically comparing the design of content system, analyzing the whole structure and comparing the application of information technology in these textbooks, this work will also explore the approaches for the embodiment of thoughts and teaching concepts in textbooks for advanced education, and study the role of textbooks both in supporting the traditional course teaching and E-learning. By using the statistical methodology, the main features and characteristics of excellent foreign textbooks in nanomaterials and technology are pointed out. It finds that foreign textbooks usually account for a better design structure and the abundant vividness and readability of illustrations, which can provide an openminded reading experience for students and will contribute to the self-learning. Therefore, this work can reveal the relevant shortcomings in Chinese textbooks of nanomaterials science major and pave a way for the organization and written of future textbooks.
2 Selection of Chinese and Foreign Textbooks The basic information of these Chinese and foreign textbooks majored in nanomaterials and technology, e.g., the contents, chief editors, publishing agency and the reasons for selecting as one of the objectives in this research work, are listed as follow. 1) Chinese textbooks. a. Nanomaterials and Their Preparation Techniques (1st Edition), Manhong Liu, Metallurgical Industry Press. This book was published by Metallurgical Industry Press in 2014 as a textbook for the 12nd Five-Year Plan of national higher education in China. The book is a basic textbook or teaching auxiliary book commonly used by students in colleges or universities majores in nanomaterials and technology. This makes the book with the certain representativeness and authority, thus been chosen as one of the objectives in this research. The book firstly introduces the background and development of nanomaterials science and then introduces the preparation and surface modification principles and methods of nanoparticles and nanomaterials, etc. Application principles in rubber and plastic nanomaterials, textile materials, optical materials, etc., are also included in this book. Therefore, this book which has a complete knowledge system has a certain research value. b. Fundamentals of Nanomaterials (2nd Edition), Yaojun Zhang, Chemical Industry Press. This book was published by Chemical Industry Press in 2014 as a textbook for the 12nd Five-Year Plan of national higher education in China. It can be used as a textbook for basic courses of nanomaterials and technology major. The book has been widely
Comparative Study on the Development of Chinese and Foreign Textbooks
835
used in lots of famous universities in China, such as Beijing University of Aeronautics and Astronautics, etc. Thus, this book is with the certain representativeness and authority and can be used as one of the objectives in this research. The textbook systematically introduces the basic concepts and classification of nanomaterials, e.g., nanoeffects, properties of nanomaterials, “top-down” and “bottom-up” preparation methods of nanomaterials, self-assembly of nanomaterials and so on. The book also includes the applications of nanotechnology in the research area as clean energy. Besides, there are two editions of the textbook (1st edition in 2011 and 2nd edition in 2015). The second edition is a bilingual edition, which has a value in researching the trend of development in Chinese textbooks. 2) Foreign textbooks. a. Fundamentals of Nanotechnology (1st Edition), Gabor L. Hornyak, CRC Press. This book, a typical foreign textbook majored in nanomaterials, which is widely used by many American universities (such as University of California, Los Angeles, etc.). Therefore, the book has the universality and representativeness among foreign textbooks in this major. The main author of this textbook, Gabor L. Hornyak, is a wellknown professor in the research area of nanomaterials. Besides, CRC Cham is one of the largest science and technology publishing agencies in the world. Thus, this textbook is also authoritative in the textbooks majored in nanomaterials, which makes this book suitable for this research work. At the same time, the book covers most of the basic knowledge of nanomaterials, including the basics of nanotechnology, practical applications of nanomaterials and knowledge of nanomaterials in drugs, etc. The book covers a wide range of knowledge for nanomaterials and technology with a complete system, which has the value for researching. b. Nanotechnology: Principles and Applications (3rd Edition), S. Logothetidis, Springer Cham Press. This book is a typical textbook widely used by universities in Europe and the United States (such as Harvard University, Oxford University, etc.) which have the major as nanomaterials and technology. Therefore, the book has the universality and representativeness among the foreign textbooks. The main author, S. Logothetidis, is also a wellknown professor in the research area of nanomaterials and technology. The publishing agency, Springer Cham, is one of the largest and most famous science and technology publishing agencies. Thus, this book is also authoritative in nanomaterials and technology. The book introduces nanostructures and nanomaterials for their applications in energy and organic electronics, knowledges of advanced nanomaterials deposition and processing methods, etc. It also focuses on the optical, electronic, surface and mechanical properties of nanomaterials. It provides a complete system of basic knowledges in nanomaterials and deserves to be carefully compared with Chinese textbooks.
836
Y. Ding et al.
3 Comparation of the Design of the Structure of Content in Chinese and Foreign Textbooks 3.1 Comparation of the Design of Knowledge System in Chinese and Foreign Textbooks In order to compare and analyze the design of knowledge system of the above Chinese and foreign textbooks in nanomaterials and technology, statistical analysis of the proportion of backgrounds, basic and expanded knowledge (including applied knowledge and frontier knowledge) in the whole textbook are listed as Table 1. Detailed data processing methods are also presented below. Table 1. Comparation of the proportion of backgrounds, basic and expanded knowledge in Chinese and foreign textbooks Textbooks
Chinese textbooks
Foreign textbooks
Backgrounds
Basic knowledge
Expanded knowledge Applied knowledge
Frontier knowledge
Nanomaterials and 0.10 Their Preparation Techniques
0.60
0.13
0.10
Fundamentals of Nanomaterials
0.18
0.42
0.14
0.16
Fundamentals of Nanotechnology
0.11
0.35
0.29
0.25
Nanotechnology: Principles and Applications
0.10
0.31
0.38
0.21
The calculation formula for the data of each observation point in this table is shown as Formula 1. ϕi =
n ain 1
An
× ωn
(1)
Here, ϕi is the proportion of backgrounds, basic or expanded knowledge, respectively. ain is the number of pages for each type of knowledge in each chapter. An is the number of pages in corresponding chapters. ωn is the weight of each chapter, ωn equals to the proportion of chapter pages in total pages; n is the number of chapters. Besides, the definition of each type of knowledge is listed as follow: a. Backgrounds. It includes the introduction of backgrounds in related basic subjects, e.g., mathematics, physics, chemistry and all related natural science basic courses. b. Basic knowledge. It includes the important concepts, definitions and basic theories in the book.
Comparative Study on the Development of Chinese and Foreign Textbooks
837
c. Expanded knowledge. It includes the comprehensive analysis and application of basic knowledge, including the case analysis, market prospects, frontier scientific researches, new technology application, etc. From Table 1, it can be found that the knowledge system of Chinese and foreign textbooks in nanomaterials and technology is significantly different, which is mainly reflected in the proportion of basic, applied and frontier knowledge. The similarities and differences are as follows: • Common points. Chinese and Foreign textbooks include all the three parts as background, basic and expanded knowledge in the knowledge system of whole book, which can enrich the teaching material and offer help for students to understand the knowledges in nanomaterials step by step. • Differences. 1) Depth of the knowledge system. The proportion of background and basic knowledge (>0.4) in domestic textbooks is higher than that in foreign textbooks, while foreign textbooks usually involve more expanded knowledge points (> 0.2). Here, more elaboration of basic knowledge can provide a deeper understanding in the nanomaterials for students when learning the relative courses with the help of textbooks. 2) Breadth of the knowledge system. In foreign textbooks, the proportion of applied knowledge is slightly larger (basically > 0.29) than that of domestic textbooks (only around 0.13–0.14). At the same time, foreign authors pay more attention to the content of frontier knowledge, the proportion of frontier knowledge (> 0.2) is much higher than that of Chinese textbooks (around 0.1). In Chinese textbooks, the applied knowledge and frontier knowledge are relatively lacking, which is reflected in the lacking of foresight and timeliness. To conclude, by comparing the design of knowledge system in both Chinese and foreign textbooks, we find that Chinese textbooks prefer to show the importance of “background” and “foundation”, while foreign textbooks highlight the “applications” and “frontiers”. Especially, the analysis of industry applications and market prospects in foreign textbooks is very distinctive, which is worth to be adopted in the organization of future Chinese textbooks. 3.2 Comparation of the Design of Content of Examples and Exercises in Chinese and Foreign Textbooks In order to explore the differences in the presentations of the above four types of knowledge, in this part we compare the design of content of examples and exercises in Chinese and foreign textbooks in nanomaterials and technology. Specially, the proportion of memorizing/conceptual and comprehensive/analytical questions in the examples and exercises in the textbooks are statistically compared respectively in Table 2. The calculation formula for the data of each observation point in this table is shown as Formula 2. n cin × ωn (2) αi = 1 Bn
838
Y. Ding et al.
Table 2. Comparison of memorizing/conceptual and comprehensive/analytical questions in the examples and exercises in Chinese and foreign textbooks Textbooks
Chinese textbooks
Foreign textbooks
Examples
Exercises
Memorizing /Conceptual questions
Comprehensive /Analytical questions
Memorizing/Conceptual questions
Comprehensive/Analytical questions
Nanomaterials and Their Preparation Techniques
/
/
0.40
0.60
Fundamentals of Nanomaterials
0.12
0.71
0.19
0.66
Fundamentals of Nanotechnology
0.09
0.88
0.08
0.78
Nanotechnology: Principles and Applications
0.06
0.79
0.11
0.75
Here, αi is the proportion of memorizing/conceptual and comprehensive/analytical questions in the examples and exercises in the textbooks, respectively. cin is the number of questions for each category in each chapter. Bn is the number of examples and exercises in each chapter. ωn is the weight of each chapter. n is the number of chapters. Besides, the definition of each type of observation point is listed as follow: a. Memorizing/conceptual questions. Objective questions that include an examination of concepts, nouns, definitions and other basic knowledge in the textbook. b. Comprehensive/analytical questions. Including the comprehensive use and analysis of basic knowledge in the textbook, and the opening and exploring questions about the expanded knowledge. From Table 2, it can be found that there are the following differences in the design of content of examples and exercises in Chinese and foreign textbooks in nanomaterials and technology. 1) In foreign textbooks, the proportion of memorizing/conceptual questions in examples and exercises (< 0.1) are smaller than that in Chinese textbooks (> 0.1), while comprehensive, comprehensive/analytical questions have the opposite proportion. Memorizing/conceptual questions can deepen students’ understanding of basic knowledge and strengthen their ability to grasp key information in the textbook. On the other hand, comprehensive/analytical questions can guide students to find diversified problem-solving methods from lots of problems and cultivate students’ comprehensive application ability of knowledge, which is valuable experience for preparing the future Chinese textbooks. 2) Foreign textbooks pay more attention to diversification of problem-solving methods for examples and exercises, which is helpful to improve the ability of students to think independently and respect students’ personalized solutions to the problems involved in the examples and exercises. Through students’ communication in the process of pre-class preview and after-class review, this characteristic in foreign textbook can
Comparative Study on the Development of Chinese and Foreign Textbooks
839
broaden their mind and cultivate the innovation thinking, which is very important for the improvement of students’ comprehensive ability. Therefore, adding comprehensive/analytical questions in the examples and exercises of the textbook and providing multiple problem-solving methods are helpful to improve students’ pioneer thinking, which is valuable experience for reference in the compilation of future Chinese textbooks in nanomaterials and technology. To sum up, based on the results from Table 1 and 2, it can be concluded that the similarities and differences in the design of structure of content between Chinese and foreign textbooks in nanomaterials and technology are as follow: • Common points. Both Chinese and foreign textbooks can take advantages of examples and exercises to provide backgrounds, basic and expanded knowledge for the readers. The system of knowledge is complete and fruitful, which is important for the classroom teaching. • Differences. 1) For foreign textbooks, they always highlight the “applications” and “frontiers”, which is different from Chinese textbooks that emphasize the “background” and “foundation”. Lots of extended knowledge occupy the main part of foreign textbooks, which is important for an advance textbook in nanomaterials and technology. This point should be adopted in organizing the Chinese textbooks in future. 2) Foreign textbooks pay more attention to diversification of problem-solving methods for examples and exercises, which is helpful to improve the ability of students to think independently. The comprehensive/analytical questions occupy the most in examples or exercises in foreign textbooks. This point is significant for a textbook which can guide the students to study independently, which should also be adopted in future Chinese textbooks.
4 Comparation of the Morphology of Chinese and Foreign Textbooks In order to make a detailed comparative analysis of the morphology of Chinese and foreign textbooks in nanomaterials and technology, four objects are observed and compared from the aspects of readability, logic and self-conscious study which are provided by textbooks. Among them, the readability of the textbook is evaluated by the number of figures and tables, the proportion of color pages and art font style. Specifically, statistic results of each point are concluded in Table 3. The calculation formula and methods for the data of each observation point in this table is as follow: a. Figures and Tables. The ratio of the total number of figures and tables to the number of pages in the textbook. b. Color pages. The ratio of the total number of color pages to the number of pages in the textbook.
840
Y. Ding et al. Table 3. Comparison of the morphology of Chinese and foreign textbooks
Textbooks
Readability Figures (/page)
Chinese textbooks
Foreign textbooks
Logic
Self-conscious study
Tables (/page)
Color page (%)
Nanomaterials and 0.89 Their Preparation Techniques
0.09
0
Fundamentals of Nanomaterials
0.12
0.034
/
There is less cohesive language between chapters and sections
0.22
Fundamentals of Nanotechnology
0.18
0.066
0.89
There is lots of cohesive language between chapters and sections
0.47
Nanotechnology: Principles and Applications
0.14
0.053
0.83
There is lots of cohesive language between chapters and sections
0.43
0.64
c. Self-conscious study. This point is evaluated by Formula 3, which calculates the proportion of exercises and examples in the textbook. n δi = din × ωn (3) 1
Here, δi is the proportion of exercises and examples in the textbook. din is the number of exercises and examples in each chapter. ωn is the weight of chapter pages. n is the number of chapters. From Table 3, it can be concluded that lots of differences are existed in the morphology of Chinese and foreign textbooks, such as the proportion of figures and tables, the
Comparative Study on the Development of Chinese and Foreign Textbooks
841
use of art font words and the proportion of exercises and examples. The similarities and differences are listed as follows: • Common points. Both of Chinese and foreign textbooks can use figures and tables to help the illustration of knowledge, which will provide a clear view for the readers. • Differences. The advantages of foreign textbooks are reflected in the beautiful layout, the use of color pages, interesting pictures and ornamental art fonts. Besides, the statement in the book maintains a strict but humorous tone. The basic knowledge of nanomaterials can be introduced vividly in foreign textbooks by means of the combination of colorful tables and pictures. At the same time, foreign textbooks are more friendly to undergraduates in terms of writing style and problem solving. For example, the explanation is easy to be understood and the steps of problem solving are provided detailly. The shortcomings of Chinese textbooks are the lack of vividness and readability of illustrations, which are mainly reflected in the small proportion of tables and figures in textbooks. Many illustrations in Chinese textbooks lack clarity and timeliness, and cases are dated to a long time ago, which has caused some obstacles for students to independent learning according to textbooks. In recent years, more and more online teaching (including E-learning, online courses, etc.) has been applied to the education in universities. In particular, COVID-19 swept across the country and even the world since 2020, forcing all universities and colleges to delay the back to school. In order to ensure the quality of education, lots of universities have implemented online teaching. It can be predicted that online teaching will occupy an important place in future and form a complement to the traditional teaching. Facing this trend, there is a problem that needs to be seriously considered that how to serve the courses and teachers with more complete textbooks in nanomaterials and technology. What kind of textbooks should be made and how to serve the online teaching in a better way are the main problems that publishers need to consider. With the rapid development of online education, textbooks should conform to the E-learning and online courses. From the perspective of this, a textbook should not only use papers as the carrier, but also should be equipped with various digital resources by advanced information technology. Therefore, we conduct the research part about the comparation of the application of information technology in Chinese and foreign textbooks. The results are concluded as following: • For the Chinese textbooks. The textbooks are mainly traditional paper textbooks, most of which lack supporting digital resources and are only suitable for traditional offline teaching rather than online teaching. • For the foreign textbooks. 1) Almost all the professional textbooks of nanomaterials published in the past 10 years have electronic versions. Students can directly purchase authentic electronic textbooks from online platforms such as Amazon. 2) Most of the textbooks have the exercises in the electronic version, also video for explanation of the exercises and other online teaching materials, which is convenient for students to learn the knowledges independently according to the book.
842
Y. Ding et al.
5 Conclusion The main purpose of this study is to compare the Chinese textbooks in nanomaterials and technology with foreign textbooks and find a new way to improve the organization in Chinese textbooks in the related fields. Here, this study compares the books in terms of their design of structure, their morphology and the application of information technology (Fig. 1).
Fig. 1. Conclusions in this study
By comparing the structure of knowledge system in Chinese and foreign textbooks, it finds that both Chinese and foreign textbooks can contain the background knowledge, the basic knowledge and the extent knowledge while foreign textbooks have more extent knowledge to open the students’ mind during learning. Besides, by statistically studying the proportion of figures, tables and color pages in the textbooks, the readability of Chinese and foreign textbooks is examined. Furthermore, the logic and the self-conscious study of the textbook are evaluated by the proportion of exercises and examples in the textbooks. It finds that the foreign textbooks usually have a large potential of art fonts, color pages and electronic materials to improve the readability and further can help the students in self-learning, as well as a larger proportion of the exercises and examples to simulate the self-motivation of students during course teaching. All these advantages this work presented in foreign books can be adopted by Chinese textbooks in future. Acknowledgements. This work is supported by the Teaching Reform Project of Wuhan University of Technology (Grant No. w2022067).
References 1. Liu Qingxian, Y., Jingyi.: Current status and innovation attempt of compiling textbooks of Chinese medicine in medical colleges in China. TMR Theory Hypothesis 2, 478–485 (2021)
Comparative Study on the Development of Chinese and Foreign Textbooks
843
2. Rodríguez-Regueira, N., Rodríguez-Rodríguez, J.: Analysis of digital textbooks. Educ. Media Int. 59(2), 172–187 (2022) 3. Bittar, M.: A methodological proposal for textbook analysis. Math. Enthusiast 19(2), 307–340 (2022) 4. Paveši´c, B.J., Cankar, G.: Textbooks and students’ knowledge. Center Educ. Policy Stud. J. 12(2)29–65 (2022) 5. Ramia, J.M., Soria-Aledo, V.: Textbook outcome (resultado de libro): una nueva herramienta de gestión. Cir. Esp. 100(3), 113–114 (2022). https://doi.org/10.1016/j.ciresp.2021.06.002 6. Kyrpychenko, O., Pushchyna, I., Kichuk, Y., Shevchenko, N., Luchaninova, O., Koval, V.: Communicative competence development in teaching professional discourse in educational establishments. Int. J. Mod. Educ. Comput. Sci. 13, 16–27 (2021) 7. Amjad, M., Akter, H.: An android based automated tool for performance evaluation of a course teacher (CTE)[J]. Int. J. Mod. Educ. Comput. Sci. 12, 14–23 (2020) 8. Zhijuan, W., Xiaobin, Z., Wei, S., Antai, W.: Readability assessment of textbooks in low resource languages. Comput. Mater. Continua 10, 213–225 (2019) 9. Qiyue, C., Zhongliang, N., Zhiyu, L.: Textbook outcome as a measure of surgical quality assessment and prognosis in gastric neuroendocrine carcinoma: a large multicenter sample analysis. Chin. J. Cancer Res. 4, 433–454 (2021). (in Chinese) 10. Jiaming, Q.: A study on the representational meaning of the images in English textbooks based on visual grammar. Proceedings of Northeast Asia International Symposium on Linguistics, Literature and Teaching (2021) 11. Hongxing, J.: A brief review of root, stem and base and their application in the textbook of new senior English for China. Overseas Engl. 20, 266–267 (2020) 12. Fei, Y.: Exploring native-speakerism in EFL textbooks in China. Overseas Engl. 14, 260–262 (2018). (in Chinese) 13. Wei, W.: A comparative analysis of three versions of junior high school English textbook from the aspect of discourse[J]. Overseas Engl. 1, 242–243 (2018). (in Chinese) 14. Xiaoyuan, S., Min, W.: An Eco-critical discourse analysis of texts in English textbooks based on discourse-historical approach. Overseas Engl. 17, 221–222 (2017). (in Chinese) 15. Bakken, A.-B.: The textbook task as a genre. J. Curric. Stud. 53(6), 729–748 (2021) 16. Benjamin Ginsberg. Thoughts on textbook writing. PS: Polit. Sci. Polit.55(3): 636–637 ( 2022) 17. Kollman, K.: The textbook road taken. PS: Polit. Sci. Polit. 55(3): 638–640 (2022) 18. van Gijn, D., van Gijn, D., Newlands, C.: From proposal to publication-The process of writing an academic textbook. Br. J. Oral Maxillofac. Surg. 60(10), e58 (2022) 19. Harijanto, B., Apriyani, M,E., Hamdana, E.N.: Design online learning system for Kampus Merdeka: a case study web programming course. Int. J. Mod. Educ. Comput. Sci. 13, 1–9 (2021) 20. Long, D.T., Tung, T.T., Dung, T.T.: A facial expression recognition model using lightweight dense-connectivity neural networks for monitoring online learning activities. Int, J. Mod. Educ. Comput. Sci. 14, 53–64 (2022)
Practical Research on Improving Teachers’ Teaching Ability by “Train, Practice and Reflect” Mode Jing Zuo1 , Yujie Huang1 , and Yanxin Ye2(B) 1 Nanning University, Guangxi 530299, China
[email protected]
2 Guangxi Modern Polytechnic College, Guangxi 54700, China
[email protected]
Abstract. The key to talent training quality lies in teachers’ teaching level and teaching ability. As an essential way to develop the teaching ability of teachers, training has been emphasized by many colleges and universities. However, the training form of many colleges and universities nowadays is still mainly the traditional “lecturing” and “infusing” one. To test whether the mode of “train, practice and reflect” can effectively help teachers to improve their teaching ability, the teaching ability development program is designed and implemented from three aspects of training, practice and reflection, and the action research is carried out. The teachers are pre-tested and post-tested before and after the training, and the independent sample T-test is conducted by SPSS. The results show that the “train, practice and reflect” mode can significantly improve teachers’ instructional design, teaching implementation, teaching evaluation, reflective ability and teaching innovation ability. Keywords: Training mode · Teaching ability · Private university
1 Introduction As an essential part of China’s higher education system, private colleges and universities have played an essential role in popularizing higher education [1]. However, there are still many problems in classroom teaching in private application-oriented universities. Generally speaking, classroom teaching is still a one-way teaching indoctrination teaching method that lacks effective teacher-student interaction and student participation [2], of which the classroom atmosphere is dull, students are still in a passive learning state, and the concept of “student-centered, ability-oriented and outputoriented” needs to be further strengthened [3–5]. Maintaining the construction of teachers in private application-oriented universities, and significantly improving the teaching ability of teachers, has become the key to improving the teaching quality of private application-oriented undergraduate universities [6]. Many universities and scholars have carried out relevant research and practice, including research on classroom teaching quality [7, 8], teaching ability [9], the relationship between teaching ability and classroom teaching quality, etc. These help us © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 844–853, 2023. https://doi.org/10.1007/978-3-031-36115-9_75
Practical Research on Improving Teachers’ Teaching Ability
845
understand the connotation of teachers’ teaching ability and pinpoint the value orientation of optimizing course teaching quality. However, the following problems still lie in these studies and practices: First, the development of the teaching ability of teachers in application-oriented colleges and universities needs to be carried out fundamentally based on the standard requirements of application-oriented talent training. China’s higher education can be divided into research, application, and vocational skills. And the demands and modes of talent cultivation vary in different universities. Basic theoretical knowledge is indispensable for the cultivation of application-oriented talents, with ability training as the core, and more attention must be paid to the comprehensive application of knowledge [10]. It is necessary to strengthen basic theoretical teaching, and practical application integration and improve students’ ability to solve practical problems through practical projects [11]. All of these require that the teaching organization of applied teachers is mainly realized within the realm of ability [12]. However, the group of applied college teachers is influenced by the legacy of the traditional education model. They believe that the university stage is mainly for knowledge learning and theoretical research. To a conscious extent, they are more inclined to enrich academic knowledge structure [13]. A considerable number of teachers classroom teaching mode is accustomed to the traditional teaching mode, and the classroom teaching method is single, which is difficult to stimulate students’ enthusiasm and initiative in learning; Second, in terms of form, the training forms for the development of teaching ability of teachers in application-oriented colleges and universities are mainly traditional “lectures” and “indoctrination”, and lack the training and practice to cultivate teaching ability. Due to the large proportion of young teachers in most private application-oriented colleges and universities, the task of teacher development is heavy, and the training methods that can quickly cover a large area of teachers are mainly “invite in, go out”. These training methods do play a specific role, but there are also specific problems: most of the lectures and reports are “lecture-style”, only emphasizing the input of educational concepts and educational method knowledge, ignoring the cultivation and training of teaching ability and teaching skills [14]. Therefore, teachers lack the practical operation to integrate training content into actual teaching. This “lecture-style” training form will also make teachers unconsciously continue to use it in their classroom teaching, attach importance to the transmission of knowledge, and ignore the cultivation of students’ abilities. To tackle the above problems, this study designs and implements teaching ability development programs from three aspsects: training, practice and reflection, starting from the classroom teaching quality standards of applied talent training, and explores the development of teachers’ teaching ability through action research. To understand the influence and role of the coupling linkage teaching ability training model of “train, practice and reflect” on teachers’ ability in private colleges and universities and provide an empirical basis for scientifically improving teachers’ teaching ability.
846
J. Zuo et al.
2 Design and Implementation of the “Train, Practice and Reflect” Coupling Linkage Teaching Ability Training Program 2.1 Scheme Design Before carrying out the scheme design, the research group members sorted out the relevant literature, such as talent cultivation in application-oriented universities, curriculum construction in application-oriented universities, and classroom teaching quality in application-oriented universities. Interviews were conducted with some full-time teachers, course leaders, directors of teaching and research offices, heads of functional departments, and leaders of schools in charge of teaching in three private universities in Guangxi. And discussed in-depth issues such as “application-oriented talent training”, “application-oriented classroom form” and “application-oriented classroom teaching quality standards”. According to the requirements of application-oriented talent training [15], student-centered, ability and result output-oriented, the “Teaching Quality Standards for Applied Courses” was compiled from five dimensions: teaching attitude, learning objectives, instructional design, teaching implementation, and teaching effect. According to the quality standards of applied classroom teaching, combined with the goal of talent training in a private university and the needs of teachers’ teaching ability development, the teacher-teaching ability development program is designed from three aspects: training, practice and reflection. Since the improvement of teaching ability is not achieved overnight and requires a specific development cycle, the entire training period is half a year when designing the program, of which the centralized training time is one week, and the half year after the intensive training will be accompanied by regular events, seminars and luncheon activities. 2.2 Program Implementation The use of action research aims to solve the problem of teacher teaching capacity improvement. 200 teachers were recruited to carry out training in batches and phases, emphasizing “learner-centered, competency and outcome-output-oriented”, changing the traditional training mode of only conducting lectures (lecture-based) and carrying out diversified training modes such as inquiry-based, personalized and participatory; Through assignments, tasks, projects, etc. as the carrier, let the trainees use the training content to carry out multiple teaching exercises in the training; Through unique design and specialized training on teaching reflection in training, teachers can continuously reflect and learn in practice, and constantly improve the practice of teaching reform through review and summary. At the end of the week’s intensive training, the research group will also understand the situation of teachers applying what they have learned in training to their actual teaching and organize various teaching competitions, as well as seminar-like activities such as teaching luncheons and teaching salons, to provide an excellent environmental foundation for teachers’ subsequent practice and reflection. After the end of half a year, the “Applied College Teacher Teaching Ability Evaluation Scale” was used to conduct pre-test and post-test before and after the program’s implementation to understand the development of teachers’ teaching abilities.
Practical Research on Improving Teachers’ Teaching Ability
847
3 Research Methods 3.1 Survey Subjects The research data of this paper comes from 200 teachers from a private university in Guangxi who participated in the “training, practice and reflection” coupling teaching ability training program designed by the university. Table 1 is a sample of descriptive statistical results of implementing the training plan. Table 1. Sample descriptive statistical results Variable
Gender
Age
Seniority
Job Title
Indicator Before Training
After Training
Frequency
Percentage
Frequency
Percentage
man
63
32.5
56
31.1
woman
131
67.5
124
68.9
total
194
100
180
100
30 or less
29
14.9
25
13.9
31–37
90
46.4
84
46.7
38–44
48
24.7
46
25.6
45–51
19
9.8
18
10
52 and above
8
4.1
7
3.9
total
194
100
180
100
less than three years
43
22.2
37
20.6
3–8 years
84
43.3
81
45
9–14 years
36
18.6
35
19.4
15–21 years
20
10.3
17
9.4
more than 22 years
11
5.7
10
5.6
total
194
100
180
100
beginner and below
27
13.9
23
12.8
intermediate
100
51.5
94
52.2
deputy senior
57
29.4
53
29.4
positive advanced
10
5.2
10
5.6
total
194
100
180
100
The training methods and approaches are designed in light of the standard process and scientific methods [16]. The questionnaire was divided into pre-training and posttraining distribution, and the trainees received the questionnaire survey three days before the training, and the trainees answered the questions anonymously. A total of 194 people participated in the answers, and the questionnaire recovery rate was 97%. At the end of
848
J. Zuo et al.
the half-year training, the same questionnaire was distributed again, and a total of 180 people participated in answering the questions, and the questionnaire recovery rate was 90%. The basic information of teachers before and after participation is fully displayed. 3.2 Scale Design The development of the “Applied College Teacher Teaching Ability Evaluation Scale” is mainly divided into the following three processes: First, according to Tang Yuguang’s research, teaching ability includes five dimensions: instructional design, teaching implementation, teaching evaluation, teaching reflection, and teaching innovation, and combined with relevant literature, a scale is preliminarily compiled based on the understanding of the concept of this dimension. Second, the Delphi method was used to conduct multiple rounds of consultation with relevant experts and improve the scale. Third, 80 teachers from a university were selected for the pilot test, the reliability and validity of the scale were tested, and some questions and expressions of the scale were adjusted. The resulting scale and related topics are shown in Table 2. Table 2. Examples of Scale Dimensions and Their Titles Variable
Dimension
Questions
Examples of Topics
Teaching Ability
Instructional Design
7
When preparing for lessons, I can complete the instructional design relatively smoothly
Teaching Implementation
6
I understand the diversity of students’ learning and can teach them accordingly
Pedagogical Evaluation
3
I design and conduct assessments suitable for evaluating students’ learning outcomes (achievement of learning goals)
Pedagogical Reflection
3
After teaching, I reflect on my teaching
Pedagogical Innovation
4
I would like to explore a variety of information technology-assisted teaching activities
4 Data Analysis and Discussion Sample T-test is carried out to test whether there is a significant difference between the teaching ability and its five-dimensional sub-abilities before and after the training.
Practical Research on Improving Teachers’ Teaching Ability
849
4.1 Reliability Test The reliability of each dimension was analyzed using Cronbach’s alpha coefficient, and the results are shown in Table 3. The reliability of the five measurements is greater than 0.7, which shows good scale reliability. Table 3. Scale Reliability, Average Value and Standard Deviation of Each Dimension Dimension
Quantity
Average Value
Standard deviation
Cronbach’s Alpha
Instructional Design
374
4.10
0.57
0.844
Teaching Implementation
374
3.92
0.54
0.774
Pedagogical Evaluation
374
3.88
0.60
0.722
Pedagogical Reflection
374
4.03
0.61
0.764
Pedagogical Innovation
374
3.92
0.56
0.780
4.2 Average Value, Standard Deviation and Independent Sample T-Tests To investigate whether there is a significant difference in teaching ability before and after training, an independent sample T-test was conducted. Table 4. Results of independent sample t-test in the group before and after training Dimension
Constituencies
Quantity
Average Value
Standard Deviation
T
Degree of Freedom
Instructional Design
Before training
194
3.96
0.69
4.974***
286.407
After training
180
4.24
0.34
Teaching Implementation
Before training
194
3.71
0.59
8.551***
321.241
After training
180
4.14
0.36
Pedagogical Evaluation
Before training
194
3.69
0.65
7.020***
347.162
After training
180
4.09
0.46
Pedagogical Reflection
Before training
194
3.91
0.66
4.039***
363.710
After training
180
4.16
0.52
Pedagogical Innovation
Before training
194
3.78
0.63
5.233***
341.085
After training
180
4.07
0.43
Total Teaching Capacity
Before training
194
3.83
0.58
7.130***
268.654
After training
180
4.16
0.25
Note: *** indicates significance at the 0.01 level
Table 4 shows the five dimensions of teaching ability, the mean and standard deviation of teaching ability before and after training, and the T-test results. The results show
850
J. Zuo et al.
significant differences at the 0.01 level of the five dimensions of instructional design, teaching implementation, teaching evaluation, teaching reflection and teaching innovation, and the same goes with the total teaching ability at the 0.01 level. After the training, its teaching ability and its five dimensions were significantly enhanced compared with before the training. As the data in Table 4 shows, teachers’ teaching capabilities have significantly changed after training. The improvement of these capabilities is also consistent with the design idea of the “training, practice, thinking” coupling linkage teaching ability training program in this action study, discussed as follows. (1) Instructional design ability Gagne first proposed instructional design, which refers to planning the entire teaching system and offering specific solutions [17]. Instructional design is an essential competency for teachers. To help teachers master instructional design, a training course on instructional design is set up in this action study Help teachers master the methods of instructional design. Then the training program also sets up practical activities of teaching drills so that teachers can put instructional design to the reasonable level through the subsequent teaching drill process; In the teaching exercise, teachers were allowed to observe each other, put forward suggestions for revision and guidance, and at the same time filmed the video of the drill for the teacher. The teacher could reflect on whether the instructional design was scientific and reasonable through peer communication and a personal video review. Therefore, this training mode of training, practice and reflection helps teachers quickly understand and master the essentials of instructional design and improve their instructional design ability. (2) Ability to implement teaching The ability to implement teaching refers to teachers’ ability to effectively organize and implement instruction according to instructional design and actual environment. Student competence can be divided into core competence and position-related competence, and position-related competence mostly depends on the effective teaching implementation of teachers. To help teachers to improve their teaching implementation ability, they are asked to carry out corresponding teaching drills in the process of participating in the training. After each teaching drill, teachers will receive suggestions and opinions from peers, and teachers will continue to carry out the next teaching drill after further improvement. Through such repeated practice, teaching skills are integrated into practical exercises, which enhances teachers’ confidence in implementing teaching and strengthens teachers’ skills in implementing it. Due to the repeated exercises during and after the training, most of the teachers will apply the knowledge they have learned to their own teaching for real-life teaching. And the school also cooperates with various teaching competitions to help teachers continuously polish their teaching skills and improve their teaching implementation capabilities. (3) Teaching evaluation ability According to the reverse design principle of instructional design, we first determine the learning objectives (learning outcomes), then decide how to evaluate whether the learning objectives have been achieved (what is the evidence of learning
Practical Research on Improving Teachers’ Teaching Ability
851
outcomes), and finally, design learning activities and tools. The “determine by what method to evaluate whether the goal has been achieved” refers to teaching evaluation. Teaching evaluation ability refers to the power of teachers to design scientific and reasonable teaching evaluation methods to carry out an evaluation to help students learn. In this action study, the overall design of the curriculum also reflects the concept of OBE, guiding teachers to pay attention to students’ output, think about how to evaluate students’ production based on students’ output, and then carry out an instructional design. The guidance of these concepts runs through training, practice and reflection, so that teachers can think about how well they have achieved their goals in this teaching drill and whether they have designed reasonable evaluations based on the goals every time they do instructional design and teaching drills. (4) Teaching reflection ability Dewey was the first to introduce “reflection” into teaching, and teaching reflection refers to the process of teachers themselves reviewing, thinking and summarizing their teaching. Professor Ye Lan once said that a teacher who writes a lifetime lesson plan could not become a famous teacher, and if a teacher writes three years of teaching reflection, he may become a renowned teacher. The ability to reflect on teaching is also an indispensable part of a teacher’s teaching ability. In this action study, the overall design of the plan is also to integrate reflection into all links. For example, after the teaching exercise, each teacher is required to fill out a self-evaluation form for this training, reflect on the advantages and shortcomings of the teaching just now, and guide teachers to think about how to improve in the future; At the same time, in peer feedback, teachers will unconsciously reflect on their entire teaching process again while listening to peer feedback; Videos were also recorded for each teacher’s teaching exercise for teachers to review privately to reinforce the process of self-reflection further. After the overall training, seminars such as teaching salons and teaching luncheons are also provided, allowing teachers to put forward their confusions and problems encountered in teaching practice and constantly review and review their teaching through discussions and exchanges with peers, to improve teachers’ teaching reflection ability. (5) Ability to innovate in teaching Teaching innovation has been a hot word in recent years. Teaching innovation ability refers to the power of teachers to start from real teaching problems and solve problems with innovative methods. Trying new teaching models provide teachers with a fresh approach and gives them tools to improve the teaching process. Teaching models give teachers conceptual as well as a practical technology to teach from. New teaching approaches can help them understand the view of the content to be taught, as well as to reflect on their ideas of learning, the learner, and their role of themselves. In this action study, the design of the training program did not specifically cover the training content of teaching innovation. Still, the final results showed that teachers’ teaching innovation ability increased, and the reasons might be as follows: First, because the whole training emphasises the integration of training, practice and reflection, teachers were able to grasp ideas and methods through training, bridge the gap between learning and doing through practice, and find problems in their teaching through reflection. It is known that finding problems is very important, and this lays the foundation for teachers to innovate in education. Second, because the training
852
J. Zuo et al.
process creates an atmosphere of trust and inclusiveness, it also encourages teachers to try new ways and methods to solve problems after discovering problems, which also provides environmental support for teachers to carry out teaching innovation; Third, due to the form of peer assistance in training, teachers can continue to reflect on themselves after carrying out teaching innovation, and on the other hand, they can communicate and discuss with peers, to better improve teaching. Based on the above analysis, improving teachers’ teaching innovation ability is promoted.
5 Conclusion Current research shows that the coupling linkage teaching ability training model of “train, practice and reflect” can significantly improve teachers’ teaching ability, which is mainly reflected in the improvement of five aspects: teachers’ instructional design ability, teaching implementation ability, teaching evaluation ability, teaching reflection ability and teaching innovation ability. The coupling linkage teaching ability training model of “train, practice and reflect” can effectively help teachers improve their teaching ability, which breaks through the simple training lecture form and integrates training, practice and reflection. It allows teachers to learn through diversified learning activities, deeply involved practical training, and peer-to-peer learning mode. This also helps the institutions responsible for teacher training and teacher development in our universities to better think about the design of our training programs. Due to the limitations of the research institute, this action plan is limited to teachers in a private university, the coverage needs to be wider, and the data needs to be more representative. Follow-up research can collaborate with more similar universities and cover a wider group of university teachers. Acknowledgment. This paper is the result of the 2021 private higher education research project “Research on the improvement of teaching ability of teachers in private applied universities” (No.2021ZJY667) of the Guangxi Education Science Plan.
References 1. Shi, Q., Zhang, C.: The internal logic of private universities’ development: reconstruction and transformation paths. J. Higher Educ. Manage. 14(04), 25–31 (2020). (in Chinese) 2. Pianta, R.C., Hamre, B.K., Allen, J.P.: Teacher-student relationships and engagement: conceptualizing, measuring, and improving the capacity of classroom interactions. Handbook of Research on Student Engagement, pp. 365–386. Springer, Boston, MA (2012). https://doi. org/10.1007/978-1-4614-2018-7_17 3. He, L.: Application of English situational teaching in classroom. Adv. Vocational Tech. Educ. 3(2), 142–146 (2021) 4. Ahmad, N.N.N., Sulaiman, M.: Case studies in a passive learning environment: some Malaysian evidence. Account. Res. J., (2013) 5. Gu, W.: Reform and exploration of database principle course based on OBE learning outcomeoriented. In: The 4th Annual 2018 International Conference on Management Science and Engineering (MSE2018). Francis Academic Press, UK, pp. 70–75 (2018)
Practical Research on Improving Teachers’ Teaching Ability
853
6. Steele, D., Zhang, R.: Enhancement of teacher training: key to improvement of English education in Japan. Procedia Soc. Behav. Sci. 217, 16–25 (2016) 7. Kong, L., Wang, X., Yang, L.: The research of teaching quality appraisal model based on AHP. Int. J. Educ. Manage. Eng. (IJEME) 02, 29–34 (2012) 8. Zhang, S.: Comprehensive evaluation of teaching quality based on cluster analysis and factor analysis. Int. J. Educ. Manage. Eng. (IJEME) 01, 16–23 (2011) 9. Li, N., Zhang, Y.: Improvement and practice of secondary school geography teachers’ informatization teaching ability based on the perspective of MOOCs. Int. J. Educ. Manage. Eng. (IJEME) 12, 11–18 (2022) 10. Chen, X., Yang, X.: 14 basic problems in the development of new application-oriented undergraduate colleges. China Univ. Teach. 01, 17–22 (2013). (in Chinese) 11. Pan, M., Che, R.: On the positioning of application-oriented universities. J. Higher Educ. 30(05), 35–38 (2009). (in Chinese) 12. Chen, X.: Exploration of the construction of new applied technology university, pp. 378–388. Guangxi Normal University Press, Guilin (2019). (in Chinese) 13. Cai, H., Xiong, K.: Action learning: training path of the teaching ability of “double teacher” teachers in applied undergraduate colleges. Heilongjiang Res. High. Educ. 36(06), 100–104 (2018). (in Chinese) 14. Wu, X.: The practice and reflection of the “Young teachers’ teaching contest” to improve teaching ability of young teachers in university. Theory Pract Educ. 38(09), 41–42 (2018). (in Chinese) 15. Zuo, J., Zhang, G., Huang, F.: Construction of ability and quality model of engineering talents based on analytic hierarchy process. In: Zhengbing, H., Zhang, Q., Petoukhov, S., He, M. (eds.) Advances in Artificial Systems for Logistics Engineering, pp. 356–366. Springer International Publishing, Cham (2022). https://doi.org/10.1007/978-3-031-04809-8_32 16. Hatice, Y., Murat, D.: Training program supporting language acquisition. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 03, 1–12 (2021) 17. Ma, L., Sheng, Q., ed.: Teacher Instructional Design Ability Development, vol. 36 Zhejiang University Press, Hangzhou (2016). (in Chinese)
College Foreign Language Teacher Learning in the Context of Artificial Intelligence Jie Ma1,2(B) , Pan Dong3 , and Haifeng Yang4 1 School of Education, Huazhong University of Science and Technology, Wuhan 430074, China
[email protected]
2 School of Economics and Business Foreign Languages, Wuhan Technology and Business
University, Wuhan 430065, China 3 South-Central Minzu University, Wuhan 430074, China 4 Faculty of Letters, Arts and Sciences, Waseda University, Shinjuku, Japan
Abstract. This study adopts a combination of quantitative and qualitative methods, uses the “teacher learning questionnaire” and interviews to explore and analyze the composition, characteristics and impact of AI on teacher learning of 148 college foreign language teachers in the context of AI language learning. The results are as follows: college foreign language teachers’ learning includes three dimensions: intrinsic motivation, organizational support, and feedback and evaluation, and the participants’ intrinsic motivation and feedback and evaluation reach the high-frequency experience level; the characteristics of college foreign language teachers’ learning are ecological orientation, practical orientation and individual orientation; artificial intelligence technology guarantees the autonomy of teachers’ learning through learner characteristics, supports the normalization of teachers’ learning through effective supply of learning resources, and improves the collaboration of teachers’ learning through virtual teaching and research rooms. Artificial intelligence integrates information technology and education, which helps to deepen and expand the research on teacher learning. Keywords: Artificial intelligence · College foreign language teacher · Teacher learning · Ecological orientation · Lifelong learning
1 Introduction “Teacher learning” is a concept embedded in the field of teacher education. With the increasingly prominent characteristics of teachers’ initiative, daily learning and endogenous knowledge, teacher learning gradually replaces “teacher professional development” and becomes a new concept of teacher education [1]. In addition, under the influence of constructivism, situational learning theory and socio-cultural theory, with the occurrence of the “learning revolution”, teacher learning research more clearly and consciously absorbs new learning theories to create a teacher learning environment and test its role in promoting teacher development, so as to make teacher learning become one of the frontier fields of teacher cognitive research and teacher education practice [2]. These provide a theoretical basis for further research on teacher learning. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 854–864, 2023. https://doi.org/10.1007/978-3-031-36115-9_76
College Foreign Language Teacher Learning
855
In 2018, the CPC Central Committee and the State Council issued the Opinions on Comprehensively Deepening the Reform of the Construction of Teachers in the New Era, which proposed: “By 2035, teachers’ comprehensive quality, professional level and innovation ability will be greatly improved, and millions of backbone teachers, hundreds of thousands of outstanding teachers and tens of thousands of educators will be trained”. Foreign language teachers are the subjects to “build new majors or directions, exploring new training modes, building new courses, and building new theories” [3]. Teacher learning is the only way to achieve these goals. At present, a social ecosystem based on the flow of data and knowledge among social subjects is gradually taking shape, and big data driven education research, education management and decision-making have become the theme of the times [4]. With the further development of information and technology, AI participates in the breadth and depth of teachers’ learning methods, quality assurance mechanisms, organizational support and operation modes, information and communication technology approaches and social support systems. Based on these changes, this study aims to provide some inspiration and reference for the further promotion of teacher learning research driven by big data in the process of deepening the integration of information technology and education through sorting out foreign language teacher learning, especially the analysis of teacher learning turn in the context of artificial intelligence.
2 Literature Review 2.1 Teacher Learning Teacher learning refers to “a holistic activity in which teachers actively seek to improve their overall quality under the support of external environment and continue to pursue the mutual unity of professional development and personal development” [5]; As an alternative concept of teachers’ professional development, teacher learning emphasizes “teachers’ initiative in their own professional development” [6], “the daily nature of the learning process” [7], and “the self-sufficiency of teachers’ knowledge and ability” [8]. In the 1950s and 1960s, Taylor, an American scholar, prospectively put forward that “the future in-service training will not be regarded as ‘training’ teachers, but will help, support and encourage each teacher to develop the teaching ability he values and hopes to increase. The guiding and universally recognized spirit will be to put learning itself in the most important position” [9]. This study uses Fred Korthagen’s The Onion Reflection Model of Teacher Development, which reveals the process of teacher learning: (1) In teacher learning, the cognitive, emotional and motivational dimensions of instructional behavior are intertwined, so teacher learning is multi-dimensional learning. (2) In the model, the learning process occurs at all levels, that is, teacher learning is multi-level learning. (3) At the core of the model is the core quality of teachers, that is, the role expertise in psychology [10]. Teacher learning has three characteristics: independent self-concept, rich individual experience, realistic demand and problem-solving drive. First of all, teachers as learners have independent self-concept. As adults and educators, teachers have gradually formed a mature self-concept from their previous growth and teaching experience, and have
856
J. Ma et al.
a clear understanding of their own abilities, personalities, attitudes, etc., so as to have an accurate grasp of the reasons and goals of their own learning. Secondly, teachers as learners have rich individual experience. The rich experience and corresponding practical knowledge that teachers have accumulated in the past learning process, social life and teaching career is an important basis for mobilizing teachers’ thirst for knowledge. Finally, in addition to teachers as learners, teachers have multiple roles such as managers, organizers, communicators, etc., which endow teachers with multiple responsibilities such as caring for students, communicating with parents and leaders. In the process of assuming these responsibilities, teachers are faced with a variety of situational problems. The need to solve these problems, to a certain extent, urges teachers to actively participate in, consciously and automatically learn. 2.2 College Foreign Language Teacher Learning in the Context of AI In the field of teacher learning research, AI brings a series of benefits by virtue of the advantages of big data computing. The typical applications of AI in the field of education include intelligent tutors to assist personalized teaching, educational robots, intelligent evaluation of real-time tracking and feedback, etc. In this research, foreign language teacher learning in the context of AI refers to foreign language teacher learning assisted by AI devices, which mainly include three categories: robots, professional software and online learning platforms. Robots, such as iFLYTEK translator, chat robot, etc.; Professional software, such as Google Translation, Kingsoft PowerWord, etc.; Network teaching platform, such as Chaoxing Fanya, moocs.unipus.cn, etc. In this study, foreign language teacher learning in the context of artificial intelligence is to investigate teacher learning based on the use of the above three types of devices with different degrees of intelligence.
3 Research Design This study attempts to answer the following questions: (1) What are the constituent factors of college foreign language teacher learning in the context of AI? (2) What are the characteristics of foreign language teacher learning in the context of AI? (3) What are the effects of AI on college foreign language teacher learning? 3.1 Subjects The research includes two stages, questionnaire stage and interview stage. Subjects in questionnaire stage are 148 college foreign language teachers, with an average teaching year of 11. The average time for participants to use the three types of artificial intelligence language learning devices: robots, professional software and online learning platforms is 2.33 years. Based on the principle of maximum differentiation, 10 foreign language teachers from universities in different areas are interviewed (see Table 1).
College Foreign Language Teacher Learning
857
Table 1. Basic information of subjects Subject
Gender
Age
Teaching Experience
Degree
Academic Title
University Type
Area
T1
Female
41
14
Master
Lecturer
PGU-C
Central
T2
Male
50
24
Doctor
Professor
PGU-C
Southwest
T3
Male
39
15
Master
Associate Professor
PGU-C
East
T4
Female
25
1
Master
Assistant
PGU-C
Central
T5
Female
34
8
Master
Lecturer
PGU-F
South
T6
Female
40
15
Master
Lecturer
DFU-C
Northeast
T7
Male
40
12
Doctor
Lecturer
PMKU-C
East
T8
Female
39
14
Doctor
Associate Professor
DKU-SE
Central
T9
Female
33
2
Master
Assistant
PGU-C
Central
T10
Female
34
8
Master
Lecturer
PGU-C
Central
In the analysis of basic information about subjects, there are differences in teachers’ gender, teaching experience, degrees, professional titles, school types and regions. The age distribution of participating teachers is uniform (1 teacher over 50 years old, 3 teachers between 40 and 49 years old, 5 teachers between 30 and 39 years old, and 1 teacher between 25 and 29 years old). The teaching age covers the early career (2 teachers between 0 and 3 years old), the middle career (7 teachers between 4 and 20 years old), and the late career (1 teacher above 20 years old). The proportion of gender (30% male and 70% female), degree (70% doctoral and 70% master) and professional title (20% junior, 50% intermediate, 20% associate senior and 10% senior) of teachers participating in the program is consistent with the basic situation of foreign language teachers in China’s universities (Ministry of Education, 2017). In order to facilitate the concise expression of the University type in the table, “Provincial General University” is abbreviated to PGU, “Provincial Key University” is PKU, “Provincial and Ministerial Key Universities” is PMKU, “Double First class” Universities “is DFU, and “Comprehensive” is represented by the letter C”, “Normal Education” is NE, “Finance” is F, and “Science and Engineering” is SE. 3.2 Instruments The questionnaire “Teacher Learning Questionnaire” is composed of two parts: the first part is background information, including teaching age, gender and usage time of AI language learning equipment; The second part is the “teacher learning questionnaire”, which has 23 items in total. The questionnaire is compiled after referring to Sun Chuanyuan’s “teacher learning questionnaire” and combining the specific situation of college teachers. The questionnaire adopts the Likert 5-point measurement method from “completely
858
J. Ma et al.
inconsistent” to “completely consistent”. The data collected from the questionnaire was processed using SPSS 26.0. In-depth interviews with 10 teachers were conducted through WeChat and QQ voice calls, involving the teacher learning of the interviewees, including the content and mode of learning, learning objectives, factors affecting learning, and the impact of AI on teacher learning. Interview questions are as follows: (1) (2) (3) (4)
Please describe your learning content, learning duration and schedule. What goals do you want to achieve or achieve through learning? Please describe the people or things that affect you in the learning process. Do you use AI in your learning? If so, what is it? What effect does it have on your learning?
3.3 Data Collection and Analysis The questionnaire was distributed online through the website https://www.wjx.cn in December 2022. A total of 150 questionnaires were collected. After eliminating the questionnaires with high repetition rate and abnormal response time, 148 valid questionnaires were retained with a recovery rate of 98.67%. The interview was recorded and translated into words with the consent of the teachers interviewed in January 2023. The transcribed content shall be further confirmed by the teacher. In addition, the researcher also collected the personal profiles, published academic papers, weblogs and other materials of the interviewed teachers on the school’s official website to supplement and verify the interview data. Data analysis refers to the practice of Wen Qiufang and Zhang Hong [11]. First, the bottom-up grounded analysis method is used to complete the first and second level coding, and then the theoretical constructs in the model are used for the third level coding. In the encoding process, the original text shall be used as the encoded byte to describe or summarize the data as much as possible (See Table 2). Table 2. Example of text analysis Coding
Text Example
First Level Coding
You can also know what is happening in the world by learning the latest foreign magazine… Learning is a channel to keep pace with the times and is helpful for class teaching… On the one hand, I am very lazy and don’t want to work hard. On the other hand, I still have a desire for new knowledge. I also want to make progress and learn some new knowledge. Once a week, I am not oppressed. If I can accept it, I can also implement my love of learning (continued)
College Foreign Language Teacher Learning
859
Table 2. (continued) Coding
Text Example
Second Level Coding Perception of environments, awareness of behavior change, awareness of ability improvement, supplement and update of beliefs, positioning of professional identity, sense of responsibility for undertaking a mission, and description of personal character Third Level Coding
Environmental factors, changes in behavior, ability and belief, professional identity, teacher’s mission and personal/Core quality
Theme
Learning objectives and content, learning process and influencing factors
4 Results and Discussion 4.1 Factors of Foreign Language Teacher Learning The first author of this article verified the structural validity and reliability of the Teacher Learning Questionnaire. The KMO value of the scale data is .811, and the associated probability value of the Bartlett sphere test is .000, indicating that the data is suitable for factor analysis. Exploratory factor analysis adopts principal component analysis method, and performs factor analysis on the data according to the maximum variance method. The load value of each item is above .40, and there is no item with a load above .40 on both factors, which is classified into three factors, explaining 64.309% of the total variance. The internal reliability analysis of the scale shows that the overall Cronbach’s alpha coefficient of the scale is .872, and the Cronbach’s alpha coefficients of the subscales are .906, .770 and .780 in order. The questionnaire structure is reliable. The three factors are named Intrinsic Motivation, Organizational Support, and Feedback and Evaluation according to variance contribution rate (see Table 3). Table 3. Descriptive statistics of the teacher learning Scale
Factor
Mean
Std. Deviation
Cronbach’s alpha
Teacher Learning
Intrinsic Motivation
4.201
0.089
.906
Organizational Support
3.363
0.134
.770
Feedback and Evaluation
3.556
0.167
.780
According to Oxford and Burry-Stock’s classification standard of Likert’s 5-point scale, the average value is equal to or higher than 3.5 for high frequency use, and the average value is introduced. The level between 2.5 and 3.4 is moderate, and the average value is equal to or lower than 2.4 is low-frequency use [12]. Table 3 shows that the participants’ Intrinsic Motivation and Feedback and Evaluation reach the highfrequency experience level. First, teachers’ the high-frequency experience on Intrinsic
860
J. Ma et al.
motivation corresponds to andragogy. Andragogy incorporates elements of cognitive and social constructivism, is problem-centered, goal oriented, practically-based, and although sometimes mandated (e.g., through an employer), usually self-initiated [13, 14]. Notably, it is a systematic and self-sustaining learning activity that is primarily based on self-motivation and commitment. The realism is that in all learning activities, intrinsic motivation, is an unavoidable paradigm that evokes and sustains effective learning, and as a result, positive job performance [15]. Second, the high level of teachers’ experience of the Feedback and Evaluation dimension indicates that evaluation is an important factor to test the achievement of learning objectives and to maintain learning persistence. One of the important objectives of teachers’ learning is to promote teaching, and the improvement of students’ performance is an important and direct way to promote teachers’ learning. Evaluating teacher’s performance is essential to improve learning system as it is related to student’s learning [16]. 4.2 Characteristics of Foreign Language Teacher Learning The three factors explored in the questionnaire research show that teacher learning presents three orientations: (1) ecological orientation. (2) practice orientation. (3) individual orientation, and this finding is also supported in the interview. Intrinsic Motivation, Organizational Support, and Feedback and Evaluation represents the process of teacher learning and reflects the interaction between external education environment and internal motivation. Specifically, ecological orientation is manifested as the purpose of adapting to the social and educational environment. For example, T1 describes two purposes of learning: one is to adapt to the needs of social development and growth, the other is because “I am a little scared, I always feel that English is going backwards quickly, afraid of being eliminated”; the second is to “know what is happening in the world, feel disconnected from the world, and the ability of information retrieval and reception is declining. Learning is a channel to keep pace with the times”. The practice orientation shows that the learning goal is to solve practical problems. For example, T3 believes that learning is “helpful for class. After taking over extensive reading, the textbook has not made a breakthrough. This is just a supplement for students, who also need it. More and more foreign articles are selected for reading in the grade examination. Students are more interested in this than textbooks”. This is consistent with the research on adult learning theory of teachers’ learning motivation, that is, as adults, one of the main learning characteristics of teachers is the internal drive based on problems and practical needs [17, 18]. Individual orientation shows that teachers have a high degree of self cognition and independent self-concept. These are consistent with the theory of Korthagen’s teacher reflection model. The purpose of teacher learning is to adapt to the surrounding environment, and then changes in behavior and belief. The dimensions of self-identity are also more abundant. The most important thing is to cause teachers to reflect on themselves and think about the quality of their own advantages. For example, T1 thinks that she “loves learning, so she studies. To be honest, on the one hand, she is lazy and does not want to work hard. On the other hand, she is eager for new knowledge, wants to make progress, and wants to learn some new knowledge. She does not oppress me once a week. If she can accept it, she can also implement his love of learning”. This is consistent with the discovery that teacher learning has “independent self-concept”.
College Foreign Language Teacher Learning
861
4.3 The Effects of AI on Foreign Language Teacher Learning 4.3.1 Learner Feature Mining and Intelligent Modeling Supported by AI Through the collection, analysis, aggregation of multiple heterogeneous data, and the use of the information complementary mechanism between multimodal data, the external learning state and internal cognitive structure of learners can be characterized. Its logic is through multimodal data perception, learner state representation, and deep feature mining. AI focuses on teachers as learners’ feature mining, provides personalized learning services for learners, and optimizes the models and strategies of personalized learning support services. AI uses teachers’ learning evaluation data, psychological evaluation data, external behavior data, physiological information data and human-computer interaction data to realize intelligent evaluation and analysis of learners’ behavior, knowledge, cognition, emotional state, and explore potential characteristics of learners’ learning interest, learning preference, learning motivation, learning style, etc. For example, T1 describes that “mobile phone and network search and information retrieval functions are used. Artificial intelligence should be used in Rubik’s Cube dubbing. Each sentence you dub will give you a score. If the pronunciation is inaccurate, it will be marked as artificial intelligence. A little bit (similar to machine review), but at the beginning, it was a machine review, but only scores. Now it has been upgraded. The wrong words are marked red, the length of rhythm stress is marked yellow, and the correct green. Then add scores. “ 4.3.2 Normalization and Innovation of Teacher Learning Maintained by AI Artificial intelligence technology supports the construction of a new mechanism for the representation, aggregation and supply of teachers’ learning resources, and provides technical support for teachers’ learning embedded work and continuous learning. First, AI has built a structured learning content system. The knowledge map is used to identify the upper and lower relationship, implication relationship, and fore and aft relationship between teachers’ learning content, and AI is used to build a teacher’s learning content system with clear structure and strict logic; Second, man-machine cooperation supports intelligent learning resource annotation. AI can solve the problems of diversified presentation methods of massive learning resources at the current stage, unclear resource labeling, and unclear logical relationship between resources, carry out humancomputer collaborative classification, labeling and evaluation of educational resources, build a dynamic mapping relationship between learning resources and teaching practice, achieve quantitative evaluation of potential characteristics of learning resources, and provide support for systematic reorganization and personalized push of learning resources; Third, AI pushes intelligent learning resources based on learners’ level and needs. AI builds an association analysis model between “learner knowledge resources”, realizes intelligent matching of learning resources and learning needs, and provides accurate resource push according to specific practice situations, learners’ needs, learning styles and other characteristics. For example, T9 believes that AI plays the role of “resource guarantee service provider” in her learning. She, a young teacher with two years of teaching experience, pays attention to the teacher’s classroom language, teaches the course Introduction to Chinese Culture, and often uses the retrieval function. These
862
J. Ma et al.
websites and applications will occasionally push her information about China’s excellent cultural traditions, cultures of various countries Professional knowledge, ideological and political elements and other related topics; T4 focuses on “learning new teaching methods, including translation software, machine translation, etc.”. Her English learning software will push relevant articles according to her interests, and will also form a list of new words, develop regular review plans, etc. 4.3.3 Teachers’ Collaborative Learning Promoted by AI Artificial intelligence, through the virtual teaching and research room, has contributed to a network learning community, enabling foreign language (mainly English) teachers in colleges and universities at different career development stages to develop a platform for cross regional, interdisciplinary and normalized collaborative innovation and development with the help of QQ groups, WeChat groups and other media for the purpose of resource sharing and academic exchanges. The influence of online learning community on the ability development of foreign language teachers in colleges and universities focuses on three aspects: teachers’ professional cognition and knowledge enrichment, professional skills enhancement, and professional quality improvement. Under the leadership of the person in charge of the task community, teachers orderly carry out and complete all teaching and research activities and tasks on time, and conduct formal and informal learning in the practice/knowledge community through self interaction, peer interaction and interaction with the community in terms of resources, cognition, emotion, value and behavior. No matter what form of interaction, reciprocity has become its prominent feature. T2 met teachers with the same interest through an online meeting, established and carried out cross regional, interdisciplinary and cross stage cooperation through QQ groups, WeChat groups and other media, and “cooperated with workers who pay attention to English education in rural primary schools and teachers in charge of this work in primary education colleges, and planned to publish a book (about) English education in rural primary schools”. T1 got to know “English Rubik’s Cube Show (dubbing APP) part-time operation, (her job is) selecting articles, typesetting, publishing videos after class, sending notices in groups, etc. through online classes”.
5 Conclusion and Implications Teacher learning has gradually replaced “teacher professional development” as a new concept of teacher education. Artificial intelligence integrates information technology and education, which helps to deepen and expand the research on teacher learning. College foreign language teachers’ learning includes three dimensions: intrinsic motivation, organizational support, and feedback and evaluation, showing the characteristics of ecological orientation, practical orientation and individual orientation. Artificial intelligence technology guarantees the autonomy of teachers’ learning through learner characteristics, supports the normalization of teachers’ learning through effective supply of learning resources, and improves the collaboration of teachers’ learning through virtual teaching and research rooms. Teacher learning in the context of artificial intelligence emphasizes the construction of suitable soil for promoting teachers’ professional development through the change of
College Foreign Language Teacher Learning
863
time, leadership, system, mechanism and teacher culture. The school-based research and practice community promoted in the current practice field all point to the paradigm of ecological transformation. AI promotes the research and analysis of individual teachers as learners and the construction of multidisciplinary teacher learning community. Facing the coming of knowledge economy and the development of information technology, teachers should be “learners” and lifelong learners. Teachers are not only adult learners, but also learners who lead students to learn, learn to teach and pursue their own initiative. The role of teachers as “learners” highlights the proposition that “teachers are the subject of learning”, and highlights the life growth and development of “teachers as human beings”. Acknowledgments. This paper is supported by supported by Special Funding in 2021 of Hubei Educational Science Planning Project (2021ZA15), Humanities and Social Science Research Project of Hubei Provincial Department of Education (QSY17007) and Teaching Research Project of South-Central Minzu University (Jyx16033).
References 1. Miao, P., Xiaoyan, L.: Interpretation of “teacher learning” from the perspective of adult learning theory: returning to the adult identity of teachers. Res. Teach. Educ. 26(6), 16–21 (2014). (in Chinese) 2. Zhnetsky, V.: The social role of intellectuals. Translated by Bin Xiang, pp. 7–8. Nanjing: Yilin Press (2000). (in Chinese) 3. Fan Liming “New liberal arts”: the needs of the times and the focus of construction. China Univ. Teach. (5), 4–8 (2020). (in Chinese) 4. Zhanjun, W., Gang, Q.: New paradigm of education research driven by big data. Peking Univ. Educ. Rev. 16(1), 179–185 (2018). (in Chinese) 5. Guorui, F.: Educational Ecology, p. 27. People’s Education Press, Beijing (2000). (in Chinese) 6. Chuanyuan, S.: Teacher learning: expectations and Realities-Take Shanghai primary and secondary school teachers as an example. Shanghai: Shanghai Normal University (2010) 7. Easton, L.B.: From professional development to professional learning. Phi Delta Kappan 89(10), 755–761 (2008) 8. Fullan, M.: The new meaning of educational change (4th edition), pp. 283–291. Teachers College Press, New York (2007) 9. Lieberman, A., Mace, D.: Teacher learning: the key to educational reform. J. Teach. Educ. 59(3), 226–234 (2008) 10. Gang, Z., Yuting, W., Jingliu, H., Yujuan, L.: How to understand the multiple essence and multiple levels of teachers’ professional learning – a dialogue with Professor Fred Kosagen, a world-renowned educator. Mod. Distance Educ. Res. 3, 32–43 (2021). (in Chinese) 11. Qiufang, W., Hong, Z.: Listening to the voice of young college English teachers: a qualitative study. Foreign Lang. Teach. 1, 67–72 (2017). (in Chinese) 12. Oxford, R.L., Burry-Stock, J.A.: Assessing the use of language strategies worldwide with the ESL/EFL version of the strategy inventory for language learning. System 23(2), 1–23 (1995) 13. Merriam, S., Brockett, R.: The Profession and Practice of Adult Education: An Introduction. Jossey-Bass, San Francisco (2007) 14. Spencer, B.: The Purposes of Adult Education: a Short Introduction, 2nd edn. Thompson Educational Publishing, Toronto (2006)
864
J. Ma et al.
15. Sweden, S.D.M., Brendan, P.K.: A case study on adult and workplace learning. Int. J. Educ. Manage. Eng. (IJEME) 1, 11–19 (2020) 16. Mahfida, A., Nusrat, J.L.: A web based automated tool for course teacher evaluation system (TTE). Int. J. Educ. Manage. Eng. (IJEME) (2), 11–19 (2020) 17. Yanju, J., Shaokang, X.: On the construction of adult learners’ learning motivation. Rural Adult Educ. 8, 7–8 (2000). (in Chinese) 18. Gilakjani, A.P., Leong, L.-M., Ismail, H.N.: Teachers’ use of technology and constructivism. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 5(21), 49–63 (2013)
The Innovation Integration Reform of the Course “Single Chip Microcomputer Principle and Application” Chengquan Liang(B) School of Intelligent Manufacturing, Nanning University, Nanning 530200, China [email protected]
Abstract. The course of Single Chip Microcomputer Principle and Application is a required course for most colleges and universities in electrical engineering and automation, mechanical design and manufacturing and automation, robot engineering, artificial intelligence and other related majors. It is a comprehensive course that integrates the basics of electronic technology, computer language programming, electronic product design, etc. This paper is based on the teaching reform of the principle and application course of single-chip computer with the integration of specialty and innovation, and integrates the innovation and entrepreneurship project into the teaching process of the course. Students are no longer limited to memorizing theoretical knowledge, but combine the learning and practice of the course. The project training of innovation and entrepreneurship has stimulated students’ enthusiasm for learning and confidence in innovation and entrepreneurship. They will be more interested in the course content and their comprehensive ability can be greatly improved. Such reform activities can cultivate more innovative and entrepreneurial talents and lay a solid foundation for their growth. Keywords: One-chip computer · Professional and creative integration · Curriculum reform · Innovation and entrepreneurship training program
1 Introduction To ensure that the course teaching achieves good results, the focus is on continuous construction and continuous optimization [1–4]. Only by clarifying the nature and characteristics of the curriculum and understanding the needs of talent training can we know the direction of curriculum reform and make the teaching process truly play its due role [5, 6]. Microcomputer integrating CPU, RAM, ROM, I/O, etc. with single chip microcomputer [7], its minimum system includes single chip microcomputer chip, power supply circuit, reset circuit of single chip microcomputer chip, clock signal generation circuit necessary for work, display module circuit, key circuit, etc. The minimum system of STC89C52 single chip microcomputer is shown in Fig. 1. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 865–875, 2023. https://doi.org/10.1007/978-3-031-36115-9_77
866
C. Liang
Fig. 1. Minimum System of Single Chip Microcomputer
The course of Single Chip Microcomputer Principle and Application is very practical. Microcontroller occupies a very important position in the field of industrial control, and its application is very extensive, almost everywhere, such as intelligent instruments, household appliances, intelligent robots, industrial control process equipment, aerospace equipment, medical equipment, etc. are widely used [8, 9]. Graduates of electrical, mechanical, robot engineering, artificial intelligence and other majors have many jobs related to SCM, such as electrical equipment development and installation, instrument detection, robot development, etc. Therefore, it is necessary for students to master the course of principle and application of single chip microcomputer [10]. The purpose of implementing the integration of specialty and innovation is to cultivate students’ innovation and entrepreneurship ability, and to reconstruct and design the existing content, so that students can form innovation ability in future work and have a certain entrepreneurial ability [11]. How to take innovation and entrepreneurship as the guidance, take curriculum knowledge as the basis, reconstruct knowledge points in combination with practical projects, and achieve the organic integration of professional education and innovation and entrepreneurship education is the key to curriculum reform [12–14]. Therefore, it is imperative to explore an effective teaching method based on the integration of specialty and creativity in the principle and application of single-chip microcomputer.
2 Main Problems in the Course of Single Chip Microcomputer Principle and Application The traditional teaching mode of the principle and application of single chip microcomputer is the theory teaching in the multimedia classroom and the experimental teaching based on the laboratory verification experiment. Generally, the theory and experimental teaching is carried out with the traditional 8051 chip as the core [15].
The Innovation Integration Reform of the Course
867
In the theoretical teaching, the teaching process of the principle and application of SCM in many schools is shown in Fig. 2. First, introduce the basic knowledge of SCM, including the basic concept, conversion of number system, and the development history of SCM [16]; Secondly, the internal structure and working principle of the hardware of 8051 single chip microcomputer (including the internal composition of the single chip microcomputer, clock circuit, reset circuit and internal memory) are introduced; Thirdly, the instruction system and C language programming of single chip microcomputer are introduced, and the data type, syntax and structure of C language programming are also introduced; Then, it introduces the application of SCM’s internal resources, such as IO structure, interrupt system, timer, serial port, etc.; Finally, the memory expansion and parallel expansion technologies of single chip microcomputer are introduced [17, 18].
Fig. 2. Teaching Process of the Course of Single Chip Microcomputer Principle and Application
In the experimental teaching, we mainly completed several simple experiments as shown in Table 1: (1) The installation and simple application of keil and proteus software; (2) Use SCM to light LED water lamp; (3) Single chip microcomputer independent keyboard, 4 × 4 Matrix keyboard programming; (4) LCD1602 displays simple characters; (5) Single chip external interrupt independent key programming application; (6) The timer is used to realize the buzzer to play a piece of music; (7) Single chip microcomputer realizes serial communication with PC, and multi computer communication between single chip microcomputer and single chip microcomputer; (8) C language programming of analog to digital converter and digital to analog converter [19]. If only theoretical knowledge and operating procedures are taught in a straightforward manner, such teaching theory courses will be boring, the depth of practical training courses will be insufficient, and theory and practice will be disconnected. In the long run, the technology of single chip microcomputer can not be linked with the actual project and is separated from the actual application, so the technology of single chip microcomputer will stagnate, or even go backwards [20]. In the era of “mass entrepreneurship and innovation”, the integration of innovation and entrepreneurship training is almost zero.
868
C. Liang Table 1. Main Experimental Contents of Single Chip Microcomputer Experiment
No.
Serial number
Contents of the experiment
1
Experiment 1
Installation and use of software
2
Experiment 2
LED stream lamps
3
Experiment 3
Independent keys and matrix keyboard
4
Experiment 4
LCD1602 display driven by Single chip
5
Experiment 5
MCU external interrupt
6
Experiment 6
Single chip timer
7
Experiment 7
serial communication of Single chip
8
Experiment 8
Application of ADC and DAC
The theory has not been thoroughly tested. In future employment, there are often poor practical ability, lack of practical experience, and cannot solve practical problems. What you learn in school has little connection with the work content, and the deviation from the talents required by the enterprise is too large [21]. Therefore, teachers should reform their teaching, break through the traditional teaching methods and adopt new teaching methods.
3 Reform of the Course of the Integration of Specialization and Innovation in the Principle and Application of Single Chip Microcomputer In the reform and practice of the course “Principle and Application of Single Chip Microcomputer”, the cultivation of innovation consciousness is taken as the starting point [22]. The innovation and entrepreneurship competitions and professional competitions that students participate in mainly include the China International “Internet plus” Undergraduate Innovation and Entrepreneurship Competition (referred to as “Internet plus” competition), Undergraduate Innovation and Entrepreneurship Training Program, National Undergraduate Electronic Design Competition, National Undergraduate Smart Car Competition, China Undergraduate Engineering Practice and Innovation Ability Competition (referred to as “Work Training Competition”), Guangxi University Innovation Design and Production Competition, etc. The projects and topics involved in innovation and entrepreneurship competitions and professional competitions are often very new, and many innovative projects emerge every year. The curriculum reform is based on the innovation and entrepreneurship competition and professional competition projects to rebuild the projects suitable for curriculum teaching. The main reconstruction projects are: (1) design of multi-function LED water lamp; (2) The design of intelligent car race timer; (3) “Explorer” smart car design; (4) Design of waveform acquisition and display system based on PC display.
The Innovation Integration Reform of the Course
869
3.1 Design of Multi-function LED Stream Lamp The reference circuit of the multi-function LED water lamp design project is shown in Fig. 3. In the figure, P1 port of single chip microcomputer is connected to matrix keyboard, P3.2 - P3.5 is connected to independent keyboard KEY1 - KEY4, and P0 is connected to LED1 - LED8.
Fig. 3. Reference circuit of multi-function LED flow light
Project purpose: Learn the basic knowledge, internal structure and working principle of single chip microcomputer through the design of LED water light, understand the relationship between various decimal systems, complete the C language design of multiple modes display of LED water light driven by IO port, and complete the independent key press and matrix keyboard to change the LED water light mode. Project requirements: (1) Press KEY1 on the independent keyboard to display binary numbers 10100101, 00001111, etc. on the LED 1-LED 8 water lamps; (2) Press KEY2 on the independent keyboard to turn on the LED 1 - LED 8 running lights in turn (using an array); (3) Press KEY3 on the independent keyboard to realize the display of LED 1 - LED 8 running lights with different requirements (each student is required to have different LED display functions); (4) Press KEY4 on the independent keyboard to enter the matrix keyboard to control the flow light speed mode (it is required to use the matrix keyboard to set at least 10 levels of different speeds). 3.2 Design of Intelligent Car Competition Timer Since 2017, our school has organized students to participate in the National College Students’ Smart Car Competition, in which a very important device - timer has been used. Therefore, LCD1602 display, external interrupt, timing counter knowledge, etc. are integrated into the SCM course with the timer of the competition as the starting point, which not only stimulates students’ enthusiasm for the competition, but also improves
870
C. Liang
their interest in learning. The reference circuit diagram of the intelligent car race timer is shown in Fig. 4.
Fig. 4. Reference Circuit of Smart Car Race Timer
Project purpose: Through the design of timer, understand the basic knowledge of timer sensor, realize the independent key start, pause, and reset timer, display the timing time with LCD1602, use the internal timer counter of SCM to achieve timing, and use the SCM counter to record the number of turns. Project requirements: (1) LCD1602 is used to display the time of the timer; (2) Press KEY1 on the independent keyboard to start, pause and reset the timer (using the external interrupt function); (3) Press KEY2 on the independent keyboard to increase the number of turns (using the external counting function of the SCM timer counter); (4) Press KEY3 to reset the number of cycles. 3.3 Design of “Explorer” Smart Car It is an effective way to apply the learned knowledge to the practice link, so that students can really master the knowledge and remind themselves of their abilities and comprehensive qualities [23–25]. The history of education and countless facts can prove the important value of practice for knowledge learning [26]. The design of the “Explorer” intelligent vehicle originates from the project of “All terrain engineering vehicle design and production” of the 2020 engineering training competition. The design gives full play to the innovative thinking of students. The smart car passes through narrow bridges, pipes, stairs, grid carpets and other obstacles, and its performance is determined by the number and time of obstacles. The reference diagram of “Explorer” smart car race track is shown in Fig. 5.
The Innovation Integration Reform of the Course
871
Fig. 5. Reference Diagram of the “Explorer” Smart Car Race Track
Project purpose: through the design of the “Explorer” smart car, master the working principle of the smart car tracking sensor and the tracking C language programming, the working principle of the smart car obstacle avoidance sensor and the obstacle avoidance C language programming, and the timer to achieve PWM control. Project requirements: (1) 2–10 infrared sensors are used to track the black line; (2) 1–3 obstacle avoidance sensors are used to avoid obstacles; (3) PWM is used to control the intelligent vehicle to travel at different speeds in different sections. 3.4 Design of Waveform Acquisition and Display System Based on PC Display The reference circuit diagram of waveform acquisition and display system design based on PC display is shown in Fig. 6. This circuit integrates functions such as ADC waveform acquisition, DAC waveform restoration, and serial port transmission of waveform data. It is a highly comprehensive project.
Fig. 6. Waveform acquisition and display system circuit based on PC display
Project purpose: Through the design of waveform acquisition and display system based on PC display, master the principle and programming of serial port transmission, master the waveform collected by ADC, and master the waveform output by DAC.
872
C. Liang
Project requirements: (1) Adopt 1Hz sine wave for ADC acquisition and transmit it to PC through serial port; (2) Single chip microcomputer controls DAC to restore the waveform collected by ADC. 3.5 Matrix of Teaching Content Based on the Reform of Integration of Specialty and Creativity The teaching content based on the curriculum reform of the integration of specialty and creativity is shown in Table 2. Table 2. Teaching content based on the curriculum reform of specialized and creative integration of single-chip microcomputer
Design of multi-function LED stream lamp Design of intelligent car race timer Design of “Explorer” Smart Car Design of waveform acquisition and display system based on PC display
Basic Structure C IO Interrupt Timer Serial ADC/ SC knowledge and programming structure system port DAC expansion working principle √ √ √ √
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
√
Based on the innovation and entrepreneurship competition and professional competition projects, the curriculum reform of single-chip microcomputer based on the integration of specialty and creativity integrates the knowledge points and most experiments of traditional theoretical teaching, combines theory with practice, implements students’ innovation and entrepreneurship, and cultivates students’ innovation ability to achieve twice the result with half the effort.
4 Phased Achievements of the Reform of Integration of Professional and Creative Work Through the reform of the integration of specialty and creativity in the course of the principle and application of single-chip microcomputer, we can expand and extend the teaching project based on the college students’ electronic design competition, college students’ smart car competition, college students’ engineering training competition,
The Innovation Integration Reform of the Course
873
Guangxi college innovation design and production competition, and encourage students to actively participate outside the class. From 2019 to 2022, students from the Intelligent Manufacturing School of Nanning University participated in various college student competitions. Among them, our students have achieved good results in the National Undergraduate Electronic Design Competition, the National Undergraduate Smart Car Competition, and the Guangxi University Innovation Design and Production Competition, which are among the best in Guangxi. The results of students from the School of Intelligent Manufacturing of Nanning University in recent three years are shown in Table 3. Table 3. Some competition results of students of Intelligent Manufacturing College of Nanning University in recent three years Items
Undergraduate Electronic Design Competition
Awards 2 National second prizes; 4 District level first prizes; 5 District level second prizes; 8 District level third prizes
Undergraduate Smart Car Competition
Guangxi University Innovation Design and Production Competition
One national second prize; 2 district level first prizes; 2 district level second prizes;
3 district level first prizes; 3 district level second prizes; 4 district level third prizes
5 Conclusion Traditional SCM teaching has many problems, such as boring theoretical courses, insufficient depth of practical training courses, and disconnection between theory and practice. Reform is imperative. Exploring the reform of integration of specialty and innovation in the course of principle and application of single chip microcomputer, based on the case of innovation and entrepreneurship competition and discipline competition, reconstruct several projects suitable for the teaching of single chip microcomputer, and effectively integrate the teaching content of single chip microcomputer into the project. This paper introduces the purpose and requirements of LED water light, competition timer, “Explorer” smart car, waveform acquisition and display system and other projects, and gives the project knowledge matrix (Table 2). Of course, these projects are not immutable, but constantly add new projects and new knowledge. The innovation integration reform of the principle and application of single-chip microcomputer curriculum is based on the cultivation of students’ project practice ability and innovation consciousness. It carries out the research and optimization of the teaching of single-chip microcomputer curriculum. In the teaching, it uses a variety of teaching methods and means to effectively combine the innovation ability with the curriculum system, effectively mobilizing the students’ initiative and enthusiasm for learning, which not only improves the students’ innovation ability, but also improves their theoretical knowledge and practical ability. This will provide a reference for the teaching of SCM
874
C. Liang
courses in various colleges and universities. In the era of “mass entrepreneurship and innovation”, we will continue to move forward and actively explore a more effective way to reform the principle and application courses of single chip. Acknowledgment. This paper is the phased research result of the third batch of specialty innovation integration curriculum reform project of Nanning University, “Research on Teaching Reform of Single Chip Microcomputer Principle and Application Based on Specialty innovation Integration” (No. 2021XJZC03).
References 1. Yuqing, H.: The network course construction of microcomputer principles and applications. Int. J. Educ. Manag. Eng. (IJEME) 1(1): 51–57 (2011) 2. Ning, G.: Construction and implementation of innovation computer network practical teaching system. Int. J. Educ. Manag. Eng. (IJEME) 1(2), 30–35 (2011) 3. Denggao, G., Jinghui, L., Jianping, L., Qiaoming, Y., Junfeng, L., Mei, Y.: Material physics & chemistry quality network curriculum construction and teaching practice. Int. J. Educ. Manag. Eng. (IJEME) 2(5), 24–30 (2012) 4. Wang, X., Cao, Z.: Discussion reform of forestry panorama course teaching. Int. J. Eng. Manuf. (IJEM) 2(12), 41–45 (2012) 5. Li, D., Zhu, W., Chen, Z., Wang, C.: Research of overall optimization based on a series of data structure courses teaching content. Int. J. Educ. Manag. Eng. (IJEME) 2(11), 18–23 (2012) 6. Xu, X.: Training strategy of key innovation and entrepreneurship talents. Educ. Asia-Pac. Reg. 60, 295–339 (2021) 7. Min, X.: Design and implementation of buzzer based on stc89c52 single chip microcomputer. Wireless Internet Technol. 19(16), 95–98 (2022). (in Chinese) 8. Haiyan, H.: Practical research on teaching reform of single chip microcomputer technology course for vocational undergraduate students based on OBE concept. Mod. Vocat. Educ. 39, 62–65 (2022). (in Chinese) 9. Huijun, L.: Reform and exploration of project-based SCM course. Educ. Teach. Forum 43, 37–40 (2022). (in Chinese) 10. Yanhong, Y.: Research on teaching reform and practice of single chip microcomputer course integrating ideological and political education. Comput. Knowl. Technol. 18(29), 153–155 (2022). (in Chinese) 11. Andrea, G.: Entrepreneurship and innovation design in education. an educational experience to train the new entrepreneurial designers. Des. J. 22, 203–215 (2019) 12. Qichun, J., Yanli, Z.: Teaching reform of integrated curriculum design of single-chip microcomputer and embedded system under OBE-CDIO engineering education mode. J. Nantong Vocat. Univ. 35(02), 57–61 (2021). (in Chinese) 13. Jianhua, Q., Zhicheng, Z., Zhang Xiong, W., Xiaojia.: Integrated teaching reform of “Microcomputer Principle” and “Single Chip Application” courses based on OBE concept. China Elect. Power Educ. 06, 63–64 (2021). (in Chinese) 14. Dalin, Z.: Exploration of PAD teaching mode in teaching reform of SCM application technology course. J. Sichuan Vocat. Tech. Coll. 31(02), 10–13+49 (2021) (in Chinese) 15. Fen, W.: Course construction and practice of “Microcontroller Application Technology” based on the integration of learning, learning and teaching. Jiangsu Educ. 21, 69–74 (2021). (in Chinese)
The Innovation Integration Reform of the Course
875
16. Hongru, W., Wei, T., Houlian, W., et al.: Teaching reform and practice of “Microcontroller” course for mechanical major based on OBE concept. Sci. Technol. Innov. 05, 66–68 (2021). (in Chinese) 17. Anyu, Z., Jun, W., Yanping, F.: Discussion on the teaching mode of SCM course based on the cultivation of innovation and entrepreneurship ability. J. Anhui Electr. Eng. Vocat. Tech. Coll. 25(04), 108–111 (2020). (in Chinese) 18. Jie, X., Zhigao, L.: Research on the experimental reform of the principle and application of Single Chip Microcomputer based on the CDIO project background. Comput. Knowl. Technol. 16(34), 141–143 (2020). (in Chinese) 19. Gaoqiang, L., Peng, L.: Teaching reform and practice of the course “Microcontroller Application Technology”. Sci. Technol. Innov. 22, 134–135 (2020). (in Chinese) 20. Yunyu, J.: Reform ideas for the course of single chip microcomputer application technology. Comput. Netw. 46(20), 45 (2020) (in Chinese) 21. Fen, L., Huaizhuang, H., Guangqiang, L.: Exploration on the reform and innovation of the teaching mode of the course “Single Chip Application Technology”. Popular Sci. Technol. 22(09), 127–129 (2020). (in Chinese) 22. Xueming, L., Min. T.: Teaching reform and practice of single chip microcomputer principle and application course. China Educ. Technol. Equipment (17): 101–102+110 (2020) (in Chinese) 23. Yuxiang, Y., Wei, S., Ruipeng, G., et al.: Teaching reform and practice of single chip microcomputer course under the background of professional certification. Educ. Teach. Forum 35, 197–198 (2020). (in Chinese) 24. Yinghua. Z.: Research on the course reform of Single Chip Microcomputer in the application oriented talent training mode. Comput. Prod. Circ. (09): 259+261 (2020) (in Chinese) 25. Shuting, G., Hua, S., Cuili, Y., Fengwei, Z.: Taking students as the center and starting from the course point - course reform and exploration of single chip microcomputer principle and application. Shanxi Youth 21, 60–62 (2022). (in Chinese) 26. Wei, X.: Discovery and practice of EDA experimental teaching reform. Int. J. Educ. Manag. Eng. (IJEME) 1(4), 11–19 (2011)
Discussion and Practice on the Training Mode of Innovative Talents in Economics and Management in Women’s Colleges Zaitao Wang1(B) , Ting Zhao2 , Xiujuan Wang3 , and Chuntao Bai4 1 School of Business Administration, Shandong Women’s University, Jinan 250300, China
[email protected]
2 Library, Shandong Women’s University, Jinan 250300, China 3 Fundamental Science Section in Department of Basic Courses, Chinese People’s Armed
Police Force Logistics College, Tianjin 300309, China 4 Management Co., Ltd., Tianjin 300405, China
Abstract. How to carry out the supply side reform of talent training and improve the social adaptability of applied innovative talent training is also one of the important issues faced by colleges and universities that set up economic management majors. Women’s colleges and universities are an important part of China’s higher education system. It is an inevitable choice for women’s colleges to insist on running schools with characteristics and cultivate innovative talents. This paper explains the connotation of “economic and management applied innovative talents” through the method of “species plus genus difference”, and on this basis, comprehensively reforms the objective elements, content elements and method elements in the talent training model. Under the guidance of the concept of talent training, we have formulated talent training objectives and specifications to adapt to the development of the industry, built a spiral integrated curriculum system and practical teaching system, explored research based teaching methods driven by teaching and research collaboration and school enterprise collaboration projects, and improved and built a diversified practical environment and a team of teachers with rich experience in enterprise management. Through the practice of some majors in Shandong Women’s College, the reform of innovative talents training mode in economic management has achieved good results. Keywords: Women’s colleges · Economic and management applied innovative talents · Cultivation mode
1 Introduction In China, the educational status of women has been greatly improved. According to , which is showed as Table 1, it is not difficult to see from the table that women account for nearly half of all educational stages, and even more than men in the higher education stages of universities and masters. There is no doubt that the development level of women’s higher education © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 876–885, 2023. https://doi.org/10.1007/978-3-031-36115-9_78
Discussion and Practice on the Training Mode
877
reflects the civilization and progress of society [1]. The mission of modern women’s colleges and universities is to strive for women’s equal access to higher education [2]. They have made great contributions to the training of women talents and the promotion of women’s socio-economic status. Women’s universities have become an indispensable part of the higher education system. The channels of cooperation and exchange between women’s colleges and universities are not smooth enough and the fields are not wide enough; the social status and existing value of women’s colleges and universities have not been fully recognized and widely recognized, and the training of innovative talents in women’s colleges and universities is still facing many severe challenges [3–5]. Table 1. Proportion of women in different education stages
Proportion of women (%)
Female postgraduate D 41.32
M U 52. 17
General U and C C 53.9
Adult U and C U 48.74
Secondary education C 61.9
Pri education
H 55.3 3
J
50. 71
Pre education
Pri 46. 46.42 55
Pre 46. 94
Note: Doctor abbreviated as D; Master abbreviated as M; Undergraduate abbreviated as U; College abbreviated as C; High school abbreviated as H; Junior high school abbreviated as J; Primary school abbreviated as Pri; Preschool education abbreviated as Pre.
Women’s colleges and universities emphasize to cultivate women’s comprehensive and applied talents suitable for social needs according to women’s physiological and psychological characteristics. The implementation of applied innovative talent training in women’s colleges and universities not only helps to promote gender equality in the field of higher education, promote the perfect development of higher education itself and cultivate high-level professionals with modern literacy, but also deconstructs gender bias, eliminates gender inequality, builds advanced gender culture, promotes women’s development and promotes gender equality, promoting social civilization and progress will play an important role in leading and promoting [6, 7]. Different talent types should build different talent training models. Therefore, clarifying the connotation and essential characteristics of innovative talents in economic management of women’s colleges and universities plays a key role in studying the innovative talent training model in economic management of women’s colleges and improving the quality of talent training.
2 Definition of Economic and Management Applied Innovative Talents This part mainly explains the connotation of “economic and management applied innovative talents” through the method of “species plus genus difference”. First of all, we should put the defined object of “economic and management applied innovative talents” into a broader concept of “talents”, which is called “species” [8]. Then, find out the unique nature of “economic and management applied innovative talents”, which is
878
Z. Wang et al.
different from other “talents” concepts. This nature is called “poor”. “Economic and management applied and innovative talents” has three concepts: “innovation”, “application” and “economic and management”. The connotation of economic and management applied innovative talents is analyzed according to the “species” - “genus” relationship of talents - innovative talents, innovative talents - applied innovative talents, and applied innovative talents - economic and management applied innovative talents [9, 10]. The so-called talents refer to the workers who have accumulated more human capital in the form of knowledge and ability through learning and practice, and can create more social value in the same working time [11]. The word “innovation” comes from Latin, and its original meaning includes three meanings: update, create new things, and change. Innovative talents refer to those who constantly make breakthrough contributions to social progress or scientific and technological development by comprehensively using their own innovative thinking, innovative ability, and innovative quality and corresponding knowledge [12]. In the process of transforming objective laws into direct social interests, there are scientific “research” categories (requiring academic talents) that transform objective laws into scientific principles and scientific “application” categories (requiring applied talents) that apply scientific principles to social practice to transform them into products (material or non-material). Applied talents refer to the talents who directly apply the knowledge of applied science or newly discovered knowledge to the practice fields closely related to social production and life to seek direct benefits for mankind [13]. It can be seen that applied innovative talents are those who have diversified and overlapping knowledge structure, profound professional and technical ability, strong sense of social responsibility, critical spirit and innovative research consciousness, and can comprehensively use their own innovative thinking, innovative ability, innovative quality and corresponding knowledge to directly apply them to practical innovation activities closely related to social production and life, so as to make breakthrough contributions to the smooth transformation of the whole society. At present, the training objectives and requirements for economic and management talents put forward by colleges and universities are mostly described as: innovative and application-oriented talents who have good humanistic quality and innovative spirit, have a solid theoretical basis of economics (management), skillfully use modern economic (management) analysis methods, and have strong practical ability to analyze and solve problems, and can engage in economic management in economic management departments, government departments, enterprises and institutions. To sum up, economic and management applied innovative talents are innovative talents between academic talents and technical skills applied talents. They have broad basic theoretical knowledge, professional knowledge and technical reasoning ability, strong economic and management awareness, practical ability, independent learning ability and system thinking ability, team cooperation and communication ability, and economic and management design and innovation ability. Secondly, what is “talent cultivation”? Cultivating talents is the primary task of higher education. Seven problems must be solved in talent cultivation: first, the proposition of educational concept; second, the determination of talent training objectives; third, the selection of talent training objects; fourth, the development of talent training
Discussion and Practice on the Training Mode
879
subject; fifth, the use of talent training channels; sixth, the optimization of talent training process; seventh, the institutional guarantee of talent training. It can be seen that talent training is a systematic project, which includes seven elements of talent training: the concept, subject, object, goal, approach, mode and system [14]. Talent training mode is closely related to talent training, but talent training mode and talent training are two different concepts. As a systematic project, talent cultivation involves seven elements of talent cultivation, namely, the concept, subject, object, goal, approach, mode (process) and system. In order to cultivate innovative talents, the above seven elements must be reasonably developed and reorganized at the same time, including further updating the educational concept, further improving the work efficiency of the training subject, further mobilizing the learning initiative of the training object, further defining the training objectives of innovative talents, further innovating the talent training mode, further enriching the training channels, further reforming the training system and optimizing the training conditions. The talent training mode mainly refers to the design and construction of the training process, that is, the consideration of the education subject on the selection of majors and curriculum, the structure and procedures of teaching activities, and the determination of teaching organization and management forms. The so-called “talent training mode” refers to the theoretical model and operation mode of the talent training process designed by the training subject under the guidance of certain education concepts and the guarantee of certain training systems in order to achieve specific talent training goals, which is composed of several elements and has the characteristics of purposefulness, intermediary, openness, diversity and imitation.
3 Analysis of the Elements of the Training Mode of Economic and Management Applied Innovative Talents The talent training mode is the most changeable and dynamic subsystem in the “talent training” system, and it is also the most complex subsystem of the constituent elements. The change of talent training mode is essentially the change of its constituent elements; The innovation of talent training mode is also mainly the innovation or reorganization of each constituent element. Therefore, in order to innovate the talent training model, we must carefully analyze the elements of the talent training model. 3.1 Talent Cultivation Concept The concept of talent cultivation here refers to the educational concept at the macro (college) and micro (teacher) levels, that is, the rational understanding of the training subject on the essential characteristics, target values, functions, tasks and activity principles of talent cultivation, as well as the ideal pursuit of talent cultivation and various specific educational concepts formed by it. Such as quality, teachers and students, teaching, scientific research, activities and evaluation. The concept of talent training aims to answer the questions of “what should talents be in colleges and universities” and “how should talents be trained”. From the perspective of philosophy, the function of talent training concept is to reveal the internal logic and ultimate value of talent training; from the operational level, it aims to guide the talent training process, including the design
880
Z. Wang et al.
and conception of training procedures and links. The concept of talent training has an extremely important influence on the selection and determination of other elements of the talent training model [15]. 3.2 Professional Setting Mode Specialty setting mode is an important part of talent training mode. Major is mainly divided according to disciplines. Generally, major settings can be designed for shape change in terms of setting caliber, setting direction, setting time, setting space, so on and so forth [16]. The professional caliber refers to the coverage of the main disciplines or the basis of the main disciplines and the scope of business specified when the disciplines are divided. Setting direction refers to whether and how much specialized direction is differentiated within the professional caliber to stiffen or activate the specialty. The setting time refers to the time of specialty setting, whether the specialty is determined as soon as the students enter the school or the specialty training is determined after a certain stage of study. Setting space refers to whether there is still room for wandering and possibility of change after students’ majors are determined, and whether students are allowed to change majors, departments, colleges or cross majors, departments or colleges. 3.3 Teaching Evaluation Methods Teaching evaluation is an objective judgment and evaluation of the talent training process and its quality and benefits based on certain standards. Teaching evaluation is an important part of the talent training process, and it is also an effective form to test the effectiveness of talent training and an important means to stimulate teachers and students. The teaching evaluation involves two aspects: the middle evaluation of running a school and the micro evaluation of teaching and learning in teaching. Whether at the macro level or at the micro level, the current problems in teaching evaluation are shown in the following aspects: first, in the scope of evaluation, the evaluation of results is emphasized, while the evaluation of process is ignored; second, in terms of the purpose of evaluation, emphasis is placed on identification, selection and elimination rather than feedback, correction and regulation; third, on the basis of evaluation, pay more attention to the score of the examination than to creative thinking and practical ability; fourth, in terms of evaluation methods and means, we should pay more attention to examinations than other methods and means. This evaluation method restricts the teachers and students’ independent choice of teaching and learning, constrains the free development of human personality, and cannot well adapt to the requirements of cultivating innovative talents. From single evaluation to multiple evaluation are inevitable, for the innovation of evaluation methods in training innovative talents.
4 Difficulties in the Implementation of Innovative Talent Training Mode in Economic Management The School of Business Administration and the School of Innovation and Entrepreneurship are the two players in the game. Because of the administrative structure of their own policies and systems, inter-departmental competition is inevitable. At the same time,
Discussion and Practice on the Training Mode
881
because participating in the issue of policies to solve major problems can enhance the power and authority of the department, the department will still choose to cooperate with other departments. The two colleges are both interconnected and relatively independent. Although they are smaller than the main body of the school, they can still serve as a unit for pursuing interests. Generally speaking, any institution pursues the maximization of its own interests, and colleges are no exception. The pursuit of departmental interests is the main driving force of its actions, and is also the core reason for departmental conflicts and inter-institutional games. In view of this, it is assumed that each participant is self-interested, and the purpose is to maximize its own interests, without caring about the interests of another participant. Both sides are in a cooperative relationship when cultivating talents, but the resources they invest in the process of cooperation can be controlled, and they are faced with the choice of more investment and less investment. The School of Business Administration is now marked as G and the School of Innovation and Entrepreneurship as C. There is a strategic relationship between the two. G invests more and C invests less, and C gains more income with less investment; on the contrary, if G invests less and C invests more, C will get less income with more investment. Or if both participants invest less resources, the training effect of innovative talents is poor, and both of them suffer losses. Both participants have invested more resources. At this time, the training effect of innovative talents is better, and both of them have obtained better benefits. The game problem between the two is expressed in Table 2 as follows with payoff table: Table 2. Game between innovative talent training colleges
Player C Player G
More investment
Less investment
More investment
50, 50
-100, 80
Less investment
80, -100
-50, -50
Use mathematical expectation to analyze the income of the School of Business Administration: EGM =
1 1 × 50 + × (−100) = −25 2 2
(1)
1 1 × 80 + × (−50) = 15 2 2
(2)
EGL =
Compare the expected payoffs in two cases: EGM < EGL
(3)
It can be seen that the best strategy of the School of Business Administration is to invest less resources.
882
Z. Wang et al.
Similar to the expected income of the School of Innovation and Entrepreneurship: ECM =
1 1 × 50 + × (−100) = −25 2 2
(4)
1 1 × 80 + × (−50) = 15 2 2
(5)
ECL =
Compare the expected payoffs in two cases: ECM < ECL
(6)
It can be seen that the best strategy of the School of Business Administration is to invest less resources. From the above analysis, it can be seen that the best strategy for both the School of Business Administration and the School of Innovation and Entrepreneurship are to invest less resources. Finally, both parties choose to invest less resources as the Nash equilibrium point. At this point, the total income of the school in the cultivation of innovative talents is the lowest, at - 100, which is less than the total income of one party investing more resources, and the total income of one party investing less resources is 20, which is far less than the total income of both parties investing more resources.
5 Taking the School of Business Administration of Shandong Women’s University as an Example The talent training mode refers to the theoretical model and operation mode of the talent training process, which is designed by the training subject in order to achieve specific talent training objectives, under the guidance of certain education concepts and under the guarantee of certain training systems, and is composed of several elements with the characteristics of system, purpose, intermediary, openness, diversity and imitation. The change of talent training mode is essentially the change of its constituent elements, and the innovation of talent training mode is mainly the innovation or reorganization of each constituent element. 5.1 Thoughts on the Reform of Talent Training Mode Comprehensive reform should be carried out according to the purpose, content and method elements of the talent training model. Specifically, the objective elements include the training objectives and training specifications, the content elements include the curriculum system and its teaching content, and the method elements include the teaching and learning methods, teachers and practice environment. The idea of comprehensive reform of talent training mode is to determine the training objectives based on the training concept. The training specifications are designed according to the training objectives, and the curriculum system is designed according to the training specifications. Under the guarantee of a certain training system, certain teaching and learning methods are adopted according to the requirements of the curriculum system, as well as the construction of the teaching staff and practice environment. The training results reflected after
Discussion and Practice on the Training Mode
883
the implementation of the training means will be transmitted to the society. The evaluation will be carried out according to the feedback of the society on the quality of talent training and the requirements of the school itself. The training objectives, training specifications, curriculum system and training means will be further optimized according to the evaluation results. 5.2 An Integrated Curriculum System Based on Spiral Ability Training According to the talent training concept of “demand guidance, project driven, strengthening practice, teaching and research unity, highlighting innovation, and comprehensive development”, according to the talent training specifications and characteristics, as well as the consistency, completeness, integration, progressiveness, and practicality of the curriculum system, and in combination with our own advantages and characteristics, we will build a talent training oriented, theory and practice oriented quality education is integrated into the training of professional knowledge and ability, and the knowledge, ability and quality are coordinated and unified. The curriculum system is oriented by ability cultivation, supported by centralized practice links, and based on core courses. A group of centralized practice links and core courses cultivate an application or design ability and related knowledge. 5.3 Establish an Effective Evaluation System for Training Innovative Talents in Engineering Application The teaching quality assurance system is an important means to ensure the teaching level and the quality of talent training. It is an all-round and whole process quality management system project, which provides an effective way for talent training evaluation. Build a closed-loop teaching quality assurance system consisting of four subsystems: quality standard system, quality monitoring system, quality information and analysis system, and quality improvement system, promote the scientific, standardized, and informationbased management of teaching quality, and thus improve the quality of teaching and personnel training. The principle in the evaluation process is the training quality standard. A quality standard system covering the main process of undergraduate teaching activities should be established, with the talent training goal as the highest program, the talent training program as the basic basis, and the daily teaching norms as the code of action. Form an all-round and whole process teaching quality monitoring system to evaluate and monitor teaching quality from multiple aspects. Evaluate the quality of talent training through the quality information and analysis system, and provide the basis for the quality improvement system.
6 Conclusion How to cultivate high-quality innovative talents is a major issue facing higher education. Women’s colleges and universities are mainly specialized in humanities and social sciences, with the goal of cultivating high-quality application-oriented professionals with gender awareness, sense of responsibility, innovative spirit, practical ability, and
884
Z. Wang et al.
comprehensive development of morality, intelligence, physique and beauty, which are required by economic society and the development of women and children. The existing women’s colleges and universities mainly focus on humanities and social sciences, which is consistent with the traditional female disciplines. In terms of major categories, they are mainly concentrated in literature, pedagogy, economics, law, management and other disciplines. Women’s colleges and universities mainly think about what kind of education they should provide for women, such as “providing women with educational forms suitable for women’s characteristics” to “meet the needs of women’s growth and talent”; “cultivate women with the spirit of ‘four self’”; employment oriented; based on the development of female students, compensation education should be carried out for female college students. In terms of specialty setting and course offering, we should think about how to teach courses with women’s characteristics and cultivate talents with women’s characteristics. In view of the problems encountered in the training of application-oriented talents, we should reform the training mode of application-oriented talents, and study and practice effective ways to improve the training quality of application-oriented innovative talents. Acknowledgment. This research has been supported by the 2017 school level teaching reform research project of Shandong women’s University: discussion and practice of innovative personnel training mode of economic management in women’s University, and the Shandong Social Science Planning Project (Grant No. 18CGLJ48, 18CSJJ32).
References 1. Offutt, S., McCluskey, J.: How women saved agricultural economics. Appl. Econ. Perspect. Policy 44(1), 4–22 (2022) 2. Piva, E., Rovelli, P.: Mind the gender gap: the impact of university education on the entrepreneurial entry of female and male STEM graduates. Small Bus. Econ. 59(1), 143–161 (2022) 3. Kyrpychenko, O., Pushchyna, I., Kichuk, Y., et al.: Communicative competence development in teaching professional discourse in educational establishment. Int. J. Educ. Manage. Eng. 4, 16–27 (2021) 4. Hedija, V., Nˇemec, D.: Gender diversity in leadership and firm performance: evidence from the Czech Republic. J. Bus. Econ. Manag. 22(1), 156–180 (2021) 5. Bullough, A., Guelich, U., Manolova, T.S., et al.: Women’s entrepreneurship and culture: gender role expectations and identities, societal culture, and the entrepreneurial environment. Small Bus. Econ. 58(2), 985–996 (2022) 6. Sharm, D.P., Deng, D.K., Tigistu, G., et al.: A cloud based learning framework for eradicating the learning challenges of Ethiopian working professionals, disables and women. Int. J. Mod. Educ. Comput. Sci. 14(2), 41–54 (2022) 7. Robledo, D.M.L.: Programming as an option for females in undergraduate studies. Int. J. Educ. Manage. Eng. 9(1), 1–8 (2019) 8. Majumder, S., Chowdhury, S., Chakraborty, S.: Interactive web-interface for competencybased classroom assessment. Int. J. Educ. Manage. Eng. 1, 18–28 (2023) 9. Ke, Z.: Research on innovation of the training mode of economic and management applied talents in local undergraduate colleges. Res. Higher Educ. Three Gorges 1, 4 (2020). (in Chinese)
Discussion and Practice on the Training Mode
885
10. Ya, T., Yinghu, T., Jian, L.: Research and practice on the “cross-border integration” talent training system for application-oriented undergraduate economics and management majors in the context of new liberal arts -- taking Jinling University of Science and Technology as an example. Res. Manage. Univ. (1), 78–81 (2022). (in Chinese) 11. Adebayo, E.O., Ayorinde, I.T.: Efficacy of assistive technology for improved teaching and learning in computer science. Int. J. Educ. Manage. Eng. 5, 9–17 (2022) 12. Hao, Y.: Research on the cultivation mechanism of high quality innovative talents in high level private universities. Educ. Mod. 5(35), 28–30 (2018). (in Chinese) 13. Adigun, J.O., Irunokhai, E.A., Onihunwa, J.O., et al.: Development and evaluation of a web based system for students’ appraisal on teaching performance of lecturers. Int. J. Inf. Eng. Electron. Bus. 14(1), 25–36 (2022) 14. Nan, L., Yong, Z.: Improvement and practice of secondary school geography teachers’ informatization teaching ability based on the perspective of MOOCs. Int. J. Educ. Manage. Eng. 1, 11–18 (2022) 15. Ahmed, N., Nandi, D., Zaman, A.G.M.: Analyzing student evaluations of teaching in a completely online environment. Mod. Educ. Comput. Sci. 6, 13–24 (2022) 16. Tendo, S.N.: Multimedia pedagogy among literature lecturers in two universities in Uganda post COVID-19. Int. J. Educ. Manage. Eng. 1, 1–9 (2023)
Cultivation and Implementation Path of Core Quality of Art and Design Talents Under the Background of Artificial Intelligence Bin Feng1,2 and Weinan Pang1(B) 1 School of Journalism and Cultural Communication,
Guangxi University of Finance and Economics, Nanning 530001, China [email protected] 2 Cooperative Course of Performance, Film & Animation, Sejong University, Seoul 05006, South Korea
Abstract. AI has driven the process of industrial reform and economic development. China’s huge network population has made Internet big data have a larger application market. A large amount of information data can improve the accuracy of AI operations. With the continuous adjustment of the industrial structure, the industrial division of labor and post responsibilities as well as the training mode of art and design talents are also facing new challenges. Starting from the current situation of art and design talents under the traditional training mode, the construction of the core quality of art and design talents and the implementation path of the core quality of art and design talents, this paper discusses the countermeasures of how to build the core quality of talent cultivation in the era of artificial intelligence. It aims to provide useful ideas and practical experience for China to cultivate high-quality creative talents in art design and strengthen professional construction. Keywords: Artificial intelligence · Art and design talents · Core quality
1 Introduction At the intelligence symposium of American universities in 1956, the professional name of “artificial intelligence” was first proposed by the famous professor John Change, which was also the first time that it was publicly used in academic discussions. In the following period of time, it did not attract the attention of the industry. It was not until more than 20 years later that “artificial intelligence” entered people’s vision again and became the mainstream trend [1]. Although the development of traditional symbols is slow, scientific researchers have made great progress in speech recognition, humancomputer interaction, neural network and other aspects by adopting the more advanced statistical probability model for the first time [2–4]. In today’s increasingly mature computer technology, art and design education also gradually introduces and actively uses high-tech means such as artificial intelligence to cultivate “skilled, rich, diverse and comprehensive” talents on the basis of respecting the principles and laws of traditional art and design education, in order to pursue the best effect of art education [5, 6]. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 886–898, 2023. https://doi.org/10.1007/978-3-031-36115-9_79
Cultivation and Implementation Path of Core Quality of Art and Design
887
In the era of rapid development of mobile network, with the increasing popularity of artificial intelligence technology, the field of big data network services has also been greatly expanded. With the change of people’s information needs, media usage habits and communication behavior, image recognition technology, voice synthesis, humancomputer interactive translation and virtual reality technologies have emerged [7–9]. On August 31, 2022, China Internet Network Information Center (CNNIC) released the 50th “Statistical Report on the Development of China’s Internet Network”, which showed that as of June 2022, the number of Internet users in China reached 1.051 billion, the Internet penetration rate reached 74.4%, and the per capita online time per week was 29.5 h; The proportion of Internet users using desktop computers, laptops, tablets and televisions to access the Internet was 99.6%, 33.3%, 32.6%, 27.6% and 26.7% respectively. A large number of users and media usage habits have promoted the transformation and integration of the industry. The accelerated integration of various industries across fields has accelerated the development of mobile network technology. At the same time, the industrial division of labor, post functions and technical conditions have also changed accordingly [10]. The development of industry and society has put forward a greater and higher demand for the quantity and quality of talents, and also entrusted colleges and universities with a more sacred mission [11, 12]. The training mode of art and design talents based on meeting the needs of economic and social development has also been greatly challenged. While new technologies promote economic and social development, the education authorities should also comply with the trend of AI development and social needs, and carry out all-round changes in the university’s professional setting system, focusing on the construction of emerging professional systems such as AI, art and technology. Data thinking, scientific and technological vision and core literacy have become important factors in today’s scientific and technological innovation [13–15]. For the training of application talents related to art design, a correct analysis should be made from the aspects of knowledge structure, ability level, creativity, etc., which is the prerequisite and basis for cultivating high-level art design talents. The cultivation of core literacy of art and design courses in colleges and universities should grasp and construct its key factors, and the quality of core literacy should be continuously improved on the basis of adapting the application of new technologies to the development of the times [16–19]. To this end, the article discusses how to provide more art and design talents for the development of society and industry and provide a strong guarantee for high-quality teaching development from the perspective of the development of new technology, combining the background of the era of artificial intelligence and referring to the requirements of the cultivation of the core quality of art and design talents [20–22].
2 Challenges Faced by Art and Design Talents in the Era of Artificial Intelligence The development of artificial intelligence technology is changing the standard of demand for art and design talents by employing enterprises. Art and design talents are facing the test brought by this new technological change [23].
888
B. Feng and W. Pang
2.1 AI technology Promotes the Innovation of Art and Design Talents Artificial intelligence technology accelerates the elimination of traditional mechanical design work, and promotes artistic design talents to be more innovative. With the continuous development of artificial intelligence, more intelligent design software is emerging. The emergence of software such as ShapeFactory, Withoomph, Designevo and Markmaker has greatly shortened the threshold of software operation. Not only that, the software itself can collect data according to users’ preferences and provide users with a variety of color schemes. Users have replaced the traditional complex design and other processes by clicking and selecting, and the software operation of simple machines has been replaced by artificial intelligence, which requires contemporary art and design practitioners to have stronger comprehensive ability. In the past, talents trained in art and design education in colleges and universities often used to copy other works directly and simply with the help of Internet resources, resulting in low quality and similar styles of works. Design is only based on mechanical reproduction, and related intellectual property rights are not protected. In contrast, in the context of AI, today’s art and design practitioners should not only have traditional painting and design skills, but also exchange roles between programmers and managers. At a certain level, they should not only understand the relevant technical knowledge, but also manage the team, constantly improve their skills, cultivate their ability to look at macro issues, and also learn computer-related code when necessary. Now more and more art and design practitioners complete the design task with the help of artificial intelligence. Traditional art and design practitioners urgently need to improve their creativity and uniqueness, improve design skills and increase the diversity of design methods. Only a solid design foundation and the ability to learn from others’ strong points can we more calmly cope with the challenges brought by AI. 2.2 Intensified Contradictions in the Employment Market of Art and Design Talents in the Context of Artificial Intelligence The replacement of human beings by artificial intelligence (at least partially) is an irreversible historical process. With the emergence of artificial intelligence, the relationship between human and artificial intelligence has undergone earth-shaking changes. AI not only improves the productivity of human beings, but also tries to replace them. It “robs” some of the work that should be done by human beings in traditional industries, thus replacing the traditional labor force. This is also the difference between the “revolution” of AI and the “machine replacing man” in the previous industrial revolution. With the continuous development of artificial intelligence, the scale of this alternative is bound to grow. In the field of art and design, the number of employers has decreased, and the requirements for art and design talents have increased significantly. In the training of art and design talents in colleges and universities under the traditional training mode, due to the backward curriculum and teaching methods, it is difficult to match the requirements of employers in the context of artificial intelligence. After graduation, a considerable number of art and design talents have chosen other careers as their means of livelihood. In the questionnaire survey of “change of profession” and “not change of profession”
Cultivation and Implementation Path of Core Quality of Art and Design
889
of traditional art and design talents, the proportion of “change of profession” is as high as 46.9%; The proportion of “not changed” is 53.1% (Fig. 1). The high rate of “career change” of nearly half directly proves that there are problems in the cultivation of art and design talents in colleges and universities under the background of artificial intelligence. If this phenomenon does not change, the development prospects of the industry are worrying.
Fig.1. Proportion of career changes of art and design talents
Figure 2 is our monthly salary survey for graduates of art and design major after work.
Fig.2. Monthly salary survey
The figure shows the monthly salary status of the people who have not changed careers, as shown in Fig. 2, the monthly salary “2000–3000 yuan” accounts for 22.5%, the monthly salary “3000–4000 yuan” accounts for 51.9%, the monthly salary “4000– 5000 yuan” accounts for 16.4%, and the monthly salary “5000 yuan or more” accounts for 9.2%. Among them, the monthly income of “3000 - 4000 yuan” accounts for the most, reaching 51.9%, and the monthly income of “5000 yuan or more” accounts for the least, only 9.2%. This shows that most of the traditional art and design talents (68.3%) have a monthly salary of “3000–4000 yuan”, and “2000–3000 yuan” (22.55%) is the normal. Compared with the five-figure salary of Al’s new career in the context of AI era, it is verified that the advent of AI era has indeed lowered the value of traditional art and design talents and reduced their wages.
890
B. Feng and W. Pang
3 Core Quality of Art and Design Talents The Ministry of Education of the People’s Republic of China has repeatedly pointed out that the core quality of students’ development mainly refers to the essential character and key ability that students should have to adapt to the needs of lifelong development and social development. It puts forward the core quality of cultivating “all-round development people”, including three aspects, six qualities and 18 basic points (see Fig. 3). Its basic principles for a long time are to be scientific, pay attention to the times and strengthen the regionality.
Fig. 3. The core quality structure of “all-round development” of students
The overall requirement for the construction of the core literacy of art and design talents is to meet the all-round development of human beings in the process of studying the cultivation of the core literacy of art and design talents under the background of the artificial intelligence era, and to conduct more professional and epochal research on art and design talents. Cultivate them to use the knowledge and concepts they have learned to solve the problems encountered in the era of artificial intelligence, turn the challenges into opportunities, and strengthen the construction of regional, contemporary and scientific core literacy learning system. The cultivation of core competence is the necessary quality and key ability of art and design talents.
Cultivation and Implementation Path of Core Quality of Art and Design
891
3.1 Requirements of Art and Design Talents for Core Quality Training Art design is an aesthetic change that people make artistic innovation on the objective things in life. The discipline of art and design integrates multiple disciplines, involving science and technology, humanities and economy. Its biggest feature is to serve people. Therefore, in the education of art and design, we should pay attention to the training of professional knowledge of art and design and the cultivation of practical operation ability. Colleges and universities pay attention to cultivating students’ practical ability, so the artistic design talents are application-oriented, that is, based on theory and aiming at practice. Therefore, while meeting the daily teaching requirements, the art design course teaching must pay attention to the development trend of the industry’s demand for talents, not only grasp the applied teaching strategy of combining art design skills and theory, but also pay attention to the cultivation of students’ core quality, knowledge, ability, attitude and other comprehensive qualities. 3.2 Elements of Core Quality of Art and Design Talents The core quality of art and design specialty is reflected in the training of art and design talents, and the connotation of its core quality is combined with skill requirements. This paper discusses the core literacy of art and design professionals from four aspects: ideological character, social participation, personal growth, and high-level cognition, and puts forward two basic elements of the core literacy of art and design talents: comprehensive literacy and design practice ability. 3.2.1 Comprehensive Quality Comprehensive quality is the main embodiment of the core quality of art and design talents, mainly including ethics, value orientation, social value and legal knowledge. The ethics here refer to national consciousness, national art, patriotism, moral character and artistic morality. Art and design talents should respect the national customs and cultures of different countries, and pay attention to the integration of modern and traditional. Correct values and popular aesthetic orientation are reflected in the design works of designers. A designer without good moral quality and artistic ethics has no high design taste. Designers should have both creativity and innovation, both rights awareness and legal concepts, and rational use of design materials; Art design has the nature of cultural communication, so the practitioners must have a sense of social responsibility, active publicity, more participation in public affairs, and have more social responsibilities. 3.2.2 Design Practice The practical ability of design is the concrete expression of social participation when cultivating core literacy. Design practice mainly includes three aspects (design expressiveness, unity and cooperation ability and communication and expression ability). Design expressiveness is an important standard to measure the core quality of art and design talents, and also an important standard to measure art and design talents.
892
B. Feng and W. Pang
It includes aesthetic perception, practice and other aspects. Art and design is a process of combining art, technology, theory and practice. Team cooperation ability refers to the cooperation consciousness, organization and coordination ability of team members, while the art design industry is an industry with high demand for team cooperation and is also a team competition. Good design originates from the cooperation and division of labor of an excellent team. Therefore, art designers should not only have strong professional quality, but also have strong coordination and communication ability. Table 1. Evaluation index system of core competence of art and design talents Goal
Level I indicators
Level II indicators
Level III indicators
Weight
Evaluation of core competence of art and design talents
Comprehensive quality c1
Ethics c11
Homeland feelings c111
5
Art morality c112
5
National Art c113
3
Aesthetic orientation c121
4
Diligent in reflection c122
4
Professional certificate c131
8
Competition winners c132
8
Copyright awareness c141
4
Intellectual property c142
4
Innovative design ability c211
8
Innovative capability c212
8
Aesthetic accomplishment c213
4
Values c12
Social value c13
Common sense of law c14
Design practice c2
Design performance c21
Score
(continued)
Cultivation and Implementation Path of Core Quality of Art and Design
893
Table 1. (continued) Goal
Level I indicators
Level II indicators
Level III indicators
Unity and cooperation c22
Sense of responsibility 5 c221
Communication expression c23
Weight
Service awareness c222
5
Risk awareness c223
4
Technology application c231
8
Problem solving c232
8
English ability c233
5
Score
On the basis of the research results of the “Research Results Conference on the Development of Core Literacy of Chinese Students”, in view of the characteristics of the core literacy cultivation ability of art and design talents, according to the principles and purposes of the indicator system design, through a series of important questionnaires and interviews, the factors affecting the evaluation of the core literacy ability of art and design talents are summarized, Finally, it is determined that the indicator elements for the evaluation of the cultivation of the core competence of art and design talents should include the first-level indicators of “comprehensive literacy” and “design practical ability” (as shown in Table 1), which form an organic whole. In the process of practice, the two elements of comprehensive literacy and design expressive ability present an interactive situation, are interrelated and interact with each other. By quantifying the three-level index points, the investigation is carried out according to the A-D index. Among them, if the score is above 95 points (excluding 95 points), it is rated as A, which means “excellent”, if the score is 90–95 points, it is rated as B, it means “excellent”, if the score is 80–89 points, it is rated as C, it means “pass”, if the score is below 80 points (excluding 80 points), it is rated as D, and it means “unqualified”. Through the evaluation of teachers, training objects and relevant enterprises, the interaction between the evaluation subjects is enhanced, the training object is emphasized to be a member of the evaluation subjects, and the evaluation system of mutual participation and interaction between teachers and students is established, so as to promote the development and development of the core competence training of art and design talents with multi-channel feedback information.
894
B. Feng and W. Pang
3.3 Construction of Core Quality Training System for Art and Design Talents 3.3.1 Research Significance of the Core Quality Training of Art and Design Talents It is of great significance to study the cultivation of the core quality of art and design talents. (1) Students’ personal development needs Core quality refers to the basic quality and essential character of cultivating students’ lifelong development and adapting to social needs in the specific stage of education. It is a comprehensive reflection of students’ theoretical knowledge learning, learning ability and learning attitude. Therefore, cultivating the lifelong development ability of students majoring in art design is the most basic quality and skill, and is the necessary condition to move towards and integrate into society. (2) Teaching needs Students majoring in art and design have a wide range of employment fields. Including large, medium and small enterprises and professional design companies in various industries. These enterprises and design companies have different needs. With the progress of the times. The needs of employers are also changing. How to keep up with the pulse of the times and cultivate talents meeting the needs of the industry? These all need to cultivate students’ core quality in the course setting and actual teaching process. Make students have the necessary character and key ability for lifelong development. 3.3.2 Cultivation Concept of Core Quality of Art and Design Talents The cultivation of the core quality of art and design talents needs to be carried out under the guidance of the cultivation concept of innovative talents combined with artificial intelligence thinking. Artificial intelligence thinking logic is considered to be the principle that should be followed in the development of new products of the Internet and mobile Internet, and fully considers user design as its own use concept. With the passage of time and the progress of computer technology, new technologies and new development concepts blend and innovate constantly. In combination with AI technology and training objectives, colleges and universities should reform the training mode of art and design talents guided by social service thinking and platform logical thinking, and strive to improve the core quality of art and design talents, such as the design and production of adaptive learning environment. Colleges and universities can analyze the relationship between the needs of domestic enterprises and students’ abilities based on the thinking logic of big data, scientifically develop curriculum and teaching plans, analyze students’ social conditions in learning in combination with social service thinking, adopt team-based learning methods to cultivate team spirit and practical ability, and enable students to track actual projects through cooperation with design companies, carry out job division and realize the whole process experience of design, The platform logical thinking allows students to conduct online teaching and mixed teaching, creating a scene for students to deeply participate in learning.
Cultivation and Implementation Path of Core Quality of Art and Design
895
3.3.3 Analyze and Refine the Core Elements of Design and Art Talent Cultivation The improvement of knowledge, ability and comprehensive quality is the whole process of training art and design talents. In the process of training art and design talents, we should not only pay attention to the teaching of computer skills and theoretical knowledge, but also teach specific learning methods to students in school, so that students can form the habit of actively acquiring and mastering knowledge, and comprehensively improve cultural quality and personal taste. In the era of artificial intelligence, more attention should be paid to the update of knowledge theory and computer technology, the design of new theoretical knowledge structure and the adjustment of teaching plan should be carried out in combination with the integration and development trend of artificial intelligence technology and art, so as to cultivate students’ scientific literacy and the ability to use big data to learn. At the same time, colleges and universities should further improve the independent learning ability of art and design talents, enhance their opportunities for active learning, improve their practical ability and innovative spirit through an active learning attitude, and achieve the integration of theory and practice. In addition, they should also strengthen the humanistic quality of art and design talents, and explore the corresponding Chinese cultural elements of art and design majors through big data, Refine the core elements of art and design talent training.
4 Research on the Implementation Path of the Core Quality Training of Art and Design Talents Through the research on the implementation path of the core literacy training of art and design talents in the context of artificial intelligence, the overall problem is divided into four sub-problems, and the corresponding research ideas are proposed. It mainly includes four aspects as shown in Fig. 4.
Fig. 4. Implementation framework of core literacy cultivation
4.1 Overall Framework and Specific Content of Core Literacy Research the overall structure of the core quality training of art and design talents. Based on the core literacy of Chinese college students, the core literacy of art and design talents is defined in depth. In terms of the selection of curriculum content, according to the
896
B. Feng and W. Pang
teachers’ conditions and school positioning of each university, a comprehensive survey is conducted on relevant enterprises (large, medium and small enterprises) and graduates (new, previous employment and unemployed graduates) to determine the overall framework of individuality suitable for the cultivation and development of students in this major. 4.2 Teaching Design Strategy for Cultivating Core Literacy Based on the basic quality system and specific content of art design major in colleges and universities, and combining with the characteristics and current situation of students in relevant colleges and universities, this paper analyzes it by using the methods of group discussion, interviews between teachers and students, and paper method. Adjust the professional curriculum. On this basis, we should further improve the curriculum of art design specialty in colleges and universities to make it more in line with the basic requirements of art design education. On this basis, the teaching design of the course is discussed in depth. When carrying out the teaching design reform, the core quality and ability of the students of art design should be reflected. 4.3 Teaching Implementation Strategy of Core Literacy Cultivation In college art design majors, how to effectively cultivate and improve their core quality is an important part of the teaching design of college art design courses, and also the key to the cultivation of core quality. Therefore, the whole teaching implementation process should be reasonably designed to make the design of teaching implementation process reasonable and feasible. The teaching implementation strategy should keep pace with the development of the times, take the design industry consultation, social hot spots and students’ concerns as the starting point, use the information teaching method, mobilize students’ learning initiative, enhance the teaching quality, and cultivate the core quality and ability of art and design students. 4.4 Teaching Evaluation Strategies for Cultivating Core Literacy The research on teaching evaluation strategies for cultivating the core quality of art design talents in colleges and universities aims to further verify the rationality and effectiveness of the research on teaching design and implementation strategies. In terms of research methods, case teaching method, collective lesson preparation, comparative research and other methods are adopted to develop a teaching evaluation model for cultivating the core competence of the major. The establishment of this teaching evaluation model pays attention to multi-angles (students, teachers, experts, enterprises) and enforceability while taking into account rationality.
5 Conclusion With the popularization of artificial intelligence technology, how to cultivate a group of design and art talents that adapt to the times has become an important indicator to measure the quality of higher education in China. The development of new technology
Cultivation and Implementation Path of Core Quality of Art and Design
897
has created a new form of integration of art and science and technology. In the teaching application of art and design majors in colleges and universities, how to improve students’ professional ability while also improving their core literacy is a very meaningful topic. The talent cultivation of art design majors should not only enable students to master the knowledge structure that adapts to the times, but also have a positive creative spirit. Using new technologies such as artificial intelligence, we should cultivate students’ data literacy, scientific and technological ability, accumulate humanistic literacy, condense the spirit of innovation and entrepreneurship, and improve the core quality of talents. Through the research on the implementation path of the core literacy of the construction of art and design courses, it can provide reference for similar domestic colleges and universities in teaching theory and practice research, evaluation system and mechanism exploration, so as to make beneficial exploration for the teaching reform, curriculum design, teaching practice and other aspects of colleges and universities.
References 1. Arora, M., Bhardwaj, I.: Artificial intelligence in collaborative information system. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 14(1), 44–55 (2022) 2. Ramjeea, P., Choudhary, P.: State-of-the-art of artificial intelligence. J. Mobile Multimedia 17(1), 427–454 (2021) 3. Bhaskara, M.: Artificial intelligence: state of the art. Intell. Syst. Ref. Library 172, 389–425 (2019) 4. Ågerfalk, P.J., Conboy, K., Crowston, K., et al.: Artificial intelligence in information systems: state of the art and research roadmap. Commun. Assoc. Inf. Syst. 50(1), 21 (2022) 5. Juana, Q.: Research on artificial intelligence technology of virtual reality teaching method in digital media art creation. J. Internet Technol. 23(1), 125–132 (2022) 6. Chen, C.: Study on the innovative development of digital media art in the context of artificial intelligence. Comput. Intell. Neurosci. 1, 1004204 (2022) 7. Rathi, J., Grewal, S.K.: Aesthetic QR: approaches for beautified, fast decoding, and secured QR codes. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 14(3), 10–18 (2022) 8. Ebenezer, O.: Graphic design principles and theories application in rendering aesthetic and functional installations for improved environmental sustainability and development. Int. J. Eng. Manuf. (IJEM) 9(1), 21–37 (2019) 9. Yunxuan, W.: Application of artificial intelligence within virtual reality for production of digital media art. Comput. Intell. Neurosci. 2022, 3781750 (2022) 10. National Library Research Institute. China internet information center released the 50th statistical report on the development of internet in China. J. Natl. Library 31(5), 12 (2022). (in Chinese) 11. Jian, L., Yuan, Z.: The exploration and practice in innovative personnel training of computer science and technology. Int. J. Educ. Manage. Eng. (IJEME) 2(6), 47–51 (2012) 12. Jiang, Q., Jin, T., Chen, H., Song, W.: Research on cultivating undergraduates in the computer science based on students. Int. J. Eng. Manuf. (IJEM) 10(6), 32–39 (2020) 13. Joseph, E.: Orn The Future of Education: Educational Reform in the Era of Artificial Intelligence. Translated by Li Haiyan and Wang Qinhui. Machinery Industry Press, Beijing (2019). (in Chinese) 14. Sun, Y.: Application of artificial intelligence in the cultivation of art design professionals. Int. J. Emerg. Technol. Learn. (IJET) 16(8), 221–237 (2021)
898
B. Feng and W. Pang
15. Hu, J., Fu, L.: Innovation and development of environmental art design thinking based on artificial intelligence in culture, form and function. In: Jansen, B.J., Liang, H., Ye, J. (eds.) International Conference on Cognitive based Information Processing and Applications (CIPA 2021). LNDECT, vol. 84, pp. 635–642. Springer, Singapore (2022). https://doi.org/10.1007/ 978-981-16-5857-0_81 16. Yanzhang, X.: App interactive experience design thinking model based on “Internet+.” Art and Design (Theory) 2(08), 113–115 (2016). (in Chinese) 17. Jing, C.: Exploration and analysis of teaching strategies for cultivating core literacy of art and design talents in higher vocational colleges. Meihe Times (I) 8, 126–128 (2017). (in Chinese) 18. Xiaoqun, C., Xiaomei, T., Yu, Z.: Research on the construction of the core quality connotation of art and design talents in adult colleges. J. Jiangxi Electric Power Vocational Tech. College, 34(12), 126–128 (2021). (in Chinese) 19. Yanzhang, X.: Analysis of the core quality of design art talents in the context of artificial intelligence. Chinese Art 1, 70–75 (2021). (in Chinese) 20. Jing, C., Chenxi, Z.: Construction of the core literacy system of art and design talents in the context of artificial intelligence. Beauty and Time (I) 9, 123–125 (2021). (in Chinese) 21. Fufeng, C.: Research on art education in the context of AI era – Taking the course of cultural and creative product design as an example. Art Rev 24, 361–362 (2019). (in Chinese) 22. Xiaoli, Y.: Reform and thinking of visual design mode in the context of artificial intelligence. Grand View Fine Arts 10, 131–133 (2020). (in Chinese) 23. Amantha Kumar, J., Muniandy, B., Wan Yahaya, W.A.J.: Emotional design in multimedia learning: how emotional intelligence moderates learning outcomes. Int. J. Mod. Educ. Comput. Sci. (IJMECS), 8(5), 54–63 (2016)
Reform and Innovation of International Logistics Curriculum from the Perspective of Integration of Industry and Education Xin Li1 , Meng Wang1 , Xiaofen Zhou1 , Jinshan Dai2,3(B) , Hong Jiang1 , Yani Li1 , Sida Xie1 , Sijie Dong1 , and Mengqiu Wang1 1 School of Logistics, Wuhan Technology and Business University, Wuhan 430065, China 2 Center for Port and Logistics, School of Transportation and Logistics Engineering, Wuhan
University of Technology, Wuhan 430072, China [email protected] 3 Department of Industrial Systems Engineering and Management, National University of Singapore, Singapore 117576, Singapore
Abstract. International Logistics is a professional course of logistics management and a school-level educational reform course of industry-education collaboration, which is an important support to cultivate international applicationoriented talents. The course has the following pain points: the teaching content is too different from the enterprise practice, the cognitive load is too large due to the complicated combination of concepts and skills, and the knowledge points are too fragmented and fragmented to produce deep learning. With the general idea of school-enterprise collaborative education and deepening the concept of industry-education integration, we innovatively put forward the teaching strategy of “three-dimensional unity”, one dimension: deepening the connection between concepts with theoretical knowledge; two dimensions: defining the cognitive steps with modular implementation; three dimensions: docking with enterprise practice with project system. Through the comprehensive innovation of objectives, contents, cases, resources, process and evaluation, the pain points are effectively solved. At the same time, it pays attention to the integration of course thinking and political elements in the process, including the shaping of values such as patriotism, responsibility, dedication and humanistic care, forming a teaching innovation model adapted to the application-oriented profession. Keywords: International logistics · Industry-education integration · Three-dimensional integration · Teaching innovation mode
1 Introduction International “Logistics” is a course of logistics management in Wuhan Technology and Business College (hereinafter referred to as WTBU), which is a school-level educational reform course based on the integration of industry and education, with a total of 32 h (2 credits), open for junior students. The course aims to cultivate international logistics talents with knowledge, ability, commitment, emotion and warmth for the society. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 899–910, 2023. https://doi.org/10.1007/978-3-031-36115-9_80
900
X. Li et al.
The course includes many abstract international logistics theoretical foundation, but also covers the bill of lading, import and export processes, and international trade practices and other practical areas, and also involves the domestic and foreign research drones technology and intelligent logistics and other frontier areas. If the practical field is the “backbone” of the course, the theory is the “support point”; the frontier field is academic and the practical application needs to be further developed, which is the “compass” of practice. How to carry out the integration of industry and education, the organic integration of theory, practice and frontier is a topic that the international logistics course team has been repeatedly thinking and practicing [1]. The characteristics of WTBU Logistics College are: professional docking with industry, talents serving industry, logistics management is located in the cultivation of “highquality applied talents serving the needs of regional economy”, through the analysis of students’ learning situation research, it is found that students like the professional courses to a high degree, independent thinking ability, but think that the professional and However, they think that the connection between professional and practical work is not close, there are too many knowledge points in the course, they are afraid of difficulties, it is difficult to integrate and apply flexibly, and the depth of study of knowledge points is weak.
2 The Existing Problems of the Curriculum In the past three years, through the evaluation of school students, interviews with graduate students, peer discussion and evaluation, research of similar colleges and universities, feedback from employers, etc., the “pain points” problems that restrict the development of this course have been summarized, mainly in the following three aspects: 2.1 The Teaching Contents Are Different from the Enterprise Practice First of all, international logistics has not investigated the matching of courses and positions in terms of theoretical courses or practical course content, course hours allocation and practical effect evaluation, nor has it investigated the matching of courses and positions, so there may be greater subjectivity, which may be different from the practical requirements of enterprises or positions. Secondly, the design of practical syllabus and practical projects is generally organized, formulated and implemented by the instructors of the course, and lacks the participation of enterprises. In addition, the practical teaching process and assessment are described, guided and assessed by theoretical teachers, which may be different from the actual employment requirements of the enterprise or position. Finally, in various internships, students Although I have experienced it in enterprises (e.g., COSCO, JD.com, SF Express), the internship method mainly focuses on visiting the operation process, equipment, and explanation by enterprise personnel, and lacking practical experience and practical experience. As a result, the adaptation period of graduates is too long, and higher requirements are put forward for non-cognitive goals such as communication, coordination, and management, and students’ career development is restricted [2].
Reform and Innovation of International Logistics
901
2.2 Concepts and Skills Are Cumbersome and Complex that Cognitive Points Are Overloaded Twelve chapters of the textbook, including hundreds of international logistics proper terms, and theories, basic knowledge and skills are scattered and difficult to understand systematically. In addition, students lack intuitive perception of all links of international logistics and “needle threads” of practical cognition, it is difficult for students to learn deeply [3]. The overall logical structure of most international logistics textbooks is “theory → case → exercises”. Although the theory related to international logistics is the basis for the study of professional courses in logistics management, there are many difficult and difficult content such as “maritime bill of lading production” and “use of trade terms”, and students are very prone to be afraid of difficulties. The course knowledge is complicated and the teaching time is limited, so it is impossible to achieve “all-inclusive”. The current teaching method makes most students stay at the “introductory” level, which is not conducive to the systematic ability construction of students. 2.3 Fragmented Knowledge Points Are not Easy to Generate Deep Learning At present, teaching mainly teaches knowledge by teachers, ignoring the cutting-edge development of disciplines and social production needs. Based on the development of the Belt and Road Initiative, it has led to higher standards and requirements for the international logistics industry in all walks of life. The knowledge system lacks systematic construction. Students often memorize by rote and forgetting rates are too high, which is not conducive to the cognitive structure network. The construction of the network and the cultivation of logical thinking system [4]. The logistics industry continues to integrate and develop, national policies are constantly updated, and the content of “static” courses cannot keep up with the times. As a result, students do not pay much attention to the cutting-edge development of the logistics industry, are not sensitive to the updating of national standards and policies, and the shaping of personal values is not enough, so that students face the truth after going to work. Difficulties and conflicts at work cannot be effectively resolved.
3 Innovation Strategy In order to solve the pain points of the above courses, as presented in Fig. 1, this research takes school-enterprise cooperative education as the overall idea, deepens the concept of the integration of production and education, and innovatively puts forward the teaching strategy of “three-dimensional integration” as the main body. Two-dimensional: the theory system strengthens the correlation between concepts; Two-dimensional: cognitive steps are defined by modular implementation; Three-dimensional: to project system docking enterprise practice. Three-dimensional mutual promotion, explore the effective “three-dimensional integration” teaching strategy path [5].
X. Li et al.
Implement
Modular docking Respectively action Strengthen responsibility
Project-based implementation Summarize practical operation Perceive experience
Targets
Abutment
Conceptual recognition Integrate knowledge Build confidence
Implementation contents
Cognition
1.Extract the outline 2.Improve the model 3.Strengthen the theory 4.Learn the new by restudying the old 5.Deep learning
Completion routes
902
Personal cognition and team discussion
1.Modular decomposition 2.Step scheduling 3.Mutual assistance 4.Resource sharing 5.Detection and feedback
1.Modular decomposition 2.Step scheduling 3.Mutual assistance 4.Resource sharing 5.Detection and feedback
Online-to-offline and multiple supports
Division of project and teamwork
Fig. 1. “Three-dimensional integration” teaching strategy
Summarize from bottom to top
Depth/Height
Non-cognitive goal coordination
Promotion
Pioneering Seriousness
Classification Consolidation
Sociality
Restudy old Connect Modeling materials
Breadth/Global
Gradually refine from top to bottom Fig. 2. "Conceptual knowledge” construction
3.1 Conceptual Understanding Theoretical basis is the “support point” of professional courses, and also the basis of later modular and project-based learning. It needs to be compiled into the existing concept network of students, so as to promote the formation of long-term memory. A top-down and bottom-up approach to conceptual architecture, Firstly, based on the existing learning content, the concept connection is strengthened from top to bottom to promote the construction of mental models. Then, the industry frontier, new policies and new regulations are integrated from bottom to reach a certain depth and width. Enable students to raise their theoretical knowledge to a certain height and comprehensiveness.
Reform and Innovation of International Logistics
903
By strengthening, summarizing and improving the basic concepts, construct the collaboration of social, serious and pioneering non-identification goals, and form the conceptual understanding with the characteristics of renewal and iteration. Conceptual knowledge construction is shown in Fig. 2. 3.2 Modular Recombination In order to undertake the “conceptual cognition” content, docking “project implementation” plan. The content of the textbook is integrated and reorganized into different modules, and each module is divided into several contents, each of which has different meanings and training objectives, so as to organize project-oriented teaching by modules. According to different tasks, different teaching methods are adopted according to the logical sequence suitable for the project, the work of all parties is broken down, and the process is coherent and integrated. Integrating learning Pass and other information teaching tools, forming a modular restructuring framework [6]. The modular recombination framework is shown in Fig. 3. Modular recombination
Textbook chapter 01
Overview
02
System
03
Basic knowledge
04
Goods storage
05
Sea transport
06
Air transport
07
Land transport
08 Container transport 09
Chapters 2, 3 and 10
Module 2 International logistics trade terms
Chapters 2, 5, 8 and 9
Module 3 International water Transport
Master the operation mode of international logistics
Chapters 2, 6 and 11
Module 4 International Air Transport
Including: international logistics of all kinds of transportation, cost calculation, customs operation process
Module 5 International land transport
Merge common section Such as: international logistics system
Chapters 2, 7, 8 and 11
Module 6 customs declaration and inspection
Logistics port
10
Freight insurance
11
Freight forwarder
12
Supply chain
Understand the past and present life of international logistics
Chapter 1
Module 1 Understanding international logistics
Chapters 9, 10 and12
Chapters 5, 6, 7 and12
Module 7 Future development of international logistics
Summarize the future development of international logistics
Fig. 3. "Modular recombination” frame diagram
3.3 Project-based Implementation According to the content system of this course, the learning and growth process is planned, the comprehensive learning plan is designed and formed, and the learning path map is determined, shown in Fig. 4. Teachers ensure the implementation and promotion of project-based implementation chart through task supervision, step scheduling, process help, resource link and inspection feedback [7].
904
X. Li et al.
School teachers Enterprise mentor
Students
Digital teaching tools
Situational settings
Publish Project
Guiding observation
Supervision and guidance
Feedback and summary
Consolidate and migration
Analysis and discussion
Project selection
Group division
Collaboration completion
Communication and sharing
Learning and evaluation
Native land emotion
Professional feelings
Harmony and mutual assistance
Responsibility and ingenuity
Self-confidence and improvement
Sticking to innovation
Thematic discussion area
Background information
Autonomous learning
Reference material
Live/video
Self and mutual assessment
Fig. 4. The modular recombination framework
Assign the class to each group with a random sub-project to ensure the coverage of each group’s project topics. The responsibility of each topic selection should be assigned to each student, and the team leader should guide each group to assign tasks. Members in the group should grasp the overall direction and content of the project. During the completion of the project, the group members made independent learning records, and the group students completed each project through teamwork. The grouping, communication and evaluation process was supplemented by the learning platform, and then shared and displayed. The groups directly evaluated and scored each other [8].
4 Teaching Process Organization and Design In view of the above “pain points”, the teacher team takes school-enterprise collaborative education as the overall idea, deepening the concept of integration of production and education, and innovatively puts forward the teaching strategy of “three-dimensional integration” as the main body to solve the key problems in the course. Starting from teaching objectives, content, methods, activities, evaluation and other aspects, the organizational process and specific design are as follows: 4.1 Facing the Industry, Docking Enterprises, Reconstructing the Course Content System At present, the practical teaching of international logistics major is single and not closely connected with enterprises. Therefore, the formulation and design of the practical teaching curriculum system should be combined with enterprises, considering the actual needs of enterprises and relying on the power of enterprises, and the course content should be jointly designed and developed with enterprises for deep integration. 4.1.1 Emphasis on Application and Enterprise Integration According to the characteristics of international logistics curriculum, the curriculum system is modular and then project-based, and the curriculum ideological and political cases, international logistics frontier, industrial development trend and other contents are expanded into the integration. Curriculum modularization is not the simple recombination of traditional teaching courses, but the traditional knowledge input oriented
Reform and Innovation of International Logistics
905
teaching to knowledge, ability, quality output oriented teaching process, emphasizing the application of curriculum and enterprise integration [9]. This results in the strategy shown in Fig. 5. Textbook chapter 01
Overview
02
System
03
Basic knowledge
04
Goods storage
05
Sea transport
06
Air transport
07
Land transport
08
Container transport
Module 1 Understanding international logistics Module 2 International logistics trade terms Module 3 International water Transport Module 4 International Air Transport
Freight insurance
11
Freight forwarder
12
Supply chain
2.1 2.2 2.3 2.4 3.1 3.2 3.3 4.1 4.2 4.3
Module 5 International land transport
09 Customs declaration 10
Project-based
Modularization
Module 6 customs declaration and inspection Module 7 Future development of international logistics
5.1 5.2 5.3 5.4 6.1 6.2 6.3 6.4
Group E 1 trade term Group F 3 trade terms Group C 4 trade terms Group D 3 trade terms Import process by sea Marine export process Calculation of freight Air import process Air export process Freight calculation Operation process Document service Container transport Cost calculation Customs and tariff Port policy Compensation scheme Check procedure
Expansion and integration Ideological and political case Feelings of nation Social responsibility Bear hardships and hard work Firm and indomitable Quality awareness Sense of responsibility professionalism Professional ethics ...
Fig. 5. Modular curriculum system, project system, expansion and integration
4.1.2 The Ideological and Political Cases of the Course Are Organically Integrated with the Teaching Content to Enrich the Course Connotation Starting from the theoretical knowledge, value concept and spiritual pursuit of ideological and political education, such as dynamic attention to international logistics standards and industrial prospects, to promote students to build cultural confidence in the logistics industry and sense of responsibility, Based on the development trend of intelligent logistics, the paper analyzes the driving force of technological innovation on logistics industry and the significance of promoting industrial upgrading. 4.1.3 Industry Frontier, Policy Update into the Teaching Content, Expand the Course Extension Under the impact of the global pandemic, the international logistics industry is already facing challenges such as business transformation and service improvement. The whole international logistics industry will present both “prosperous” and “difficult” situation. Industry standards have been constantly upgraded, and laws and regulations have been
906
X. Li et al.
updated many times. Students are instructed to apply the mini program of “China Logistics Online Platform” and “Global Logistics Inquiry” to improve the efficiency of students in finding the latest policies, new laws and regulations and new guidelines in project tasks. 4.2 Three Steps to Promote, Three Steps to Transfer, the Implementation of Project-Based Teaching To make the students become the “protagonist” in the class, first of all, let the students look at the major and study from a different perspective, let the students clearly know what to learn and how to use it? Driven by the project, with the international logistics operation process as the core and team cooperation as the necessary form, students are guided to in-depth practice and form the project results in the course. Integrate the course achievements with the enterprise needs to achieve the “conversion agent” of learning style. 4.2.1 Preparation: Set up the Platform, Multiple Tutoring Combined with the requirements of knowledge ability and quality of the project task, the project learning platform is designed and constructed through learning platform (see Fig. 6). Typical cases, video resources, industry standards, virtual simulation training projects and other frameworks are provided to students. Teachers provide technical guidance and psychological guidance, so that students can successfully complete the project tasks and expand the knowledge points combined with the project tasks. 1.Communication skills 2.survey visit 3.Organization and coordination 4.project management 1.Mental management 2.Motivational video 3.Communication and mutual assistance
Skill coaching
Mental coaching Chaoxing learning APP
Knowledge coaching
Typical cases Video resources Industry standards Virtual simulation training project
1.Typical cases 2.Enterprise's instances 3.Literature review 4.curriculum materials 5.Training projects
Fig. 6. Multi-tutoring mechanism and platform construction
4.2.2 Implementation: Three Level Project, Step by Step Taking Module 3: International logistics air transport as an example, docking items 3.1–3.2: Ocean import process, export process and liner billing [10].
Reform and Innovation of International Logistics
907
Level 3 projects: In combination with online video learning, case analysis and discussion, multiple tutoring mechanisms and learning platform, the core content of shipping import and export process is completed, the liner billing method is calculated through group discussion and data collection, and finally the teacher feedback. Level 2 Project: Each group of students completed the core content of the project design. Connect with the work flow of Marine import and export, division of labor, clear tasks, clear goals, strong operability, and share and display the progress of different modules after completion. Main steps are listed: 1) Preliminary research stage: Through the investigation and analysis of the current situation of relevant enterprises (mainly school-enterprise cooperation units), the initial identification of logistics culture and professional value is formed. Through the background research of typical projects, students can feel the connection between theory and practice. From the research, understand the definition of sea transportation, its role, the formula of liner billing, factors to be considered, the causes of risks and preventive measures. 2) Medium-term promotion stage: By collecting and sorting out the preliminary research data, the team conducted internal discussion, communication and planning, and each group analyzed the current situation, market trend and future development mode in combination with the shipping industry, and promoted the writing of the project book in combination with shipping enterprises. 3) Project acceptance stage: Completion of the project, together with the results and expected value. Level I project: Based on the team, as can be seen is Fig. 7, project report and achievement presentation were carried out around each group project. The specific process included: “Project report - teacher’s questions - group mutual evaluation - project points”. To enable students to form a holistic understanding of the international logistics industry, clear interdisciplinary integration of scientific literacy, innovative thinking and team consciousness.
Project Presentation
Teacher's Questioning
Peer-assessment
Project scores
Fig. 7. The team project presentation
4.2.3 Promotion: Self-construction, Deepening Cognition Based on constructivism theory, it emphasizes student-centered, enabling students to actively explore and discover knowledge and actively construct the knowledge system
908
X. Li et al.
they have learned. Promote in-depth teaching and expand the concept. For students, their time of contact with international logistics is still short, and their abilities are relatively poor. They often concentrate on learning a certain knowledge point, but lack the overall concept, forming a bad situation of “learning while forgetting”. Deep learning is an important way to combat this adverse phenomenon. When explaining concepts, teachers should not only focus on the textbook itself, but also help students form a macro perspective, clarify cognition through the comparison of concepts before and after, construct knowledge network, and cultivate students’ core qualities such as logical reasoning. 4.3 Process Motivation, Result Evaluation, to Ensure the Learning Effect With the flexible application of tools such as Learning Pass, the trajectories of students’ learning activities are recorded completely. Based on the response of students to teaching activities, such as the feedback rate of watching course videos, the average response time of learning, and the quality of replying topic discussion, students with weak learning drive are screened and given emphasis and encouragement [11]. Stimulating students’ learning motivation has always been an important function of colleges and universities and teachers [12–14]. Only in this way can students truly become masters of learning and excellent talents [15]. The course evaluation system is shown in Table 1. The project task consists of teacher evaluation, group evaluation and self-evaluation. The evaluation includes whether the learning task has been completed as expected, whether the learning effect is good, the level of learning ability, the contribution of oneself in the group task, and the improvement suggestions for the next study, etc. Table 1. The course evaluation system Evaluation process
Evaluation contents
Proportion (%)
Usual performance (60%)
Course video
10
Final exam (40%)
Homework or quizzes
20
Topic discussion and classroom interaction
10
Project tasks
Third-level project
10
Second-level project
15
First-level project
15
Project acceptance report
20
Reform and Innovation of International Logistics
909
5 Teaching Effect of Integration of Production and Education The average score of students has been improved year by year, and students’ satisfaction with the course is more than 93%. Students have won many excellent achievements in student competitions at all levels. Through the integration of industry and education and the docking of enterprises, many students are employed in the world’s top 500 enterprises. The international logistics quality of students is also praised by employers. In the organic integration of ideological and political cases with teaching content, many students take the initiative to participate in volunteer and social practice activities such as going to the countryside and poverty alleviation, and the number of participant increases year by year. Specific materials are shown in the Fig. 8.
Fig. 8. Teaching effect at the student level
6 Conclusion According to the thought of integration of production and teaching, a complete teaching content system related to professional courses has been constructed. I have compiled the supporting teaching syllabus, teaching cases, experimental course operation process, written research reports, teaching reform plans, and published relevant teaching reform papers, which have been promoted and applied in related majors of other colleges and universities. It has been awarded the national first-class undergraduate course of virtual simulation experiment, which is shared and applied by universities all over the country. The course
910
X. Li et al.
teaching model, teaching reform plan and teaching cases have been popularized and applied in other professional courses. At present, the teaching results of the “Design of Container iron-water Combined Transport Port Operation Plan” developed by this course based on the teaching reform of the integration of production and education of this course won the second prize of school-level teaching results. Acknowledgment. This research was funded by School-level Teaching Reform Project (2022Z03) of Wuhan Technology and Business University.
References 1. Hughes, L.J., Amy, N., Mitchell, M.L.: Utilising the invitational theory provides a framework for understanding assessors’ experiences of failure to fail. Nurse Educ. Pract. 55(5), 103135 (2021) 2. Mu, M.X.: Research on application-oriented undergraduate course construction under the background of industry-education integration- take the performance management course as an example. Front. Educ. Res. 12(1), 53–57 (2022) 3. Tunviruzzaman, R., Tahera, T., Zannat, T.: Economic & geopolitical opportunities and challenges for Bangladesh: one belt-one road (OBOR). Int. J. Res. Bus. Soc. Sci. 10, 2147–4478 (2021) 4. Bogojevi, S., Zou, M.: Making infrastructure “visible” in environmental law: the belt and road initiative and climate change friction. Trans. Natl. Environ. Law 10(1), 35–56 (2021) 5. Gratchev, I., Jeng, D.S.: Introducing a project-based assignment in a traditionally taught engineering course. Eur. J. Eng. Educ. 43(5), 788–799 (2018) 6. Jian, Q.: Effects of digital flipped classroom teaching method integrated cooperative learning model on learning motivation and outcome. Electron. Libr. 37(5), 842–859 (2019) 7. Babincakova, M., Bernard, P.: Online experimentation during COVID-19 secondary school closures: teaching methods and student perceptions. J. Chem. Educ. 97(9), 3295–3300 (2020) 8. Liao, D.: Enhancing project-based learning to the teaching of control engineering education. Boletin Tecnico/Tech. Bull. 55(13), 124–128 (2017) 9. Brdulak, H., Brdulak, A.: Challenges and threats faced in 2020 by international logistics companies operating on the polish market. Sustainability 13(1), 359 (2021) 10. Magnaye, R.B., Chaudhry, S.S., Sauser, B.J., et al.: Bridging the gap between practice and undergraduate teaching of operations management: the case of public liberal arts colleges. Int. J. Oper. Quant. Manage. 26(1), 1–16 (2020) 11. Verbi, G., Keerthisinghe, C., Chapman, A.C.: A project-based cooperative approach to teaching sustainable energy systems. IEEE Trans. Educ. 60(3), 221–228 (2017) 12. Fatima, S., Abdullah, S.: Improving teaching methodology in system analysis and design using problem based learning for ABET. Int. J. Mod. Educ. Comput. Sci. (IJECS) 5(7), 60–68 (2013) 13. Kaviyarasi, R., Balasubramanian, T.: Exploring the high potential factors that affects students’ academic performance. Int. J. Educ. Manage. Eng. (IJEME) 8(6), 15–23 (2018) 14. Al-Hagery, M.A. Alzaid, M.A. Alharbi, T.S., Alhanaya, M.A.: Data mining methods for detecting the most significant factors affecting students’ performance. Int. J. Inf. Technol. Comput. Sci. (IJITCS), 5, 1–13 (2020) 15. Zhiqin, L., Jianguo, F., Fang, W., Xin, D.: Study on higher education service quality based on student perception. Int. J. Educ. Manage. Eng. (IJEME), 2(4), 22–27 (2012)
An Analysis of Talent Training in Women’s Colleges Based on the Characteristics of Contemporary Female College Students Ting Zhao1 , Zaitao Wang2(B) , Xiujuan Wang3 , and Chuntao Bai4 1 Library, Shandong Women’s University, Jinan 250300, China 2 School of Business Administration, Shandong Women’s University, Jinan 250300, China
[email protected]
3 Fundamental Science Section in Department of Basic Courses, Chinese People’s Armed
Police Force Logistics College, Tianjin 300309, China 4 Party and Government Office, Tianjin Beichen Science and Technology Park Management
Co., Ltd, Tianjin 300405, China
Abstract. How to cultivate female college students and what kind of talents they will become is a question worth discussing and must be answered by women’s colleges. By looking up the relevant literature at home and abroad, this paper clarifies the characteristics of women’s colleges and universities, and points out their mission, value and development direction. In order to have a systematic understanding of female characteristics in colleges and universities, this paper briefly introduces the structural equation model method, and points out that their growth environment, values and consumption outlook are different from each other in the past, which is conducive to teaching women’s colleges in accordance with their aptitude and completing the talent training goal better. This paper puts forward an integrated curriculum system that integrates the training specifications of knowledge, ability and quality, and takes the spiral ability training as the main line. Women’s colleges and universities emphasize that, according to women’s physiological and psychological characteristics, they should cultivate women’s compound applied talents suitable for social needs, which will play an important role in leading and promoting the deconstruction of gender bias, eliminating gender inequality, building advanced gender culture, promoting women’s development and gender equality, and promoting social civilization and progress. Keywords: Women’s colleges · Contemporary female college students · SEM · Personnel training
1 Introduction The cause of women in the world has made great achievements, but still faces many challenges. More than half of the world’s 800 million poor people are women; when wars and epidemics strike, women often bear the brunt. The main reason is that the historical stereotypes of gender inequality have not been completely eliminated, the social © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 911–920, 2023. https://doi.org/10.1007/978-3-031-36115-9_81
912
T. Zhao et al.
environment for women’s development needs to be optimized, and women’s participation in cultural education, economic participation, decision-making management, social security, legal protection and other aspects needs to be further improved [1]. According to < Outline for the Development of Chinese Women (2011–2020) >, which is showed as Fig. 1. It is not difficult to see from the table that from 2010 to 2020, the proportion of women in postgraduate education has continued to increase, and by 2020, its proportion has exceeded half. In the report of the 20th National Congress of the Communist Party of China in 2022, it was clearly pointed out that we should adhere to the basic national policy of gender equality and protect the legitimate rights and interests of women and children. Women’s colleges and universities are educational institutions that provide higher education for women and are an important part of higher education in China. Building a strong country in higher education is an important task for women’s colleges and universities. Training high-quality female talents with innovative ability should be an important task for women’s colleges and universities.
Fig. 1. Number and proportion of female graduate students in China from 2010 to 2020
2 Research on Characteristics of Women’s Colleges and Universities The characteristics of running a university is a systematic structure composed of multiple elements, and so is the characteristics of running a women’s college. In this system structure composed of multiple elements, each element is interdependent and interacts with each other, which jointly affects the formation and development of school running characteristics. The following is a literature review of each element of school running characteristics.
An Analysis of Talent Training in Women’s Colleges
913
2.1 Research on the Goal and Idea of Running a School Kim et al. pointed out that female characteristics should be highlighted as the main body of school running, gender theory should be studied and gender awareness should be spread, which will further highlight its professionalism [2]. Dasgupta et al. believes that teaching should be carried out according to the needs of women as talents and the cognitive laws, and specialties and courses should be set up and taught according to women’s characteristics [3]. The “female advantages” of specialties and courses will be paid more attention. He puts forward that we should surpass the tradition in school running goals, education models and school functions, and build a targeted environment for cultivating women’s culture, so that there will be room for upward development and sustainable vitality. 2.2 Research on the Characteristics of the Main Body of Running a School The research on the “characteristics of the main body of running a school” of women’s colleges and universities is carried out from the two main bodies of teachers and students. (1) To train teachers in women’s colleges and universities on gender awareness, so as to improve their consciousness and sensitivity of gender awareness; (2) This paper studies the scientific literacy, interpersonal communication and psychological anxiety of female college students. Sugianto pointed out that colleges and universities should take effective measures to strengthen the cultivation of college students’ scientific literacy [4]. Mosleh et al. suggested that schools should start from reality and carry out mental health education to alleviate their psychological anxiety and help them improve their mental health [5]. Mozgalova et al. proposed improvement approaches from both the students themselves and the school [6]. 2.3 Research on the Characteristics of Talent Training The research focuses on the goal, mode and significance of female talent training. The training objectives formulated by women’s colleges and universities mainly include “taking social and market needs as the basis, giving full play to women’s advantages”, “modern women who are excellent in quality, professional, physical, artistic, and aesthetic”, and “opening up new employment channels for young women”. Liu Jia and Yang Liu started from explaining the connotation and extension of the talent training model, analyzed the characteristics of the talent training model and the factors affecting the talent training model, it also puts forward some ideas on the talent cultivation of women’s universities [7]. Tang Jianxiong discussed the characteristics and construction of the talent training model of modern women’s colleges and universities, and pointed out that if women’s colleges and universities want to really go hand in hand with other colleges and universities, they must build a talent training model that meets the needs of social development and has a distinctive female education color [8]. Choi et al. pointed out the defects of the traditional talent training model and the bottleneck in the reform of the talent model in women’s colleges and universities, and discussed the employment oriented training model of applied talents for women [9]. Cai pointed out the significance of the “four self” talent training path of the school [10].
914
T. Zhao et al.
2.4 Research on the Characteristics of Disciplines and Specialties There are two articles about “discipline and specialty characteristics”. Warren et al. emphasized that the scientific setting of majors is an important way for universities to highlight the characteristics of running schools [11]. Women’s universities can only ensure the training of outstanding women talents with characteristics by scientifically standardizing, fully demonstrating and building a scientific and characteristic professional system on the premise of in-depth discussion of teaching rules and the growth laws of women talents. Kyrpychenko et al. pointed out that the communicative competence development in teaching professional discourse is of special significance for women in colleges and universities [12]. 2.5 Research on the Characteristics of Curriculum Construction The research on the curriculum construction of women’s colleges and universities focuses on the following two aspects: (1) The characteristic curriculum construction of women’s colleges and universities. Sharm et al. aimed to apply exploratory applied research design, evaluate challenges through a unified cloud-based e-learning framework, and use hybrid research methods to propose a new cloud-based solution framework for women to be challenged by accidental injuries [13]. Li analyzed the problems of the development of the course of women’s studies in China, the methods and objectives of the course construction of women’s studies, and the ways of the course construction of women’s studies to ensure the healthy development of the course construction of women’s studies [14]. In addition, ADIGUN proposed that the integrated teaching of artistic dance and aerobics has the function of cultivating correct health outlook, improving artistic accomplishment, and meeting physiological and psychological needs of contemporary female college students [15]. (2) Cultural curriculum in women’s colleges and universities. Koseoglu et al. based on the basic characteristics of the feminist curriculum, paid special attention to the gender inequality in the curriculum, advocated the construction of a curriculum serving the teaching purpose of women’s colleges and opposed the gender bias in the curriculum [16]. Through the analysis of the above research, it is not difficult to find that the research mainly focuses on narrow aspects, with more repetitive research, less innovative research and less objective research attitude. There are few research results on the characteristics of foreign women’s schools, which are not deep and thorough enough. The total number of research results on the characteristics of foreign women’s colleges in the existing literature is relatively small, but only from two aspects. On the one hand, the characteristics of famous women’s universities in the United States and South Korea are analyzed, and on the other hand, a comparative study of Chinese and foreign women’s universities is conducted. There are few empirical research results, and most of the articles are the author’s experience summary or theoretical overview. More empirical research is needed to more convincingly reflect the current situation of the running characteristics of women’s colleges and universities in China, and accordingly put forward corresponding countermeasures and suggestions, in order to make our women’s colleges and universities more robust, healthy development and growth.
An Analysis of Talent Training in Women’s Colleges
915
3 Research Method Based on the structural equation model (SEM), this paper will carry out the corresponding empirical research, build the structural equation model of college students’ personality and behavior characteristics and the factors affecting the cultivation of innovative talents, and analyze and explore the main factors that affect the cultivation of innovative talents in economics and management. Let ξ be a vector of independent variables and η be a vector of dependent variables, Λ is a matrix of factor loadings, ε a vector of residuals often known as unique variates. B, L and M arecoefficient matrices, I is the identity matrix of an appropriate order, Let us denote B = L0 00 , = MI , and v = ηζ , the following equation can be obtained: η = Lη + M ξ
(1)
v = Bv + ξ
(2)
To make sure the inverse of I–B exists, we assume the I–B is non-singular, then v = (I − B)−1 ξ
(3)
The components of the matrix G are either 1 or 0 which connects v to the observed variables x such that x = Gv. Let μ = E(x), μξ = E(ξ), = Cov(x), and = Cov(ξ). Various covariances φij are shown as two-way arrows in path diagrams. The full mean and covariance structure analysis model is given as: Mean structure: μ = G(I − B)−1 μξ
(4)
Covariance structure:
= G(I − B)−1 (I − B)−1 G
(5)
The measured variables x may be generated by the ξ x = μ + ξ + ε
(6)
ξ = Bξ + ζ
(7)
ξ = (I − B)−1 ζ
(8)
x = μ + (I − B)−1 ζ + ε
(9)
If the means are unstructured, μ = E(x). With a structure, we take μ = 0 in (9) and let μζ = E(ζ), and with the covariance matrix of the ζ and the ε given as ζ and , respectively, the mean and covariance structure of the model follow as: μ = (I − B)−1 μζ
(10)
916
T. Zhao et al.
= (I − B)−1 ζ {(I − B)−1 } +
(11)
This representation shows the confirmatory factor analysis model = ζ +
(12)
SEM is a multivariate statistical method, which combines factor analysis and path analysis, including leading variables, potential variables, interference or error variables, indirect effects or total effects for verification. Structural equation model can be divided into measurement model and structural model. It can not only study the structural relationship within variables, but also study the relationship between variables. Therefore, it is widely used in various studies by domestic and foreign scholars. In order to pursue preciseness, even if the established model can be fitted, the model must be modified according to the evaluation structure to obtain the most scientific, reasonable, most suitable and best explanation of the final model.
4 Analysis of Characteristics of Contemporary Female College Students 4.1 An Analysis of the Psychological Characteristics of Contemporary Female College Students Influenced by self-gender role and external social and cultural atmosphere, the psychology of contemporary female college students mainly has the following characteristics. (1) Strong self-awareness, incomplete self-awareness, relatively lack of team awareness, and urgent need to improve interpersonal skills. They tend to pursue and advocate individuals and personalities. In the survey, many female college students pay more attention to themselves and less attention to the feelings of others. They lack clear recognition of their roles and comprehensive understanding of their advantages and disadvantages. Many people think they lack interpersonal skills and need to improve their sense of teamwork and interpersonal skills. (2) Confident, optimistic and frank, but sometimes lazy, lacking time concept. On almost all occasions, as long as they have the opportunity, they will publicize their personality. They dare to express their views without scruple. They are keen on various campus cultural activities and competitions and actively participate in various elections; but many students think they are lazy and have a weak sense of time. (3) Strong learning ability, extensive hobbies, strong ability to accept new things, but affected by the physiological characteristics of late youth, emotional stability is not good, and lack of some self-control. Their average intelligence is higher than their previous peers, and they have strong learning ability. 4.2 An Analysis of the Characteristics of Contemporary Female College Students Compared with the past, the family environment in which contemporary female college students grow up is relatively superior, and they have no worries about food and clothing since childhood, which clearly reflects the following characteristics.
An Analysis of Talent Training in Women’s Colleges
917
(1) The vision is broad and the knowledge is rich, but sometimes the heart is empty and the goal is vague. Contemporary female college students have grown up with the Internet. Their familiarity and affinity with new media far exceed those of previous generations. They have a broader vision, more diversified channels to receive information, advanced mental development, and far more knowledge and precocity than previous generations. It also directly affects their lifestyle and values, which easily makes them feel empty. Moreover, the more familiar with such new media, the contrast between virtual and real will be felt. Some of them virtual communication ability has far exceeded their real communication ability. This virtual personality has a great influence on their socialization. (2) Diversified values and more realistic values. The value orientation of contemporary female college students is more diversified. Some social realities have made them understand that they should focus on specific things rather than beliefs. Their aesthetic taste, lifestyle and moral boundaries have also gradually changed. The inner world has become “old” since childhood, and has a better understanding of the rules of the adult world. (3) The consumption concept is ahead of time, and the consumption level is incompatible with the family economy. According to statistics, more than 99% of the nearly 15, 000 students in our college have bought mobile phones, and there are many highend mobile phones; the average monthly living expenses given by parents to students reached 600–1, 500 yuan. And many students obviously showed that they could not bear hardships, and showed varying degrees of inadaptability in the face of setbacks. For example, in the centralized internship carried out by the school, some girls rent houses and even need help from others. Girls who practice in refrigerators of JD.com and other units generally say they are tired, and some even give up the internship in processing plants.
5 An Analysis of Talent Training in Women’s Colleges Based on the Characteristics of Contemporary Female College Students In combination with the teaching practice of women’s universities and the campus culture of “advocating virtue, love, erudition, and beauty”, women’s colleges and universities have two main goals for talent cultivation: first, enrich professional education through liberal education, improve the quality of talent training, and establish a talent training model of “thick foundation, wide caliber”, so as to improve the core competitiveness of women’s colleges and universities; second, improve the liberal education curriculum system and model, and gradually form and reflect the unique and distinctive characteristics of women’s colleges. 5.1 Training Specifications Integrating Knowledge, Ability and Quality Training specification is the specific refinement and decomposition of training objectives. According to the talent training concept of “all-round development”, the training objectives are refined based on knowledge, focused on ability and expanded by quality. According to the orientation of talent training objectives, determine the knowledge
918
T. Zhao et al.
structure, ability structure and quality structure with horizontal correlation in spatial dimension. The same structure requires an integrated training specification with vertical correlation in time dimension: (1) Knowledge structure requirements: a. master the knowledge of humanities and social sciences; b. master the basic knowledge of the discipline; c. master basic professional knowledge and theory; d. master professional technical knowledge. (2) Capability structure requirements: a. have the ability to understand the theoretical frontier, development trends and independently acquire knowledge in the professional field; b. have the ability to use engineering technology for system design; c. strong ability in scientific research, engineering practice and solving practical problems with the knowledge learned; d. have the ability to learn independently, innovate, work cooperatively and organize. (3) Quality structure requirements: a. have good ideological and moral cultivation, professional quality, physical and mental quality; b. have the spirit of dedication, interpersonal awareness and solidarity; c. have a certain literary and artistic accomplishment, scientific engineering practice methods d. have a certain international vision, realistic and innovative consciousness. 5.2 Integrated Curriculum System with Spiral Ability Training as the Main Line According to the talent training concept of “demand guidance, project driven, strengthening practice, teaching and research unity, highlighting innovation, and comprehensive development”, according to the talent training specifications and characteristics, as well as the consistency, completeness, integration, progressiveness, and practicality of the curriculum system, and in combination with our own advantages and characteristics, we will build a talent training oriented, theory and practice oriented Quality education is integrated into the training of professional knowledge and ability, and the knowledge, ability and quality are coordinated and unified. The curriculum system is oriented by ability cultivation, supported by centralized practice links, and based on core courses. A group of centralized practice links and core courses cultivate an application or design ability and related knowledge. According to the talent cultivation concept of “demand guidance”, the school and the enterprise jointly design the curriculum system, and the enterprise sends engineers to set up relevant engineering design courses in the school. The curriculum system has set up a series of centralized practice links, independent class experiments and in class experiments to implement the talent cultivation concept of “strengthening practice”.
6 Conclusion For thousands of years, women have been positioned as the role of procreation, housekeeping, helping husbands and taking care of children. They have been excluded from receiving education for a long time. “women without talent are virtuous” is a true portrayal of this phenomenon. It is only in recent decades that a large number of Chinese women have entered schools to receive formal education. It is precisely because the objects of education for thousands of years are all men (or basically men), so the current
An Analysis of Talent Training in Women’s Colleges
919
educational philosophy, teaching methods, curriculum structure and examination evaluation model are designed with men as the objects of education. The current education system mainly takes into account the psychological characteristics and cognitive laws of men, and rarely takes into account the characteristics of women. This paper makes a deep analysis of the characteristics of contemporary female college students, and on this basis, puts forward some suggestions on talent training in women’s colleges. What kind of education should be given to women, such as “providing women with educational forms suitable for women’s characteristics” to “meet the needs of women’s growth and talent”; “cultivate women with the spirit of ‘four self’”; employment oriented; based on the development of female students, compensation education should be carried out for female college students. In terms of specialty setting and course offering, we should think about how to teach courses with women’s characteristics and cultivate talents with women’s characteristics. In view of the problems encountered in the training of application-oriented talents, we should reform the training mode of application-oriented talents, and study and practice effective ways to improve the training quality of application-oriented innovative talents. Acknowledgment. This research has been supported by the 2017 school level teaching reform research project of Shandong women’s University: discussion and practice of innovative personnel training mode of economic management in women’s University, and the Shandong Social Science Planning Project (Grant No. 18CGLJ48, 18CSJJ32).
References 1. Seymour, S.: In: Mukhopadhyay, C.C., Seymour, S. (eds.) Women, Education, and Family Structure in India, pp. 213–233. Routledge, New York (2021) 2. Kim, D., Yoon, M., Jo, I.H., et al.: Learning analytics to support self-regulated learning in asynchronous online courses: A case study at a women’s university in South Korea. Comput. Educ. 127, 233–251 (2018) 3. Dasgupta, U., Mani, S., Sharma, S., et al.: Effects of peers and rank on cognition, preferences, and personality. Rev. Econ. Stat. 104(3), 587–601 (2022) 4. Sugianto, E.S.: The role of collaborative learning and project based learning to increase students’ cognitive levels in science literacy. In: International Conference on Madrasah Reform 2021 (ICMR 2021), pp. 67–72, Atlantis Press (2022) 5. Mosleh, S.M., Shudifat, R.M., Dalky, H.F., et al.: Mental health, learning behaviour and perceived fatigue among university students during the COVID-19 outbreak: a cross-sectional multicentric study in the UAE. BMC psychol. 10(1), 47 (2022) 6. Mozgalova, N.G., Baranovska, I.G., Hlazunova, I.K., et al.: Methodological foundations of soft skills of musical art teachers in pedagogical institutions of higher education. Linguist. Cult. Rev. 5(S2), 317–327 (2021) 7. Jia, L., Liu, Y.: Research on guiding education of college students’ interpersonal relations. Writer’s World 22, 110–111 (2021). (in Chinese) 8. Jianxiong, T.: Research and practice on the reform of new business talents training mode in higher vocational education -- Taking the International Economy and Trade Major of Guangdong Women’s Vocational and Technical College as an example. Univ. Educ., 208– 210+220 (2022) (in Chinese)
920
T. Zhao et al.
9. Choi, S.J., Jeong, J.C., Kim, S.N.: Impact of vocational education and training on adult skills and employment: an applied multilevel analysis. Int. J. Educ. Dev. 66, 129–138 (2019) 10. Cai, X.: Research on the training mode of adaptive talents in higher vocational education under the background of educational informatization 2.0. In: 2022 6th International Seminar on Education, Management and Social Sciences (ISEMSS 2022), pp. 3542–3548, Atlantis Press (2022) 11. Warren, K., Mitten, D., D’amore, C., et al.: The gendered hidden curriculum of adventure education. J. Experiential Educ. 42(2), 140–154 (2019) 12. Kyrpychenko, O., Pushchyna, I., Kichuk, Y., et al.: Communicative competence development in teaching professional discourse in educational establishment. Int. J. Educ. Manage. Eng. 4, 16–27 (2021) 13. Sharm, D.P., Deng, D.K., Tigistu, G., et al.: A cloud based learning framework for eradicating the learning challenges of Ethiopian working professionals, disables and women. Int. J. Mod. Educ. Comput. Sci. 14(2), 41–54 (2022) 14. Li, L.: Education supply chain in the era of Industry 4.0. Syst. Res. Behav. Sci. 37(4), 579–592 (2020) 15. Adigun, J.O., Irunokhai, E.A., Onihunwa, J.O., et al.: Development and evaluation of a web based system for students’ appraisal on teaching performance of lecturers. Int. J. Inf. Eng. Electron. Bus. 14(1), 25–36 (2022) 16. Koseoglu, S., Ozturk, T., Ucar, H., Karahan, E., Bozkurt, A.: 30 years of gender inequality and implications on curriculum design in open and distance learning. J. Interact. Media Educ. 5(1), 1–11 (2020)
Solving Logistical Problems by Economics Students as an Important Component of the Educational Process Nataliya Mutovkina(B) Tver State Technical University, Tver 170012, Russia [email protected]
Abstract. The article examines teaching students to solve logistical problems and also focuses on the importance of such skills in the professional activities of economists. Methodological teaching methods are considered in preparing students toward “Economics” at Tver State Technical University. Logistics asks in the educational process are presented to students as one type of optimization tasks and are solved as transport tasks, assignment tasks, backpacks, traveling sales agent. The author of the paper defines the role and practical significance of logistics tasks in the professional activity of economists. In training, students formalize logistics tasks, choose the most appropriate solution method, and analyze the results of the solution. The primary tools in this case are software tools available to students. The article defines the effects of a problem-oriented approach to training future economists on the example of logistics tasks. The novelty of the work is to identify the features of teaching individual disciplines, considering the need to include methods of solving logistics problems in their structure. In addition, the author’s recommendations on teaching students the selected elements of logistics are new. In particular, students can choose tasks of different levels of complexity, as well as offer their tasks and methods of solving them. Keywords: Logistics · Educational process · Logistics tasks · Optimization methods · Optimization tasks · Transport tasks
1 Introduction Logistics tasks are an integral part of the economic tasks that economists-managers have to solve. First, this is because of the need for transport support for the business, delivery of raw materials, semi-finished products and finished products to both external customers and the movement of goods on the territory of the enterprise. Saving resources and time acts as a target function of most logistics tasks. Solving the tasks of delivering goods on time and at the lowest cost creates additional business opportunities for the company, attracting new customers, and improves the company’s image. The main competitive advantages of the company are: cost reduction because of the optimization of logistics processes, because of reduction of transportation costs; guaranteed fulfillment of orders in the required volume and in the required time [1]. Cargo transportation is an important © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 921–930, 2023. https://doi.org/10.1007/978-3-031-36115-9_82
922
N. Mutovkina
element of the economic system of any company. Transportation concerns not only the movement of goods outside the company but also the distribution of resources within it [2]. As practice shows, in their professional activities, graduates in Economics often have to solve problems related to the need to minimize the cost of delivering goods from suppliers to consumers, the tasks of transport logistics. To solve these problems, the following actions are necessary: 1) To collect and analyze information about the material flows available in the company. 2) To clarify the location of consumers and find out possible ways of delivering goods. 3) To determine the transport for delivery and choose the optimal transport, considering the cost of its operation and travel time. 4) To develop the optimal pathway to deliver goods to customers. The work aims to prove the effectiveness of logistics tasks in the educational process for the professional training of economists. The article is based on the results of a study designed to answer the fundamental question: “How does the solution of logistics tasks form students’ knowledge, skills, skills and professional competencies, and contributes to effective professional activity?”. 1.1 Literature Review Logistics issues and solutions to logistics problems are considered in many educational and methodological works on optimization and transport logistics. So, in the work [2] a decision support system is proposed to assess the costs associated with logistics processes. This system allows you to calculate the economic, environmental, and social costs of the logistics process to ensure sustainable logistics. In the article [3]. The authors propose a solution to the problem of planning the movement of railway transport, considering the following features: besides planning service intervals for trains, the article additionally solves the problem of securing each train to the railway track. A mathematical model and a method for solving the scheduling problem are presented. The key feature of this mathematical model is that it does not use Boolean variables, but works with combinatorial objects (sets of permutations). Sometimes delivering goods is complicated by the need to consider the specifics of the cargo being transported. For example, it can be extremely fragile or perishable goods. Here, a condition for the safety of goods is added to all other conditions. In the publication [4]. A system for monitoring the temperature regime of goods delivery is proposed. This system can be linked to the client’s system so that all data on the temperature of processing, storage, and transportation can be sent to clients via the Extranet and the suppliers’ Web server. The customer system can determine whether the temperature data is normal before the products are received. Similar technology is also considered in the works of [5, 6]. Many logistical problems are reduced to solving linear programming (LP) problems. Linear programming is a field of optimization theory that includes optimization problems in which the objective function and constraints have a linear form [7]. Linear programming is widely used to solve a variety of financial, marketing, production, and agricultural problems. In finance, LP is used for budgeting, asset allocation, financial
Solving Logistical Problems by Economics Students
923
planning, etc. [8, 9]. In marketing, LP is used to solve problems related to marketing research and media selection [10]. In addition, LP is used to compile an optimal production program, and optimize the product range. Similarly, LP is used in the agricultural sector to optimize the structure of crops, as well as the distribution of resources (water, land, fertilizers, etc.) [11]. LP is not a new concept, but the ease of use and accuracy of the results got to make it an indispensable tool in the hands of a decision-maker. 1.2 Software Tools for Solving Logistics Tasks Linear programming problems can be solved using special software. Now the user may not even know the subtleties of methods for solving optimization problems to get the optimal solution in a short time. Only knowledge of the formalization of optimization problems is enough. Software having the specified functionality includes such programs as LINDO SS [12, 13], LINGO [14, 15], MathCAD, IOSO, Approx, Xpress Optimizer, Microsoft Excel, etc. In the article [16] a comparative analysis of several software products is presented. According to the results of the analysis, the most convenient programs for solving transport logistics problems are Microsoft Excel and MathCAD. The IOSO and Approx programs have only paid versions, and the convenience of working in these software tools for untrained users is difficult. Xpress Optimizer is a suitable programming environment, but because of the lack of a Russian interface and the lack of a free version, it cannot compete with Microsoft Excel and MathCAD. The advantages of Microsoft Excel are also mentioned in [17, 18]. In this software, there is an optimization tool Solver, with a user-friendly and intuitive interface. Many researchers use Microsoft Excel to solve most optimization problems easily and quickly. For example, in [18] the solution of the problem of controlling uncertainty in income is presented from agriculture using a model of fuzzy multi-purpose linear programming. In classroom classes with students at Tver State Technical University, Microsoft Excel is most often used to solve logistical problems. Here, this is due not only to the versatility and ease of use of this software, but also to the availability of a licensed version of Microsoft Office at the university.
2 Features of Solving Logistics Problems 2.1 Logistical Tasks in the Professional Activity of Economists In the logistics activity of the enterprise and its subsequent optimization, the following mathematical problems are most often formed: forecasting demand for goods, determining the optimal stock of goods in stock, calculating the optimal order size, and forming a transportation plan from a group of suppliers to a group of buyers. The definition of the type of task and its formulation isis carried out depending on the goals of the company. Therefore, students must first analyze the subject areas according to the principles of a systematic approach. Then it is necessary to understand what data is available to the analyst and what data is missing. If the numerical information is presented in full, then the logistic problem belongs to well-formalized problems and can be
924
N. Mutovkina
solved by standard mathematical methods. If there is an incompleteness of information, then two approaches are possible here. The first approach is to restore information using interpolation, extrapolation, and other methods. The second approach is used if information recovery is impossible and comprises expert evaluation. Both methods apply to problems that are difficult to formalize. Logistics tasks are usually well formalized, but there are some uncertainties in their formulation and solution. For example, modern logistics concepts imply that determining the amount of demand for goods is the responsibility of the logistics services of the enterprise, therefore, demand forecasting can be attributed to logistics tasks [19]. However, forecasting the level of a time series is associated with difficulties in identifying trend, seasonal, and cyclical components. In addition, there is always the influence of a random deviation formed for various reasons that do not depend on the researcher. In determining the optimal stock, the storage capacity is usually taken as a constant value, and the number of stocks stored on it is a variable value. However, this statement is correct with a certain conditionality. For example, it is assumed that the warehouse area cannot be physically increased, or that it is impractical from an economic point of view. For the transport problem to be solvable, it is necessary and sufficient that the total cargo stocks at the points of departure are equal to the total needs at the destinations. The model of the transport problem satisfying this condition is called closed. If the specified condition is not met, then the model is open. If the stock exceeds the need, a fictitious destination is introduced, the cost of delivering the goods to which is zero. If the demand exceeds the stock, a fictitious departure point is introduced and delivery costs are also assumed to be zero [20]. Students must analyze the features of the problem and only after that make up its mathematical model. 2.2 Implementation of the Mathematical Model A mathematical programming problem can be written: ⎧ ⎪ ⎨ F = f (xi ) → max(min), ⎪ ⎩ ϕ (x ) ≤ b , j = 1, m, i = 1, n, x ∈ D i i j i i
(1)
The first expression is the objective function, in which xi there are the desired task variables. The second expression is the constraints of the task. This model is used to formalize the transport task. However, when setting a transport task, it is necessary to consider its features: 1) The system of constraints is a system of equations, the transport problem is given in the canonical form. 2) The coefficients for the variables of the constraint system are equal to one or zero.
Solving Logistical Problems by Economics Students
925
3) Each variable enters the restriction system twice: once for orders, the second time for needs. Implementing the model is carried out in a software environment accessible to students and teachers. As an additional work that develops the student’s thinking, implementing the model in other software environments can be considered, followed by a comparison of the results and features of the work. Students develop a mathematical model of the problem first under the guidance of a teacher, then independently. When performing independent work, students can choose problem situations themselves for subsequent analysis, formulation, and solution of the optimization problem. Students can also choose methods for solving the optimization problem. Non-standard, creative approach to solving problem situations is welcome and has a positive effect on the evaluation of the student’s work. 2.3 Analysis of the Results Obtained In optimization theory, there is a concept of an acceptable, but not optimal, solution. An acceptable solution is a distribution of the desired quantities that satisfies all the conditions of the problem but does not provide an extremum of the objective function. Such situations occur often from the subjective preferences of participants in the decision-making process. Objective factors can also influence the solution to the optimization problem. In particular, when solving a transport problem, the following circumstances may arise. 2.3.1 Blocking Delivery If transportation from sender i to consume j is prohibited (for example, because of bad roads), then some large number A (A → ∞) should be taken as the transport cost coefficient in the cell (i, j). Then, in the optimal solution (on a minimum) this cell will be zero. If this cell still turns out to be the basic one, then this shows the insolvability of the problem because of prohibited transportation. 2.3.2 Bandwidth Limitation Suppose, for example, it is required to consider the limited capacity of the road from sender i to consume j. Here, the column of the consumer j in the distribution table should be divided into two parts. The left column shows the demand equal to the limited capacity, and the right column shows the remaining amount of demand. In the cell corresponding to the transportation from the sender i to the consumer j of the right column, the transportation should be blocked by adding the coefficient A. In this case, the limited capacity of the road will be automatically considered when solving the problem. 2.3.3 Accounting of Production Costs There are tasks in which it is necessary to consider the costs not only for the transportation of products but also for the production of these products.
926
N. Mutovkina
For example, there are n points of production, from which the manufactured products must be delivered to m points of consumption. Then the generalized costs from produce i to consume j are equal to the sum of the costs of producing products at point i and the costs of transporting products from point i to point j. It is required to find such an optimal product distribution plan so that the total costs are minimal. Here, the coefficients of the objective function of a transport problem of this type are the costs equal to the sum of the costs of producing a unit of production at point i (ci ) and the costs of transporting a unit of production from point i to point j (pij ): zij = ci + pij
(2)
Students are encouraged to consider various options for development and analyze how the solution to the problem will change.
3 An Example of a Lesson Dedicated to Solving Logistical Problems The following is an example of a practical lesson conducted under the guidance of the author of the article in the course’s framework “Marketing” on the topic “Sales policy of the company”. The lesson is conducted with third-year students studying toward “Economics”. At the beginning of the lesson, students are explained the theory, then a specific task is given, which is solved in the chosen software environment. Such a software environment is Microsoft Excel. 3.1 Initial Data and Problem Statement Individual marketing strategies are usually developed for each type of market and its constituent types of goods. This is also typical for transport services, when the major objectives of marketing are the orientation of operational work to meet the needs of customers in the transportation of goods and passengers, increasing the coverage of the market with transport services. The solution to these tasks provides an increase in the company’s income. Based on the analysis of the problem area, students together with the teacher carry out the linguistic formulation of the problem. There are three computer supply points: Warehouse No. 1, Warehouse No. 2 and Warehouse No. 3. There are also five stores: “Terabyte”, “Leader”, “Expert”, “Digital Service”, “Office Equipment”, purchasing computers for retail sale. The warehouses have the following number of computers: Warehouse No. 1–200 pcs., Warehouse No. 2–250 pcs., Warehouse No. 3–200 pcs. It is required to deliver: to “Terabyte”–190 pcs., to “Leader”–100 pcs., to “Expert”–120 pcs., to “Digital Service”–110 pcs., and to the “Office Equipment” store 130 pcs. The cost of delivering one computer from each warehouse to each store is represented by the C matrix. ⎛ ⎞ 28 27 18 27 24 C = ⎝ 18 26 27 32 21 ⎠ 27 33 23 31 34 It is necessary to find the best way to distribute computers.
Solving Logistical Problems by Economics Students
927
3.2 Task Solving and Interpretation of Results According to the method of teaching students to solve logistical problems presented in clause 2.2, the next stage is to form a block of source data on a Microsoft Excel sheet. Such a block is shown in Fig. 1. Attention should be paid to which type of transport task belongs: open or closed.
Fig. 1. Initial data
Students independently form a block of source data. This work takes 5–7 min. The location of some elements on the Excel sheet may differ for different students, but this only applies to the selected design style. The requirements of data integrity and ease of perception must be fully met. Then students prepare a block to solve the problem. The distribution table has exactly the same appearance as the table with the source data. The SUM() function is used in the “Needs” row and in the “Stocks” column. The cells in the distribution table remain unfilled (Fig. 2). Filling in them will occur automatically after using the “Solver” add-in. In some cell after the distribution table, the formula of the objective function is written: SUMPRODUCT(B3:F5;B10:F12). The first range contains computer delivery costs, and the second range contains computer distribution volumes. Students are also given 5–7 min to implement this procedure.
Fig. 2. Distribution table of the transport task
After the students have implemented this stage, the teacher checks the correctness of the result got and the students search for a solution. In the “Search for a solution” window, students set all the parameters, restrictions and get the desired values for the distribution of computers in stores and the cost of their delivery (Fig. 3). When setting restrictions, it is necessary to consider that the quantity of goods transported cannot be fractional and negative. Students are given 3 to 5 min to get the optimal solution in the software environment. After that, students analyze the resulting solution and possible alternative solutions.
928
N. Mutovkina
Fig. 3. Solving the transport task
3.3 Analysis of Alternative Solutions The resulting solution shows that computers for retail sale should be delivered from the first warehouse to the stores “Leader” (100 units), “Office Equipment” (70 units) and “Expert” (30 units). The major consumer of the products stored in the second warehouse is the Terabyte store (190 pieces). From the third warehouse, computers are distributed to “Digital Service” and “Expert” stores. To increase the assessment for the work done, students are recommended to consider the situations listed in the second section of the article. The results of the analysis are drawn up by students in writing, as a report on the work performed and submitted to the teacher for verification. Depending on the number of situations considered, 20 to 40 min can be provided for this work. According to the results of the check, the teacher announces the grades. The best works can be recommended for preparing reports in the framework of the following practical classes or scientific conferences.
4 Summary and Conclusion The skills gained by students in solving logistical problems in the learning process allow them to subsequently: 1) To analyze the external and internal environment of the company, identify its key elements and assess their impact on the company. 2) To apply analytical and computational methods in practice for making managerial decisions. 3) To organize and ensure material and financial flows in a timely manner both in the company and abroad. 4) To use logistics tools in supply and distribution management. 5) To apply knowledge and solve problems in inventory management using various models of inventory control.
Solving Logistical Problems by Economics Students
929
6) To evaluate the efficiency and develop the logistics process in the company’s warehouses. In the study’s course, it was found that Microsoft Excel is a very effective tool for solving logistics problems. The built-in service “Solver” allows you to quickly and costeffectively get solutions to logistical problems related to the distribution of shipments to customers.
References 1. Anitha, P., Patil, M.M.: A review on data analytics for supply chain management: a case study. Int. J. Inf. Eng. Electron. Bus. 10(5), 30–39 (2018). https://doi.org/10.5815/ijieeb.2018.05.05 2. Benotmane, Z., Belalem, G., Neki, A.: A cost measurement system of logistics process. Int. J. Inf. Eng. Electron. Bus. (IJIEEB) 10(5), 23–29 (2018) 3. Grebennik, I., Dupas, R., Lytvynenko, A., Urniaieva, I.: Scheduling freight trains in rail-rail transshipment yards with train arrangements. Int. J. Intell. Syst. Appl. (IJISA) 9(10), 12–19 (2017) 4. Ting, P.-H.: An efficient and guaranteed cold-chain logistics for temperature-sensitive foods: applications of RFID and sensor networks. IJIEEB 5(6), 1–5 (2013) 5. Weihua, G., Tingting, Z., Yuwei, Z.: On RFID application in the information system of rail logistics center. IJEME 3(2), 52–58 (2013) 6. Weihua, G., Yuwei, Z., Tingting, Z.: Research on RFID application in the pharmacy logistics system. IJEME 2(8), 13–19 (2012) 7. Dantzig, G.B.: Linear Programming and Extensions. Princeton, NJ (1963) 8. Ajayi Dr, Ibrahim, D.: Application of linear programming in investment portfolio selection (using Microsoft Excel 13). Int. J. Acad. Res. Bus. Arts Sci. (IJARBAS.COM), 3, 1–27 (2021) 9. Silva, P.M.S., Moreira, B.C.M., Francisco, G.A.: Linear programming applied to finance - building a great portfolio investment. Revista de Gestão, Finanças e Contabilidade 4(3), 107–124 (2014). https://doi.org/10.18028/2238-5320/rgfc.v4n3p107-124 10. Uday, S.V., Hamritha Chaudhary, G.: Linear programming in market management using artificial intelligence. In: Vijayan, S., Subramanian, N., Sankaranarayanasamy, K., (eds) Trends in Manufacturing and Engineering Management. Lecture Notes in Mechanical Engineering. Springer, Singapore, pp. 845–851 (2021). https://doi.org/10.1007/978-981-15-4745-4_73 11. Alotaibi, A., Nadeem, F.: A review of applications of linear programming to optimize agricultural solutions. Int. J. Inf. Eng. Electronic Bus. (IJIEEB) 13(2), 11–21 (2021) 12. Volchkov, V.M., Tarasova, I.A., Shvedov, E.G.: Solving integer programming problems in the LINGO computer package: textbook. Volgograd: VolgSTU (2020) 13. Kaur, J., Tomar, P.: Multi objective optimization model using preemptive goal programming for software component selection. Int. J. Inf. Technol. Comput. Sci. (IJITCS) 7(9), 31–37 (2015) 14. Goldstein, A.L.: Modeling in LINGO. Bulletin of Perm National Research Polytechnic University. Electrical Engineering, Information Technology, Control Systems, 18, 25–38 (2016) 15. Mikulich, E.M., Podina, K.V., Gudkov, V.A., Volchkov, V.M.: Application of the “LINGO” linear programming package for solving logistical problems of optimizing the cargo delivery process. Bull. Transp. 11, 35–38 (2012) 16. Dubenetskaya, E.R.: Training of economic specialties students in the solving of optimization problems using specialized software. Sci. Bull. MSIIT 6(38), 73–79 (2015)
930
N. Mutovkina
17. Render, B., Stair, R.M., Hanna, M.E.: Quantitative analysis for management. 11TH EDITI. Pearson (2012) 18. Kumari, P.L., Reddy, G.K., Krishna, T.G.: Optimum allocation of agricultural land to the vegetable crops under uncertain profits using fuzzy multiobjective linear programming. J. Agric. Vet. Sci. 7(12), 19–28 (2014) 19. Sergeev, V.I., Jeljashevich, I.P.: Logistics of Supply: Textbook for Bachelor’s and Master’s Degrees. Yurayt Publishing House, Moscow (2016) 20. Volkova, I.I., Prudnikova, O.M.: Methods of Optimal Solutions for the Bachelor of Economics: Textbook. Ukhta State Technical University Publishing House, Ukhta (2015)
Exploration and Practice of Ideological and Political Construction in the Course of “Container Multimodal Transport Theory and Practice” for Application-Oriented Undergraduate Majors—Taking Nanning University as an Example Shixiong Zhu, Liwei Li, and Zhong Zheng(B) Nanning University, Nanning 530200, Guangxi, China [email protected]
Abstract. As an important professional course for logistics majors, the course “Theory and Practice of Container Multimodal Transport” is a relatively new professional course to adapt to the rapid development of container multimodal transport. This paper defines the basic thinking based on the work process orientation. On this basis, it analyzes the necessity of strengthening ideological and political education in the field of container multimodal transport on how to implement the ideological and political requirements of the curriculum, excavates the ideological and political elements, determines the weight of different ideological and political elements, clarifies the thinking of ideological and political education of the curriculum, emphasizes the pertinence of ideological and political education of the curriculum, and discusses the specific problems. The research of this paper focuses on helping to improve the training quality of application-oriented logistics professionals. It ponders and expounds the necessity, construction ideas and construction paths of the ideological and political construction of this course, and carries out relevant practice in the teaching process, and has achieved good results. Keywords: Container multimodal transport theory and practice · Curriculum ideological and political · Curriculum construction · Teaching practice
1 Introduction In the new economic situation and social environment, the construction of new land and sea routes in the west is in full swing, and new requirements are put forward for the quantity and quality of qualified personnel for multimodal transport [1–3]. It is the mission of colleges and universities given by the times to continuously deliver more and better talents to the society [4–6]. In addition to professional knowledge, qualified talents must also be politically and ideologically competent. To improve the comprehensive quality of students in an all-round way, moral education contained in professional courses © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 931–943, 2023. https://doi.org/10.1007/978-3-031-36115-9_83
932
S. Zhu et al.
is duty-bound. Curriculum ideological and political education refers to a comprehensive educational practice that integrates ideological and political education elements, including theoretical knowledge, values and spiritual pursuit of ideological and political education, into the curriculum in professional courses and general courses, and exerts a subtle influence on students’ ideological awareness and behavior, so as to achieve the goal of moral education and education [7, 8]. It is not a separate course, but a new education and teaching mode. In recent years, the national education management department has repeatedly stressed that we should adhere to the central link of moral education and the cultivation of people, carry out ideological and political work throughout the whole process of teaching, achieve all-round education for all staff, and strive to create a new situation for the development of higher education in China [9]. We should take the comprehensive promotion of the ideological and political construction of the curriculum as a strategic measure to implement the fundamental task of building morality and cultivating people, integrate values into knowledge teaching and ability training, and help students shape correct world outlook, outlook on life and values [10–14]. It can be seen that the ideological and political construction of curriculum has been promoted to an important task of comprehensively improving the quality of talent training in colleges and universities. As people’s teachers of professional courses in the new era, it is incumbent upon them to participate in and promote the ideological and political construction of courses [15]. Taking the course “Theory and Practice of Container Multimodal Transport” (hereinafter referred to as “Container Multimodal Transport”, or CMT for short) as an example, this paper explores and practices the ideological and political construction of the course of logistics engineering in Nanning University (application-oriented undergraduate college).
2 The Necessity of Ideological and Political Construction of “CMT” Course for Application-Oriented Undergraduate The training objectives of application-oriented undergraduate logistics engineering students are different from those of research-oriented undergraduate majors or higher vocational colleges [16]. The former focuses on training enough theoretical knowledge to guide production practice, strong hands-on ability to engage in post work, direct employment and “zero distance” employment. It emphasizes the direct impact of application-oriented undergraduates on production practice [17]. Therefore, the ideological and political construction of the CMT course for application-oriented undergraduates emphasizes not only knowing the actual performance of the field of the course, what problems exist, and what to do in action, but also knowing why to do so. Obviously, the ideological and political construction and implementation of curriculum are more urgent for application-oriented undergraduate students [18–20]. The necessity of ideological and political construction in the course of “container multimodal transport” for application-oriented undergraduate students is at least reflected in the following aspects. (1) The ideological and political construction of the CMT course is conducive to enhancing the four self-confidence of application-oriented undergraduates [21, 22]. Transportation, including multimodal transport, plays a key role in the development
Exploration and Practice of Ideological and Political Construction
933
of the national economy. Compared with European and American countries, China’s D multimodal transport has shown the world’s first in container manufacturing capacity, technology research and development level and industrial supporting cluster level for 25 consecutive years. The contribution of “Made in China” to global multimodal transport cannot be underestimated. During the COVID-19, China’s multimodal transport volume grew against the trend, and its export maintained a double-digit high-speed growth, resulting in a huge surplus. As an all-weather container transport product on the Eurasian Continental Bridge, which is less affected by climate and environment, has faster transport speed than sea transport, and is far cheaper than air transport, China Europe Express is increasingly welcomed by countries and regions along the “the Belt and Road”. When teaching the content of international container multimodal transport, the teacher showed the development achievements of China’s multimodal transport to students, so that students could deeply understand the important role of China’s multimodal transport in the world economic development and the economic development along the “the Belt and Road”, and more aware of China’s role as a major country in promoting the construction of a community with a shared future for mankind. In the process of knowledge learning, students should establish a global perspective, firmly adhere to the “four self-confidence”, and are willing to demonstrate the strength and responsibility of China’s road, theory, system and cultural self-confidence in future work and international exchanges. (2) The ideological construction of the CMT course is beneficial to the applicationoriented undergraduates to enhance their sense of cooperation. The operation process of container multimodal transport involves not only a variety of transportation modes such as road, railway, waterway and air, but also multi-way transport, multi-operator transport, multi-industry cooperation and the impact of multiple countries and their laws and regulations. It is necessary to establish a good cooperative relationship with all parties. It is necessary to encourage and teach students to firmly establish the principle of human kindness and improve their awareness and ability of cooperation through curriculum teaching. (3) The ideological and political construction of CMT course is conducive to strengthening the spirit of inquiry of application-oriented undergraduates. At present, China’s container multimodal transport still has many problems. For example, the application of information technology is backward, the information system supporting multimodal transport business and management has not yet been formed, and the intermodal transport business transactions still need complex paper documents; The efficiency and quality of multimodal transport services need to be improved; The loading and unloading facilities, documents and information of the freight station cannot be transferred in time. In addition, the coordination of multimodal transport system is not ideal. The combination of different modes of transport lacks research and planning, and has not formed a unified multimodal transport operation network. The lack of close cooperation between railway and maritime and inland water transport, and the poor connection between related enterprises are all problems facing multimodal transport. It is necessary for application-oriented undergraduates to participate in the research, development, upgrading and application of the system and strengthen the spirit of exploration.
934
S. Zhu et al.
(4) The ideological and political construction of the curriculum is conducive to enhancing the sense of hardship of application-oriented undergraduates. The container multimodal transport business, especially the international container multimodal transport business, often has many links, a long distance, is subject to many countries, regions and their laws and regulations, and has great risks. It is necessary to enhance the awareness of hardship and the ability to make emergency response for application-oriented undergraduates through the course of ideological and political education.
3 Basic Thinking of Ideological and Political Construction of the Course “CMT” The essence of ideological and political courses is to establish morality and cultivate people. The goal is to cultivate college students who have both professional knowledge and correct ideals and beliefs, value orientation, political belief, social responsibility and patriotism, as well as talents who have the ability to reason and distinguish right from wrong. The basic idea of ideological and political construction of this course is to take the thought of socialism with Chinese characteristics in the new era as the guidance, focus on cultivating applied multimodal talents for the national development and national rejuvenation, clarify the responsibilities and requirements of the main body of ideological and political construction of the course, and build the carrier of ideological and political construction of the course in accordance with the principle of “full integration of professional knowledge teaching, ability training and value shaping”, Control the key factors that affect the effect of ideological and political construction of the curriculum and strengthen the assessment of the implementation process of ideological and political construction of the curriculum to promote the effectiveness of ideological and political construction of the curriculum. 3.1 Mining Ideological and Political Elements of Curriculum Modules The ideological and political elements of the curriculum should be thoroughly and comprehensively excavated to provide an important basis for improving the quality of ideological and political construction of the curriculum. The factors to be considered include: whether the textbooks, syllabus and courseware are applicable, whether the ideological and political elements of the curriculum module are mined in combination with the professional characteristics, whether the teaching team has sufficient awareness and attention, and whether the division of labor and cooperation is effective. The results are shown in Table 1. In the table, “container transport”, “multimodal transport” and “container multimodal transport” are abbreviated as “CT”“MT” and” MTC” respectively. M1, M2, M3 and M4 represents different modules in turn. Task 1 to Task 8 are denoted by T1,…, T8 separately.
Exploration and Practice of Ideological and Political Construction
935
Table 1. Ideological and political elements and objectives of “container multimodal transport” course Module
Task
Mapping point
Objectives
M1: Basic knowledge of CMT
T1: Learn to know MT
1) The current situation of CT in China; 2) China’s MT increased during the epidemic; 3) Integrity of MT operators
1) Set up correct three views and enhance professional confidence; 2) Cultivate the awareness of integrity, legal system and social responsibility, and be a qualified multimodal transport operator
T2: Prepare equipment and tools for MT
1) The necessity and scarcity of resource of MT 2) Domestic equipment on automatic CT; 3) Qinghai-Tibet Railway and other projects (video)
1) Learn to cooperate; 2) The belief in building a country; 3) Establish craftsmanship spirit
T3: Familiar with MT logistics nodes
1) Global container port top 10 in 2021; 2) Automated CT domestic equipment; 3) The model team set a new world record
1) Adhere to openness and innovation; 2) Loving and dedicated; 3) Craftsman spirit
M2: CMT business process
T4: 1) China’s customs Understand customs clearance speed; clearance management (2) Customs information application; (3) The impact of RCEP on CMT
1) Keep learning; 2) Expand international vision
T5: Familiar with MT billing, documents and operation process
1) Keep improving, find and solve problems; 2) Have professional quality and meet the requirements of post ability
1) “Consignment, payment, document signing and bill” one-stop mode of MT; 2) Calculation of multimodal freight
(continued)
936
S. Zhu et al. Table 1. (continued)
Module
Task
Mapping point
Objectives
M3: Risk management of CMT
T6: Familiar with insurance and claims settlement
1) Policies for import, export and customs declaration of goods; 2) Compliance of MT; 3) MT risk management
1) Have legal awareness and bottom line awareness; 2) Serious and responsible attitude
M4: Design and implementation of CMT scheme
T7: 1) Cost budget and Design MT scheme of control of container container cargo multimodal transport; 2) Market analysis; 3) Route selection, site selection and layout planning
1) Pay attention to efficiency and fairness; 2) Good at objective and systematic analysis
T8: Organize and implement MT of container cargo
1) Patriotic and dedicated; 2) Be bold in innovation
1) The opportunity of the “the Belt and Road”; 2) Development history of international CMT; 3) The development trend of international trains
3.2 Weight Analysis of Curriculum Ideological and Political Elements Analytic Hierarchy Process (AHP) is a systematic method to solve complex multiobjective decision-making problems [23–26], which was put forward in the early 1970s for the US Department of Defense when studying the topic of “power distribution based on the contribution of various industrial sectors to the national welfare”. The core principle of the method is to decompose the problem into different components according to the requirements of the overall goal, According to the interrelation, influence and subordination among factors, factors are aggregated and combined at different levels to form a multi-level analysis structure model, so as to rank the importance or advantages and disadvantages of different elements to the overall goal [27, 28]. In order to carry out the ideological and political construction of the curriculum more pertinently, according to the principle of AHP method, the importance of different indicators is determined by comparison between two groups, and the preliminary score is made, which provides basic data for the development of AHP. The course teaching team will discuss and grade the ideological and political elements of the course, absorb the opinions of similar course experts from other institutions, evaluate the importance of the course elements through research and telephone consultation, and compare and judge different indicators. There are 8 main elements in the curriculum determined after the survey, as shown in Table 2.
Exploration and Practice of Ideological and Political Construction
937
The judgment matrix is constructed according to the results of comparison and judgment, as shown in Table 3. Table 2. Main factors in course analysis and evaluation Code
Step
Content
S1
Step 1
Understand multimodal transport
S2
Step 2
Prepare multimodal transport tools and equipment
S3
Step 3
Familiar with multimodal transport logistics nodes
S4
Step 4
Understand multimodal customs clearance management
S5
Step 5
Know multimodal transport billing, documents and operation process
S6
Step 6
Be familiar with multimodal insurance and claim settlement
S7
Step 7
Design multimodal transport scheme of container cargo
S8
Step 8
Organize and implement multimodal transport of container goods
According to the eight main factors identified in the table, the judgment matrix of the importance of ideological and political elements of the curriculum shown in Table 3 can be obtained. The analysis process is as follows. After the weight score is obtained, hierarchical ranking analysis can be carried out. The core index is the determination of the eigenvector and the weight value. The eigenvector is obtained by the square root method, and the weight value is normalized so that the sum of all elements is 1. The AHP hierarchical analysis results are carried out to calculate the maximum characteristic root of the judgment matrix, and the consistency index CI is calculated. The analysis results are shown in Table 4. According to the principle of AHP method, CI equal to 0 indicates complete consistency. The closer CI is to Table 3. Judgment matrix of the importance of ideological and political elements of the curriculum Index
S1
S2
S3
S4
S5
S6
S7
S8
S1
1
1/2
1/3
1/4
1/8
1/6
1/9
1/9
S2
2
1
1/2
1/3
1/7
1/5
1/6
1/7
S3
3
2
1
1/5
1/6
1/4
1/6
1/6
S4
4
3
2
1
1/4
1/2
1/5
1/5
S5
8
7
6
4
1
2
1/2
1/2
S6
6
5
4
2
1/2
1
1/3
1/3
S7
9
7
6
5
2
3
1
1
S8
9
7
6
5
2
3
1
1
938
S. Zhu et al.
0, the more satisfactory consistency is displayed. The larger CI value, the more serious the inconsistency is. Table 4. AHP hierarchy analysis results Items
Eigenvector
Weight (%)
S1
0.239
2.009
S2
0.344
2.88、
S3
0.493
4.138
S4
0.767
6.444
S5
2.256
18.952
S6
1.382
11.61
S7
3.212
26.981
S8
3.212
26.981
Maximum characteristic root
CI
8.241
0.034
For the purpose of further measurement of the size of CI, the random consistency index RI and consistency ratio CR (CR = CI/RI) are introduced. Generally, when CR < 0.1, the degree of inconsistency is considered to be within the allowable range. The consistency test of the analysis results is carried out according to the principle of AHP method, as shown in Table 5. Table 5. One-time inspection results One-time inspection Maximum characteristic root
CI
RI
CR
Results One-time inspection
8.241
0.034
1.404
0.025
Pass
The calculation results of the analytic hierarchy process show that the maximum characteristic root is 8.241, and the corresponding RI value is 1.404 according to the RI table, so CR = CI/RI = 0.025 < 0.1. Through a one-time test, it shows that the analysis process is effective, and the weights of different ideological and political curriculum elements are determined according to the order of weight analysis results in Table 4. 3.3 Weight Analysis Results of Curriculum Ideological and Political Elements According to Table 4 of the above analytic hierarchy process (AHP) analysis results, it is concluded that: (1) the weight of knowing multimodal transport is 2.009%, (2) the weight of preparing multimodal transport tools and equipment is 2.886%, (3) the weight of knowing multimodal transport logistics nodes is 4.138%, (4) the weight of knowing multimodal transport customs clearance management is 6.444%, (5) the weight of
Exploration and Practice of Ideological and Political Construction
939
knowing multimodal transport billing, documents and operation process is 18.952%, (6) The weight of knowing multimodal transport insurance and claim settlement is 11.61%, (7) the weight of designing multimodal transport scheme of container goods is 26.981%, and (8) the weight of organizing and implementing multimodal transport of container goods is 26.981%. The weights of (5), (7) and (8) are the highest. Therefore, in the ideological and political process of the course, it is necessary to strengthen the familiarity with multimodal transport billing, documents and operation processes; Design multimodal transport scheme of container cargo; Organizing and implementing the three aspects of container cargo multimodal transport to strengthen ideological and political construction and training is also the key link to improve the quality of ideological and political education in this course. Because these three aspects are also the core knowledge points of the course, it is necessary to combine the course elements and business links more closely, so that the ideological and political effect of the course can be more prominent.
4 Ideological and Political Construction Path and Practice of CMT Course 4.1 Select Teachers with Outstanding Ideological and Political Ability to Teach, and Create a Capable Subject of Ideological and Political Construction of Curriculum The main body of curriculum ideological and political construction is the curriculum teachers, whose ability and level are directly related to the quality of curriculum ideological and political. Therefore, the first is that the teachers put forward the ideological and political requirements of the curriculum every time the curriculum is arranged. The teachers are required to take moral cultivation as the primary task of student education and training. The second is to give priority to professional teachers with outstanding ideological and political abilities and strong sense of responsibility. The third is to take the ideological and political ability requirements as the conditions for arranging further study and evaluating the best, and promote professional teachers to consciously improve the ideological and political ability of the curriculum. The curriculum ideological and political requirements of the selected teachers are clear, forming the ability requirement orientation. The teachers consciously improve the curriculum ideological and political ability, actively shoulder the responsibility of cultivating talents and educating people, let the ideological and political awareness take root in the heart, teach by example, and help the application-oriented undergraduates form correct values and ideals. 4.2 Compile Applicable and Visible Teaching Materials, Outlines and Courseware, and Build a Useful Course Construction Carrier Reflecting Ideological and Political Content Textbooks, syllabus and courseware are the necessary carriers of curriculum teaching. They are not only required for the teaching of professional knowledge and vocational skills training, but also for the ideological and political construction and implementation
940
S. Zhu et al.
of the curriculum. They need to be well constructed. In the process of teaching, we can use the task book, case analysis and other documents to arrange teaching activities, group training, watch with the help of network video, and further expand the teaching content, or transmit information, or assign tasks, enrich students’ horizons, so that students can increase their interest in the content of ideological and political education, and improve the effect of ideological and political learning of the course. 4.3 Plan and Design a Number of Curriculum Implementation Rules, Including Teacher Evaluation and Collaborative System, to Form a Guarantee Mechanism for Establishing Morality and Cultivating People The ideological and political construction of curriculum is a systematic project, which needs the support and help of all relevant aspects to achieve good results. At present, in addition to the practical needs and the requirements of the Party and the country, a number of curriculum construction and implementation rules need to be included, such as teacher evaluation, collaborative education system, interconnection mechanism, constraint mechanism, incentive mechanism, and so on. It needs to emphasize top-level planning and design, form a security mechanism based on moral education, promote the healthy and sustainable development of curriculum ideology and politics, and truly cultivate existing patriotism, broad international vision, innovation Craftsmen and patriots with good team spirit, professional quality and strong social responsibility, as well as socialist cause builders and new logistics comprehensive talents with rich knowledge and solid professional skills. 4.4 Add Value-Leading Indicators and Strengthen the Process Assessment and Evaluation of Ideological and Political Teaching The application-oriented undergraduate students are the object of the ideological and political construction of the curriculum and the main body of the teaching class. All measures of the teaching reform are carried out around the adult talents of students, which is to make them move through the assessment of the main body of the teaching class. The assessment idea is to add an assessment index of “ideological and political performance”, including the students’ ideological and political performance (60 points) into the assessment scope, and assess them together with their personal performance (professional quality, 60 points), work performance (100 points), methods and skills (40 points) and others (40 points). Each task will be assessed once, with a full score of 300 points. At the beginning of the course, students should be clearly informed of the detailed rules of the process assessment to improve their learning initiative, enthusiasm and participation. The main observation points of the “ideological and political performance” assessment indicators are positive ideological performance, support for the Party, love for the motherland, and the absence of rebellious words and deeds of love for the Party and popular “good people in the workplace”. There are multiple versions of the ten characteristics of “good people”. After sorting out 15 characteristics, the 10 characteristics and full marks approved by the majority of applied undergraduates are selected as the evaluation indicators and criteria. The 10 indicators and full score criteria are: integrity: adhere to the truth and act with conscience; Honesty: seek truth from facts
Exploration and Practice of Ideological and Political Construction
941
and do not resort to fraud; Kindness: always be grateful, be kind and do good deeds; Kindness: be willing to suffer losses and think for others; Humility: look down on yourself; Tolerance: can tolerate people with different opinions; Persistence: determination, patience and perseverance in doing things; Keep faith: words must be done, and faith must be rewarded; Capable: positive, optimistic, proactive, and able to handle things well; Innovation: strong sense of breakthrough, good at finding and solving problems.
5 Conclusion The purpose of implementing curriculum ideological and political education is to train students to become excellent talents with morality, ideals and all-round development. The ideological education of the “container multimodal transport” course explored the practical significance, basic thinking and path of the course construction, and implemented the practice of “creating the main body of ideological and political construction of lean courses, mining the ideological and political elements of the course modules, diversifying teaching forms, building a useful course carrier, forming a moral education guarantee mechanism, and strengthening the process assessment of ideological and political teaching of the course”. Along the way, our teaching team has experienced the teaching of professional knowledge The integration of ability training and value shaping is indeed an effective way to build morality and cultivate people. After analysis, we need to be familiar with multimodal transport billing, documents and operation process; Design multimodal transport scheme of container cargo; Organize the implementation of three aspects of container cargo multimodal transport, strengthen ideological and political construction and training, strengthen the combination of courses and key business links, and improve the teaching effect of ideological and political education. Acknowledgment. This research is supported by (1) The fourth batch of undergraduate core curriculum construction project of Nanning University “Exploration and Practice of Course Construction of ‘Container Multimodal Transport Theory and Practice’” (2022BKHXK06); (2) Guangxi Higher Education Undergraduate Teaching Reform Project “Research and Practice on the Construction of Logistics Virtual Teaching and Research Office from the Perspective of Fourdimensional Linkage of ‘Point, Line, Area and Body’” (2022JGB440); The subject of ideological and political construction of curriculum of Nanning University (2020SZSFK14).
References 1. Hacène H.: Action research to enhance quality teaching. In: Arab World English Journal (AWEJ) May 2019 Chlef University International Conference Proceedings, pp. 4-12 (2019) 2. Vijayalakshmi, V., Venkatachalapathy, K.: Comparison of predicting student’s performance using machine learning algorithms. Int. J. Intell. Syst. Appl. 11(12), 34–45 (2019) 3. Rifat, M.R.I., Imran, A.A., Badrudduza, A.S.M.: Educational performance analytics of undergraduate business students. Int. J. Mod. Educ. Comput. Sci. 11(7), 44–53 (2019) 4. Wang, X., Yingjie, W.: Exploration and practice for evaluation of teaching methods. Int. J. Educ. Manage. Eng. 2(3), 39–45 (2012) 5. Zhiqin, L., Jianguo, F., Fang, W., Xin, D.: Study on higher education service quality based on student perception. Int. J. Educ. Manage. Eng. 2(4), 22–27 (2012)
942
S. Zhu et al.
6. Skedsmo, G., Huber, S.G.: Measuring teaching quality, designing tests, and transforming feedback targeting various education actors. Educ. Assess. Eval. Accountability 32(3), 271– 273 (2020) 7. Xuejian, W., Yan, S.: The connotation, characteristics, difficulties and countermeasures of ideological and political education in the new era curriculum. J. Xinjiang Normal Univ. (Philos. Soc. Sci. Ed.) 2, 50–58 (2020). (in Chinese) 8. Jiwei, Z.: Curriculum ideological and political: meaning, concept, problems and countermeasures. J. Hubei Univ. Econ. 17(2), 114–119 (2019). (in Chinese) 9. Jinyao, R., Qing, W., Chao, G.: Exploration and practice of curriculum ideological and political systematic design in vocational colleges. China Vocat. Tech. Educ. 29, 27–29 (2021). (in Chinese) 10. Yunbo, O.: Foster the sense of Chinese national community in the ideological and political curriculum. Ref. Polit. Teach. Middle Sch. 42, 83 (2022). (in Chinese) 11. Liwei, P.: Shi Xiaorong exploration of the “four full coverage” model of curriculum ideological and political construction in the context of “new engineering.” J. Nat. Insts Educ. Adm. 11, 63–70 (2022). (in Chinese) 12. Shangzi, Z.: The organic combination of ideological and political courses and ideological and political courses: the three-dimensional proof of ideological and political principles. J. Henan Normal Univ. (Philos. Soc. Sci. Ed.) 49(06), 124–130 (2022). (in Chinese) 13. Bailin, C., Yi, L., Jinqiu, Z.: Using data mining approach for student satisfaction with teaching quality in high vocation education. Front. Psychol. 12, 1–8 (2022) 14. Suoming, H., Lijuan, L.: Problems and countermeasures in ideological and political teaching of new engineering courses. Educ. Theory Pract. 42(36), 39–42 (2022). (in Chinese) 15. Chengwen, C., Yiming, W., Jibin, Z., et al.: Research and practice of ideological and political teaching of characteristic professional courses for engineering certification. Packag. Eng. 43(S2), 54–60 (2022). (in Chinese) 16. Adam Lindgreen, C., Di Benedetto, A., Brodie, R.J., Zenker, S.: Teaching: how to ensure quality teaching, and how to recognize teaching qualifications. Ind. Mark. Manage. 100, A1–A5 (2022) 17. Sal, C.: Practitioner research in a UK pre-sessional: the synergy between exploratory practice and student motivation. J. Engl. Acad. Purp. 57, 1–6 (2022) 18. Jiao, L., Ji, F.: An analysis of the path of the ideological and political systematization of university curriculum. China Univ. Teach. 11, 64–71 (2022). (in Chinese) 19. Lin, W.: A probe into the new theory and practical mechanism of curriculum ideological and political education. Ref. Middle Sch. Polit. Educ. 41, 91 (2022). (in Chinese) 20. Fei, L., Dong, L.: Open a new perspective of “three complete education” in the ideological and political practice of college curriculum. Ref. Polit. Teach. Middle Sch. 40, 110 (2022). (in Chinese) 21. Tao, J., Yujuan, S.: Problems and countermeasures in ideological and political construction of university curriculum. Sch. Party Constr. Ideological Educ. 20, 44–46 (2022). (in Chinese) 22. Anna, S.: Perceptions of student motivation and amotivation. The Clearing House J. Educ. Strat. Issues and Ideas 94(2), 76–82 (2021) 23. Xiaomin, X.: Application of analytic hierarchy process. Statistics and Decision (1), 156–158 (2008) (in Chinese) 24. Kong, L.A., Wang, X.M., Yang, L.: The research of teaching quality appraisal model based on AHP. Int. J. Educ. Manage. Eng. (IJEME) 9(29), 29–34 (2012)
Exploration and Practice of Ideological and Political Construction
943
25. Haji, E., Azmani, A., Harzli, M.E.: Using AHP method for educational and vocational guidance. Int. J. Inf. Technol. Comput. Sci. (IJITCS) 9(1), 9–17 (2017) 26. Cheng, J., Chen, S.: A fuzzy Delphi and fuzzy AHP application for evaluating online game selection. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 5, 7–13 (2012) 27. Listyaningsih, V., Utami, E.: Decision support system performance-based evaluation of village government using AHP and TOPSIS methods: secang sub-district of magelang regency as a case study. Int. J. Intell. Syst. Appl. (IJISA) 10(48), 18–28 (2018) 28. Lotfi, F., Fatehi, K., Badie, N.: An analysis of key factors to mobile health adoption using fuzzy AHP. Int. J. Inf. Technol. Comput. Sci. (IJITCS) 12(02), 1–17 (2020)
A Study on Learning Intention of Digital Marketing Micro Specialty Learners Under the Background of New Liberal Arts—Based on Structural Equation Model Yixuan Huang1(B) , Mingfei Liu1 , Jiawei You1 , and Aiman Magde Abdalla Ahmed2 1 School of Management, Wuhan University of Technology, Wuhan 430070, China
[email protected] 2 Faisal AI Islam Bank, Khartoum 999129, Sudan
Abstract. In the context of the new liberal arts, the trend of interdisciplinary integration is becoming increasingly clear. Micro specialty model is a more indepth exploration on the basis of interdisciplinary integration. Taking digital marketing micro specialty as an example, this paper constructs a study model of learning intention of digital marketing micro specialty learners based on stimuliorganism-response theory (S-O-R) and the technology acceptance model (TAM), and explores the relationship between variables using a structural equation model and a test of mediating effect. The findings show that: Self-efficacy, perceived ease of use and perceived usefulness directly affect learning intention, while perceived ease of use indirectly affects learning intention through self-efficacy and perceived usefulness. Educators significantly affect self-efficacy. Instructional media directly affect perceived ease of use and perceived usefulness, as well as selfefficacy and perceived usefulness through the mediating role of perceived ease of use. Instructional content significantly affects perceived usefulness. Keywords: Digital marketing micro specialty · Learning intention · Structural equation model · S-O-R theory · TAM model
1 Introduction Since the official launch of the “Six Excellence and One Top-notch” plan 2.0 of the Ministry of Education in October 2018, the construction of new liberal arts has attracted widespread attention from the society. The new liberal arts are characterized by strategy, innovation, integration and systematization. It is a real revolution in liberal arts education from discipline orientation to demand orientation, from adaptation service to support and guidance, and from specialty segmentation to cross-integration [1]. In this context, the construction of digital marketing micro specialty will undoubtedly be a feasible path for advancing liberal arts education. Facing the demand for innovative talents in national strategic emerging industries and enterprises’ digital transformation,
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 944–953, 2023. https://doi.org/10.1007/978-3-031-36115-9_84
A Study on Learning Intention of Digital Marketing Micro Specialty Learners
945
the digital marketing micro specialty closely combines cutting-edge theories and practices such as the Internet, big data and artificial intelligence, with the goal of cultivating senior compound innovative talents who adapt to the development of new technologies, new industries, new business forms and have the basic theoretical knowledge of digital marketing, market analysis and the ability to solve practical problems in marketing. The concept of micro specialty was first proposed in 2013 by an educational organization co-founded by Harvard and Massachusetts Institute of Technology. By determining 5–10 core courses, micro specialty provides a targeted curriculum system for the cultivation of talents in corresponding majors or positions [2]. Guided by career and competency development, micro specialty focuses on the integration of teaching content and the practicality of training effectiveness. It is not only refined and targeted, but also emphasizes the interdisciplinary [3–5]. Regarding the existing research achievements of micro specialty, Zhu Jie et al. took the construction of new engineering as the background and explored the construction path of micro specialty of computer [6]. Zhang Zhiping et al. studied the construction of big data micro specialty from the aspects of specialty positioning and planning, faculty team and curriculum construction [7]. In the context of MOOCs, Wang Xiaomin et al. took rail transit signal and control major of Southwest Jiaotong University as an example to build a micro specialty course system [8]. From the perspective of curriculum developers, researchers have mainly discussed the cultivation patterns of micro specialty and the construction of curriculum systems, while there has been little research from the perspective of curriculum learners. In addition, most researchers focus on the theoretical deduction and induction level, and little present research results through data analysis. Therefore, from the perspective of course learners, this paper analyzes the learning intention of digital marketing micro specialty learners based on survey data from course learners, in order to provide some reference for the specific construction of micro specialty.
2 Theoretical Basis and Research Hypothesis 2.1 Theoretical Basis 2.1.1 Stimuli-Organism-Response Theory (S-O-R) S-O-R theory holds that environmental factors, as external stimuli, will affect individuals’ internal cognition and emotion, and then affect their behavioral responses [9]. The S-OR theory has been applied mainly to the study of consumer decision-making behavior in the field of economic management, and in recent years it has also been applied to the study of learner learning behavior in the field of education [10–12]. The S-O-R theory provides a crucial theoretical foundation for this research. 2.1.2 Educational Communication Theory Educational communication is an activity in which educators transmit knowledge, skills, and ideas to specific educational objects by selecting appropriate information content and using effective communication channels in accordance with certain requirements [13].
946
Y. Huang et al.
Berol’s communication model holds that the communication process should include receivers, information sources, channels and information, while the corresponding educational communication process should include learners, educators, instructional media and instructional content [14]. Therefore, the study of educational communication theory can help to identify the relevant factors or dimensions that affect the learning intention of digital marketing micro specialty. 2.1.3 Technology Acceptance Model (TAM) The TAM model, or the technology acceptance model, was first proposed by Davis et al., drawing on the intention model and planned behavior theory of social psychology [15]. This model is used to predict the propensity of subjects to accept, use, or reject new information technology. Perceived ease of use and perceived usefulness are core variables in the TAM model, through which external environment variables in information system affect individual behavior and decision intention [16]. This model has been widely used by educational researchers to study learning intention and behavior. For example, Vidanagama suggested that university students tend to focus more on perceived ease of use on e-learning and the students’ attitude has more influence on intention to use e-learning based on the modified TAM model [17]. 2.1.4 Self-efficacy Theory The self-efficacy theory was proposed by Bandura, who believed that self-efficacy refers to an individual’s subjective judgment on whether he can complete a certain task or behavior [18]. Self-efficacy has an impact on learners’ learning intention and learning outcomes. Nandang et al. found that computer self efficacy has a significant influence on computer anxiety, therefore it is expected to the university to be more often use the computer to provide lecture materials and assignments to students [19]. Liu Zhenyu et al. constructed a hypothetical model that self-efficacy influences learning outcomes based on the mediating effect of flow experience in the context of desktop virtual reality environment [20]. Lin Hongyi et al. found that self-efficacy can cultivate positive learning habits, enhance learning confidence and motivation, and thus reduce students’ learning burnout [21]. 2.2 Research Model and Hypothesis In this research, the S-O-R theory is used as the basic framework to combine educational communication theory, TAM model, and self-efficacy theory, and to use educators, instructional media, and instructional content in educational communication links as external stimulus variables for the learning intention of digital marketing micro specialty. The self-efficacy, perceived usefulness and perceived ease of use of the learners in the link of education communication are taken as the variables of psychological experience, and the learning intention of digital marketing micro specialty is taken as the variable of behavioral response, so as to study the influencing factors and influencing mechanism of the learning intention of digital marketing micro specialty. The model constructed in this research is shown in Fig. 1.
A Study on Learning Intention of Digital Marketing Micro Specialty Learners
947
Fig. 1. The study model of learning intention of digital marketing micro specialty learners
2.2.1 Stimulus Variables Stimulus variables are external environmental factors that influence the learning intention of digital marketing micro specialty, including educators, instructional media, and instructional content in the educational communication process. Due to the existence of educational tasks and training objectives, educators are in the position of leading, controlling and teaching in instructional activities and are indispensable subjects in micro specialty learning [22]. Instructional media refer to modern electronic media that can carry and transmit instructional information, including slide shows for offline learning, course platforms for online learning, etc. It facilitates the transfer of pedagogical information from the source to the learner and is an essential element in micro specialty learning. Instructional content is the primary message intentionally conveyed in the teaching process and is a critical prerequisite and conditional basis for the development of micro specialty learning. This research argues that external environmental stimuli have a direct or indirect effect on the learning experience and learning intention of digital marketing micro specialty. 2.2.2 Organism Variables As learners’ inner perception variables, organism variables include self-efficacy, perceived ease of use and perceived usefulness. In this research, perceived ease of use refers to learners’ comprehensive judgment on the difficulty of knowledge learning and online platform operation in the learning process of digital marketing micro specialty. Perceived usefulness refers to the comprehensive value judgment of learners that educators, instructional media and instructional content can effectively improve learning outcomes. Self-efficacy is the level of confidence a learner has in their ability to successfully complete a digital marketing micro specialty with the skills they have. This research believes that the analysis of learners’ internal experience is conducive to predicting their intention to learn digital marketing micro specialty.
948
Y. Huang et al.
2.2.3 Response Variables Response variables refer to the intention to learn a digital marketing micro specialty that a learner possesses after receiving an external stimulus and producing a corresponding mental state. Based on the above analysis, the following hypotheses are proposed in this research: H1/H2/H3: Educators significantly affect learners’ perceived ease of use/selfefficacy/perceived usefulness. H4/H5/H6: Instructional media significantly affect learners’ perceived ease of use/self-efficacy/perceived usefulness. H7/H8/H9: Instructional content significantly affects learners’ perceived ease of use/self-efficacy/perceived usefulness. H10: Perceived ease of use significantly affects perceived usefulness. H11/H12: Perceived ease of use/perceived usefulness significantly affects selfefficacy. H13/H14/H15: Perceived ease of use/self-efficacy/perceived usefulness significantly affected the intention to learn.
3 Research Design 3.1 Research Object This research was conducted for universities in Wuhan area, using a combination of stratified sampling and simple random sampling method to distribute online questionnaires to undergraduates. A total of 480 questionnaires were distributed, and 416 valid questionnaires were collected, with an effective recovery rate of 86.7%. 3.2 Research Tools The questionnaire consists of seven main parts: educator, instructional media, instructional content, perceived ease of use, perceived usefulness, self-efficacy and learning intention. The items of the questionnaire are adapted from the maturity scale combined with the specific situation of digital marketing micro specialty. Likert’s 5-point scale was used on all scales, with 1 to 5 indicating “strongly disagree” and “strongly agree” respectively. SPSS 26.0 was used to test the reliability of the data, and the Cronbach’s α coefficient values of seven variables were respectively 0.702, 0.777, 0.776, 0.809, 0.830, 0.898, 0.873, and the overall Cronbach’s α coefficient was 0.927, indicating good internal consistency of the questionnaire. The KMO value was 0.936 > 0.7, and the significance level was 0.000 < 0.05, indicating that all variables had good structural validity. AMOS 17.0 was used for the confirmation factor analysis, and the metrics for each fit scale were obtained to an acceptable level, with good overall fit results.
A Study on Learning Intention of Digital Marketing Micro Specialty Learners
949
4 Data Analysis 4.1 Common Method Deviation Test and Correlation Analysis Bias in common approaches is controlled programmatically by anonymous measurements and random permutations of latent variables. At the same time, Harman’s singlefactor test is used to detect common methodological biases in the collected data. The results of unrotated exploratory factor analysis showed that the maximum factor variance interpretation rate was 33%, lower than the judgment standard of 40% proposed by Harman [23], indicating that the common method deviation of the data in this study was within the acceptable range. Pearson correlation coefficients were used to find that there were significant correlations among all variables shown in Table 1, suggesting that external environmental stimuli may have a large impact on learners’ internal learning experience and learning intention. Based on this conclusion, the structural equation model can be further tested. Table 1. Structures and sources of the scale Educ
IM
IC
PU
PEU
SE
Educ
1
IM
0.67**
1
IC
0.72**
0.67**
1
PU
0.56**
0.56**
0.57**
1
PEU
0.34**
0.34**
0.42**
0.62**
1
SE
0.46**
0.42**
0.50**
0.59**
0.72**
1
LI
0.69**
0.60**
0.68**
0.62**
0.54**
0.65**
LI
1
Note: Educ: Educator; IM: Instructional media; IC: Instructional content; PU: Perceived usefulness; PEU: Perceived ease of use; SE: Self-efficacy; LI: Learning intention; ** indicates a significant correlation at 0.01 level;
4.2 Structural Equation Test This research tested the hypotheses of each path of the theoretical model by AMOS 17.0. The parameter test results show that in the model, educators → perceived ease of use (β = 0.083, P = 0.246), educators → perceived usefulness (β = 0.071, P = 0.228), instructional media → self-efficacy (β = 0.015, P = 0.339) and other paths do not meet the significance criteria of the parametric test. That is, H1, H3, H5, H7, H8, and H12 are not valid. After deleting the above paths and modifying the model, the final validated model is shown in Fig. 2, and the test results of main fit indexes are shown in Table 2.
950
Y. Huang et al.
Fig. 2. Research model and path coefficients
Table 2. The modified model fit Fitting index
χ 2 /df
Required 1–3 value Actual value
2.295
RMR
SRMR
RMSEA GFI
AGFI
TLI
CFI
< 0.05
< 0.05
< 0.08
> 0.9
> 0.9
> 0.9
0.035
0.041
0.056
> 0.9 0.923
0.908
0.954
0.905
4.3 Mediating Effect Test The mediating effect analysis of variables was shown in Table 3 and it was found that self-efficacy, perceived ease of use, and perceived usefulness have significant mediating effects between external learning environment stimuli and learning intention. Table 3. Results of mediating effect test (Bootstrap = 5000) Mediating path
Coefficient
Bootstrap standard error
Bootstrap confidence interval
Significance
Educ → SE → LI
0.192
0.026
[0.142,0.244]
***
IM → PEU → LI
0.235
0.046
[0.148,0.329]
***
IM → PU → LI
0.168
0.024
[0.086,0.176]
***
IC → PU → LI
0.202
0.043
[0.122,0.287]
*** (continued)
A Study on Learning Intention of Digital Marketing Micro Specialty Learners
951
Table 3. (continued) Mediating path
Coefficient
Bootstrap standard error
Bootstrap confidence interval
Significance
IM → PEU → PU → LI
0.044
0.015
[0.019,0.076]
***
IM → PEU → SE → LI
0.090
0.020
[0.053,0.131]
***
Note:Educ: Educator; IM: Instructional media; IC: Instructional content; PU: Perceived usefulness; PEU: Perceived ease of use; SE: Self-efficacy; LI: Learning intention; *** indicates a significant correlation at 0.001 level (bilateral)
5 Conclusions and Suggestions Based on the above analysis, the following conclusions and suggestions are drawn: Firstly, perceived ease of use, perceived usefulness and self-efficacy significantly affect the learning intention of digital marketing micro specialty. Meanwhile, perceived ease of use significantly affects perceived usefulness and self-efficacy. These results indicate that perceived ease of use plays a key role in enhancing learning intention to learn digital marketing micro specialty by promoting perceived usefulness and selfefficacy, and is an influential source of motivation for the generation and development of learning intention. Therefore, curriculum providers should take learner experience as the center, focus on the “zone of proximal development” of learners, and carry out the instructional content step by step, from the shallow to the deep. The selection or development of online learning platforms should also pay more attention to simplicity and convenience, so as to maximize the perceived ease of use for learners. Secondly, the role of the educator significantly affects the learner’s self-efficacy and ultimately their intention to learn. Educators positively influence learners’ selfefficacy, that is, by providing instructional support, educators help to create a harmonious learning atmosphere, promote learners’ confidence in completing learning activities and achieving learning goals, improve their internal motivation and sense of competence in learning, and ultimately generate positive intention to learn. Therefore, educators should give learners all-round and continuous attention in cognitive support, technical support and emotional support, so as to enhance learners’ intention to learn digital marketing micro specialty. Thirdly, the instructional content significantly affects the perceived usefulness, and ultimately affects the learners’ intention to learn the digital marketing micro specialty. Perceived usefulness reflects the performance that learners believe learning digital marketing micro specialty will bring to their own development. The analysis results show that the learner’s expectation of the instructional content lies in its practical value. Therefore, the design and implementation of digital marketing micro specialty for instructional content should constantly focus on the trends of learners’ needs so that learners can feel its prominent role in improving learning outcomes. Fourthly, instructional media have a positive impact on perceived ease of use and perceived usefulness, and further affect the intention to learn. In addition to directly affecting
952
Y. Huang et al.
perceived ease of use and perceived usefulness, instructional media also indirectly affect self-efficacy and perceived usefulness through perceived ease of use. Instructional media are the enlargement and extension of the senses and the brain of the learner, and play an indispensable role in knowledge teaching, skill training and intellectual development. Therefore, course providers can enhance learners’ perceived ease of use by continuously developing existing instructional media resources, simplifying and optimizing the operation process, so as to improve learners’ learning confidence and perceived benefits, and finally realize the enhancement of learners’ learning intention. Acknowledgment. This project is supported by Research on the Construction of Teaching Team of Marketing Specialties in the Context of “Double First-Class” (w2019037) and Exploration of the Cultivation System of Excellence Talents in “Digital Marketing” Micro Specialty in the Background of New Liberal Arts (Teaching Research Project of Wuhan University of Technology).
References 1. Junzong, Z.: The explaining of the new liberal arts from the four dimensions. J. Northwest Normal Univ. (Soc. Sci.) 56(5), 13–17 (2019). (in Chinese) 2. Pasad, R., Traynor, C., Alabina, A.: Engaged IT experience course to enable the future workforce. In: Proceedings of the 18th Annual Conference on Information Technology Education, pp. 7–12 (2017) 3. Long, C., Nan, W., Lili, F.: A study and exploration on the talent training model of “Microspecialty” in local universities: taking D University as an example. J. Shijiazhuang Univ. 23(02), 152–155 (2021). (in Chinese) 4. Yanyang, W., Yuan, Z., Yongming, W., Yangfeng, P.: Construction of micro specialty curriculum system in chemical engineering. High. Educ. Chem. Eng. 38(05), 11–16+51 (2021). (in Chinese) 5. Xiaodong, T., Chengbo, Y., Linlin, D.: Research on interdisciplinary integration mode based on micro specialty. J. Hunan Post Telecomm. College 20(04), 108–110 (2021). (in Chinese) 6. Jie, Z., Haiping, H.: Construction of micro-credential in Chinese universities under the background of new engineering. Softw. Guide 18(11), 172–175+179 (2019). (in Chinese) 7. Zhiping, Z., Xiaoxiao, L.: A preliminary study on the construction of big data micro specialty. Comput. Era 08, 68–70 (2019). (in Chinese) 8. Xiaomin, W., Wudong, Y., Qian, W., Yang, Y.: The construction of micro specialty curriculum system of rail traffic signal and control. Educ. Teach. Forum 05, 35–37 (2019). (in Chinese) 9. Lazaeua, R.S.: Emotion and Adaption, pp. 212–215. Oxford University Press, New York (1991) 10. Wanxin, P., Qishen, L., Haiying, J.: Research on the influence of college students’ willingness to learn under the online education model. Shanxi Youth 17, 13–14 (2021). (in Chinese) 11. Xuesong, Z., Minjuan, W., Ghani U.: The SOR (stimulus -organism- response) paradigm in online learning: an empirical study of students’ knowledge hiding perceptions. Interact. Learn. Environ. 28(5), 586–601 (2020) 12. Huajun, W., Wenshuang, G., Juhou, H.: Research on the effect of teacher support on willingness to continue learning in MOOC courses. Mod. Distance Educ. 03, 89–96 (2020). (in Chinese) 13. Guonong, N., Yunlin, L.: Educational Communication. Higher Education Press, Beijing (2005). (in Chinese)
A Study on Learning Intention of Digital Marketing Micro Specialty Learners
953
14. Daqing, Z.: The construction of “Four-in-one” integrity education mode for college students based on Berlo’s propagation mode. J. Jilin Agric. Sci. Technol. Univ. 28(01), 18–21+116 (2019). (in Chinese) 15. Soepriyatna, Pangaribuan, C.H.: The direct and indirect influence of gamification on learning engagement: the importance of learning goal orientation (a preliminary study). Int. J. Inf. Eng. Electron. Bus. (IJEEB) 14(04), 39–46 (2022) 16. Meihao, S.: A study on college students’ willingness to use online learning behaviour in the post-epidemic era - based on technology acceptance model. Agric. Henan 30, 32–34 (2021). (in Chinese) 17. Vidanagama D U. Acceptance of e-learning among undergraduates of computing degrees in Sri Lanka. Int. J. Mod. Educ. Comput. Sci. (IJMECS), 8(04), 25–32 (2016) 18. Guanru, H., Xin, L.: Exploring the effects of self-efficacy on entrepreneurial intentions. Shanghai Manage. Sci. 44(5), 74–79 (2022). (in Chinese) 19. Nandang, R., Budiman.: The influence of computer attitude, grade point average and computer self-efficacy against computer anxiety. Int. J. Educ. Manage. Eng. (IJEME) 9(05), 10–17 (2019) 20. Zheyu, L., Yujing, L., Jihui, Z.: Research on the influence of self-efficacy on learning outcomes in desktop virtual reality environment: the mediating based on flow experience. J. Dist. Educ. 40(04), 55–64 (2022). (in Chinese) 21. Hongyi, L., Fengyan, W.: The effect of discrimination perception on academic burnout among part-time graduate students: time management disposition and academic self-efficacy as chain mediator. China J. Health Psychol., 1–10 (2022). (in Chinese) 22. Le, Z.: On the status and function of educator and educate in vocational education. Hum. Resource Dev. 16, 46–47 (2018). (in Chinese) 23. Podsakoff, P.M., MacKenzie, S.B., Lee, J.Y., et al.: Common method biases in behavioural research: a critical review of the literature and recommended remedies. J. Appl. Psychol. 88(5), 879–903 (2003)
Comparisons of Western and Chinese Textbooks for Advanced Electronic Packaging Materials Li Liu1 , Guanchao Yin1(B) , Jin Wen1 , Qilai Zhou1,2 , Yao Ding1 , and Liqiang Mai1(B) 1 School of Materials Science and Engineering, Wuhan University of Technology,
Wuhan 430070, China {guanchao.yin,mlq518}@whut.edu.cn 2 Faculty of Science, Shizuoka University, Shizuoka 422-8529, Japan
Abstract. Textbooks are the importance carrier of teachers’ teaching and students’ learning process. Recently, the traditional textbook forms cannot fully meet these needs, which requires more researches. This study chosen western and Chinese textbooks on the field of advanced electronic packaging materials to fully analyze and compare in the aspects of format structure, knowledge system design, knowledge point introduction and information technology application. From the general comparison results, this Chinese textbook only surpasses the western textbooks in the online course construction of information technology application. Unfortunately, there are several drawbacks that needs to be improved for the future textbook constructions, which are smaller numbers of references, pictures and tables, relatively old references, too much background knowledge and narrative content, less analysis of case studies, insufficient width and depth compared to the western textbook in this study. Through this comparison study between western and Chinese textbooks, the improvement directions of Chinese textbooks for materials and related majors were pointed out, which includes format structure, knowledge system design, knowledge point introduction and information technology application. Keywords: Electronic Packaging Materials · Textbooks · Comparison
1 Introduction Textbooks are the carrier of knowledge, the medium for students to learn knowledge and the foothold of teaching content and curriculum system [1, 2]. As an important medium for imparting subject knowledge, university textbooks need to not only highlight the basic knowledge of the discipline and reflect the cutting-edge knowledge in related fields, but also need to have a clear understanding of the teaching rules. However, the organization and compilation of textbooks require enormous knowledge including discipline professional theory, engineering application knowledge and modern education theory with Chinese characteristics and world-class advanced teaching concepts [3–5]. Thus, the researches on the teaching textbooks have always been a hot topic among the educators. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 954–962, 2023. https://doi.org/10.1007/978-3-031-36115-9_85
Comparisons of Western and Chinese Textbooks
955
Especially for the training of novel engineering talents facing modernization in China, it is important to combine the development direction of China’s new economic form with the requirements of international and future-oriented talents from an international perspective. Moreover, establishing the requirements for the training of outstanding talents in engineering that adapt to the industrial transformation and upgrading of China’s traditional engineering industry and the development of strategic emerging industries is also of vital importance. However, the knowledge system of a single field is difficult to meet the needs of diversified knowledge structure of talents. For instance, the simple theoretical knowledge is difficult to fulfill the complex ability structure, while the traditional textbook forms cannot fully meet the needs of the new ecology of digital learning [6–8]. Therefore, the construction of teaching textbooks for engineering majors should closely follow the development of engineering sciences and related disciplines, especially for combining the new trends of the industry and the characteristics of China’s engineering majors to build teaching textbooks that reflect the international and future-oriented for new engineering talent training. One the one hand, the traditional textbook design mainly focuses on teacher teaching skills and knowledge system constructions. For cultivating high-level innovative talents, it is required to start from the student-centered and ability-oriented educational concept. Textbooks should provide space for students to take the initiative to build knowledge and the cultivation of students’ comprehensive ability [9, 10]. On the other hand, with the application of modern education technology and information technology, the application of teaching methods such as heuristic teaching and research teaching has given birth to the construction of new forms of teaching materials. Therefore, it is necessary to absorb and learn from foreign advanced experience to enhance the construction and form of textbooks. In this study, western (Materials for Advanced Packaging) and Chinese textbooks (Advanced Packaging Materials for Integrated Circuits) on advanced electronic packaging materials were fully compared in the aspects of format structure, knowledge system design, knowledge point introduction and information technology application to show the improvement directions for Chinese textbooks.
2 Descriptions of the Chosen Textbooks In this study, the selected western textbook was with two edition versions that was written by Daniel Liu and C.P. Wong and published by Springer publisher in 2009 and 2017, respectively. This book is one of the few important works on electronic packaging materials. The authors of the original book are all authoritative experts in the field of electronic packaging. From the content point of view, this book not only includes the latest insights of internationally renowned scholars on packaging materials, including wire bonding materials, lead-free solder, substrate materials, flip chip underfills, epoxy molding compounds, conductive adhesives, thermal interface materials, nano-encapsulation materials, etc. It also covers the latest developments in electronic packaging technology, including three-dimensional integration, system packaging (silicon thinning, hole filling), nano-packaging and interconnect, wafer-scale packaging, MEMS packaging, LED packaging and other frontier
956
L. Liu et al.
fields. This book provides a large number of references, providing readers with comprehensive background information for reference to relevant research at home and abroad [11, 12]. In contrast, the Chinese textbook chosen in this study is written by Dr. Qian Wang and published by Publishing House of Electronics Industry in 2021. This book is one of the integrated circuit series of academic books. Recently, the overall scale of China’s integrated circuit industry is experiencing unprecedented rapid development. However, due to the relatively weak foundation and some restrictions in foreign industries and technologies, China lags the international advanced level in most product subdivisions in all aspects of advanced electronic packaging materials, which still has very high foreign dependence. Table 1 lists the general details of above western and Chinese textbooks for comparisons. In the following chapters, the comparisons of and will be fully discussed from the aspects of format structure, knowledge system design, knowledge point introduction and information technology application. Table 1. Comparisons and details of Western and Chinese Textbooks Name
Materials for Advanced Materials for Advanced Advanced Packaging Packaging Edition 1 Packaging Edition 2 Materials for Integrated Circuits
Language
English
English
Chinese
Publication year
2009
2017
2021
Publishers
Springer
Springer
Publishing House of Electronics Industry
Author Number
44
57
5
Page Number
719
969
286
Chapter Number 19
22
13
References
1267
1789
289
Ref Year
Within 10 years
Within 10 years
Within 15 years
Figure Number
475
679
179
Table Number
97
113
44
3 Comparisons Between Western and Chinese Textbooks 3.1 Format Structure By comparing the above two examples of Chinese and western electronic packaging material textbooks in the numbers of authors, chapters, references, pictures and tables listed in Table 1, the differences of format structure between Chinese and foreign textbooks can be summarized as follows.
Comparisons of Western and Chinese Textbooks
957
It is noted that the author number of western textbooks is 57, which is far more than it of Chinese textbook (5 people). This is because that each chapters of this western textbook were written by experts in each specialized field, resulting in enormous people participating in the compilation. Moreover, the knowledge background and depth of this western textbook are wide. Since electronic packaging materials are emerging cross-courses, the current knowledge system has not been fully built. Thus, there are no exercises in both Chinese and English textbooks as some theories were not been clarified totally. Moreover, the shortcomings of domestic textbooks are also obvious. The numbers of references, pictures and tables are much smaller than it of western textbooks, which indicates that this Chinese textbook lag western textbook in terms of vividness. For example, there are chapters with fully plain text without figures and tables in this Chinese textbook. The narrative of the text with charts will make the teaching process of knowledge points more vivid to enable students learn professional knowledge more clearly. In addition, the gap in the reference number is also enormous. As reference can help readers expand their knowledge points during learning. Students would explore more about the knowledge they are interested in. Therefore, the new references can improve readers’ enthusiasm for independent learning, which effectively completes the teaching goals. 3.2 Knowledge System Design As shown in Fig. 1, the composition of different knowledge including background knowledge, general knowledge, application knowledge and frontier knowledge in western and Chinese textbooks is provided. The background knowledge includes introduction to industry background knowledge, pre-professional knowledge, etc. Basic knowledge means common knowledge and difficult points in professional knowledge. The definition of application knowledge is the application examples of basic knowledge, engineering application knowledge, industry application prospects and so on. Frontier knowledge includes cutting-edge knowledge of scientific research in emerging technology fields.
Fig. 1. The proportion of different types of knowledge in western and Chinese textbooks
958
L. Liu et al.
In Fig. 1, the obvious difference between the western and Chinese textbooks for advanced electronic packaging materials lies in the aspects of basic knowledge, application and frontier knowledge. The length of background knowledge for Western and Chinese textbooks is quite similar. In terms of basic knowledge and extended knowledge, the proportion of application knowledge and frontier knowledge of western textbook is higher than its general knowledge. In contrast, domestic textbooks account for a higher proportion of basic knowledge than application and frontier knowledge. Therefore, the focus of domestic and foreign textbooks is different. This Chinese textbook describes the basic knowledge more extensively and meticulously, so that students can learn and master the professional knowledge of related knowledge areas faster. However, for expanding knowledge, the content of domestic textbooks is relatively lacking due to the less and shallow descriptions of new technologies, frontiers, applications and corresponding extensions compared with the foreign electronic packaging material textbook. As a result, it can be concluded that this Chinese textbook can be significantly improved by introductions of expanding knowledge to further broaden students’ horizons.
Fig. 2. The proportion of narrative content and case study in western and Chinese textbooks
Moreover, the proportion of narrative content and case study in western and Chinese textbooks is listed in Fig. 2. The narrative includes textual explanations of important terms, formulas, charts and so on. Case study is the analysis of knowledge point expansion and the illustration of applied knowledge by cases. As shown in Fig. 2, the proportion of narrative content and case analysis in Chinese textbook is relatively even, while it is biased towards case analysis in foreign textbooks, which is quite similar to the conclusion of Fig. 1. The elaboration of basic knowledge points in domestic textbooks is detailed and solid, which take students to steadily learn step by step due to more proportion in narrative content. However, it is still slightly inferior to this foreign textbook for the expansion of knowledge points. In similar, there are more case studies in this foreign textbook, which also highlights the characteristics of
Comparisons of Western and Chinese Textbooks
959
foreign textbooks to enable readers to analyze and explore together. Therefore, Chinese textbook can add more case studies in the new edition to improve the interests of students. 3.3 Knowledge Point Introduction Through carefully reading through < Materials for Advanced Packaging (English)> and textbooks, the comparative summary of Chinese and western textbooks in knowledge point introduction can be listed in Table 2. Since the authors of each chapter in foreign textbook are independent, they all have their own introduction when introducing knowledge points. As a result, each chapter of foreign textbook is basically around a theme to describe from many aspects. Without the limitation to a fixed direction, a wider range content can be involved with more freedom. As for the Chinese textbook, it is basically introduced to students according to the three aspects of application, characteristics and development of electronic packaging materials, which are very clear and standardized at a glance. Table 2. Comparison of Chinese and western textbooks in knowledge point introduction. Western textbook
Chinese textbook
Structure
Chapters are written in different styles as they are written by varying authors
Similar structure and writing styles
Depth
More case studies to enhance students’ independent learning skills
Insufficient depth
Breadth
High proportion of frontier field
Insufficient breadth
3.4 Information Technology Application Recently, due to the worldwide epidemic impacts, offline teaching across the world, especially for China, has become more and more difficult. To ensure that teaching work is not affected, growing offline classes are shifting to online teaching. After several years of online teaching exploration, the online teaching methods of various colleges and universities are also gradually improving and exploring growing suitable online teaching methods [13–15]. The combination of online and offline teaching can improve the implement teaching. However, it is quite important to require teachers and universities to constantly explore and improve how to combine online and offline teaching to achieve the best teaching effect [16–18]. The combination of paper textbooks and electronic materials can bring out the best teaching results. Therefore, in addition to paper textbooks, it is necessary to make a richer teaching resource library and more comprehensive teaching resources, such as electronic materials and teaching videos corresponding to paper textbooks. Students can gain more knowledge from the rich resource base and carry out deeper learning
960
L. Liu et al.
by themselves. As a result, teachers’ teaching efficiency can be increased. For example, Beijing Institute of Technology designed a virtual simulation experiment of gold wire welding ball. Harbin Institute of Technology opened an English online course on microelectronic manufacturing technology. Moreover, the University of Maryland in the United States also offers online courses related to electronic packaging. In terms of information technology applications in textbooks, universities from Chinese and foreign countries have not jointly established a digital resource library and quality evaluation system that can be shared. When college students want to find information, they not only have to face the problem that textbook resources cannot be shared between universities, but also must judge the quality of data by themselves, which greatly reduces students’ learning efficiency. At present, the digital educational resources owned by colleges and universities are mainly based on text and video. Most students just enter the video interface and fast forward or drag the progress bar to complete the viewing of the video, which makes the educational resources greatly wasted. If colleges and universities want to conduct online teaching, establishment of a rich digital education resource base can be a good option. The rich educational resource library can not only allow teachers to have more abundant teaching methods but also enable students to learn independently and efficiently. Colleges and universities should accelerate the construction of digital resources and develop various platforms to facilitate the use of teachers and students. In addition, some universities can also cooperate to jointly build a digital resource library, share teaching resources, and greatly enrich teaching content (Table 3). Table 3. Comparison of Chinese and western textbooks in information technology application Western textbook
Chinese textbook
Electronic textbook
Can be downloaded online
Basically accompanied by electronic textbooks but with fewer obtain ways
Website/QR code
Has specific links through publish house website
No
New forms
The construction is late than the Chinese textbook
Online course websites
4 Effects of Textbook Improvement By comparing the differences between Chinese and foreign materials textbooks, this study provides guiding suggestions for optimizing Chinese textbooks, which promotes the improvement of innovative talent training, high-level curriculum construction, and teacher training skills. Based on the results, Wuhan University of Technology has carried out practice and reform in the following aspects. Firstly, for the main course of , its textbook is updating to the new version, which integrates the latest application knowledge and case studies. Moreover, the modern science and technology are combined with latest theory can not only broadens students’ knowledge, but also stimulates students’ creativity. Besides, all teaching courses in Wuhan University of Technology now require to complete the construction of online courses before their start. Therefore, with the advantages of online platforms, the construction of three-dimensional teaching materials can be promoted, improving students’ self-learning efficiency and the teaching interaction between teachers and students.
5 Conclusion The main purpose of this study was to compare the western and Chinese textbooks on topic of advance electronic packaging materials, which provides a direction for Chinese textbooks to improve afterwards. The conclusions can be drawn as follows. 1) In the aspect of format structure, although Chinese textbooks is concise, clear and logical due to small authors, the English textbook has rich information, knowledge and distinct level. 2) For the design of knowledge system, this Chinese textbook can be significantly improved by bringing more expanding knowledge to further broaden students’ horizons. 3) As for knowledge point introduction, the Chinese textbook shows its insufficient in depth and breadth of textbooks, which can be enhanced by adding more case studies in various aspects. 4) The Chinese textbook surpasses the western textbooks in the online course construction of information technology application. Acknowledgment. This work is supported by the Teaching Reform Project of Wuhan University of Technology (Grant No. w2021063).
References 1. Vojíˇr, K., Rusek, M.: Science education textbook research trends: a systematic literature review. Int. J. Sci. Educ. 41(11), 161354 (2019) 2. Sikorová, Z., Bagoly-Simó, P.: Textbook as a medium: impulses from media studies for research on teaching materials and textbooks in educational sciences. In: International Conference on Textbooks and Educational Media: Perspectives from Subject Education, 1–22 (2015) 3. Rottensteiner, S.: Structure, function and readability of new textbooks in relation to comprehension. Procedia Soc. Beh. Sci. 2, 3892–3898 (2010) 4. Marsono, W.M.: Design a digital multimedia interactive book for industrial metrology measurement learning. Int. J. Mod. Educ. Manag. Eng. 8(5), 39–46 (2016) 5. Fan, L.: Textbook research as scientific research: towards a common ground on issues and methods of research on mathematics textbooks. ZDM 45, 765–777 (2013)
962
L. Liu et al.
6. Scott, K., Morris, A., Marais, B.: Medical student use of digital learning resources. Clin. Teach. 15(1), 1–86 (2018) 7. Shafiee, N.S.M., Mutalib, S.: Prediction of mental health problems among higher education student using machine learning. Int. J. Educ. Manage. Eng. 10(6), 1–9 (2020) 8. Adebayo, E.O., Ayorinde, I.T.: Efficacy of assistive technology for improved teaching and learning in computer science. Int. J. Educ. Manage. Eng. 12(5), 9–17 (2022) 9. Sievert, H., Ham, A.K.V.D., Heinze, A.: The role of textbook quality in first graders’ ability to solve quantitative comparisons: a multilevel analysis. ZDM Math. Educ. (53): 1417–1431 (2021) 10. Hadar, L.L.: Opportunities to learn: mathematics textbooks and student’ achievements. Stud. Educ. Eval. 55, 153–166 (2017) 11. Lu, D., Wong, C.P.: Materials for Advanced Packaging. Springer, New York (2009). https:// doi.org/10.1007/978-0-387-78219-5 12. Lu, D., Wong, C.P.: Materials for Advanced Packaging, 2nd edn. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-45098-8 13. Hilton, J.: Open educational resources and college textbook choice: a review of research on efficacy and perceptions. Educ. Tech. Res. Dev. 64, 573–590 (2016) 14. Berry, T., Cook, L., Hill, N., Stevens, K.: An exploratory analysis of textbook usage and study habits: misperceptions and barriers to success. Coll. Teach. 59(1), 31–39 (2010) 15. Bowen, W.G., Chingos, M.M., Lack, K.A., Nygren, T.I.: Interactive learning online at public universities: evidence from a six-campus randomized trial. J. Policy Anal. Manage. 33(1), 94–111 (2014) 16. Krämer, B.J., Neugebauer, J., Magenheim, J., Huppertz, H.: New ways of learning: comparing the effectiveness of interactive online media in distance education with the European textbook tradition. Br. J. Edu. Technol. 46(5), 965–971 (2015) 17. Almazova, N., Andreeva, S., Khalyapina, L.: The integration of online and offline education in the system of students’ preparation for global academic mobility. In: International Conference on Digital Transformation and Global Society, pp. 162–174 (2018) 18. Cheung, S.K.S, Lee, K.K.W., Chan, K.K.L.: A review on the development of an online platform for open textbooks. In: International Conference on Hybrid Learning and Continuing Education, pp. 196–207 (2014)
Innovation and Entrepreneurship Teaching Design in Application-Oriented Undergraduate Professional Courses – Taking the Transportation Enterprise Management Course as an Example Yan Chen1 , Liping Chen2(B) , Hongbao Chen3 , Jinming Chen4 , and Haifeng Yang5 1 College of Traffic and Transportation, Nanning University, Nanning 530200, China 2 Pubei Middle School, Qinzhou 535300, China
[email protected]
3 School of Architecture and Engineering, Guangxi City Vocational University,
Chongzuo 532100, China 4 School of Electronic Information and Automation, Civil Aviation University of China,
Tianjin 300300, China 5 Faculty of Letters, Arts and Sciences, Waseda University, Shinjuku, Japan
Abstract. Integrating innovation and entrepreneurship teaching into applicationoriented undergraduate professional courses can help college students connect with career development and cultivate application-oriented professional and technical talents for the development of local economy. By analyzing the current development situation of innovation and entrepreneurship education for college students, analyzing the students’ learning situation, and analyzing the role and significance of transportation enterprise management as the core course of transportation specialty in cultivating application-oriented talents with integrating innovation and entrepreneurship education, this paper proposes to build a link between the teaching content of the course and the work content of the vocational post with innovation and entrepreneurship as a bridge, and excavate the innovation and entrepreneurship elements in the transportation enterprise management course. According to the work content of the management post of the transportation enterprise and the teaching methods such as team-based learning and task-driven approach, the innovation and entrepreneurship teaching design is carried out. Keywords: Applied undergraduate · Professional courses · Innovation and entrepreneurship teaching · Transportation enterprise management
1 Introduction Promoting innovation and entrepreneurship education in university education and cultivating college students’ innovation and entrepreneurship ability can help college students better connect with career development and promote employment. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 963–973, 2023. https://doi.org/10.1007/978-3-031-36115-9_86
964
Y. Chen et al.
1.1 The Development Status of Innovation and Entrepreneurship Education for College Students The proposal and promotion of innovation and entrepreneurship education for college students has developed for more than 20 years. In 1998, the United Nations Educational, Scientific and Cultural Organization (UNESCO) clearly proposed that college students should be trained to be excellent college students with entrepreneurial thinking, entrepreneurship and entrepreneurial skills. After graduation, they can not only find jobs, but also create jobs themselves. Driven by this educational concept, innovation and entrepreneurship education for college students at home and abroad has shown different characteristics through experiments and reforms. American colleges and universities take the concept of innovation and entrepreneurship as an indispensable ideological concept for infrastructure construction, set innovation and entrepreneurship education as a professional research direction, and formulated a complete set of class system and teaching plan for characteristic innovation and entrepreneurship education. According to 2016 statistics, the entrepreneurship rate of American college graduates has reached nearly 25%. The innovation and entrepreneurship education in Japanese universities has gone through the stage from utilitarian education to mass education. It is led by the government, assisted by universities and society, and trained from high school students, with strong coherence and regional characteristics. Innovation and entrepreneurship education in Britain is more about creating a cultural atmosphere. In China, there are three main ways for colleges and universities to carry out innovation and entrepreneurship education, namely, college students’ innovation and entrepreneurship courses, college students’ innovation and entrepreneurship training programs and college students’ innovation and entrepreneurship competitions, which can guide students to carry out career planning in combination with their majors and clarify their career development direction, which can play a role in enhancing students’ innovative thinking, practical ability, communication ability, etc., but only 1.5% of college students actually start businesses after graduation, There are few successful cases of entrepreneurship [1]. 1.2 The Relationship Between Application-Oriented Talent Training and Innovation and Entrepreneurship Teaching The cultivation of application-oriented talents is a comprehensive reform of the teaching model based on capacity output. The specific reform path is based on career orientation, industry-teaching integration, workshops, professional spirit, industry background, etc. The ultimate goal of innovation and entrepreneurship education is to build the professional ability of college students. The goals of the two are the same. However, in the college education system, at present, these two parts of teaching work are still two teaching systems, not well integrated, and there are bottlenecks such as lack of entrepreneurial experience of professional teachers, parallel innovation and entrepreneurship curriculum system and professional curriculum system, and lack of platform support [2–4]. In order to solve this problem, experts and scholars put forward relevant countermeasures. In addition to reforming the talent training mode, in the aspect of curriculum construction, they propose to build dual-qualified curriculum, specialized and innovative integrated curriculum with integrating innovation and entrepreneurship education into professional
Innovation and Entrepreneurship Teaching Design
965
construction based on the countermeasures of systematic and applied courses in the work process [5, 6]. 1.3 Problems to be Solved in this Paper The application-oriented curriculum is the carrier of cultivating application-oriented talents. Transportation enterprise management (hereinafter referred to as “the course”) is a basic course of transportation specialty, and also a practical course of profession and innovation & entrepreneurship integration reform supported by Nanning University. It is one of the key courses for the cultivation of application-oriented talents in transportation specialty. The course covers all aspects of enterprise operation and management, such as strategy, corporate culture, brand, market research, marketing, production operation, team management, customer management, human resource management, equipment management, quality management, financial management, et al., corresponding to relevant positions in production & business & administration & other departments of the enterprise, and is closely related to career development. The reform of the teaching content and teaching method of this course based on the idea of innovation and entrepreneurship can explore how to carry out the integration of specialty and entrepreneurship, cultivate students’ innovation and entrepreneurship ability in professional development based on the course, connect professional development and job requirements, lay the foundation for employment and entrepreneurship, and also provide reference for how to integrate professional courses into innovation and entrepreneurship education and teaching.
2 The Role and Significance of Integrating Innovation and Entrepreneurship Teaching into the Course of Transportation Enterprise Management Integrating innovation and entrepreneurship education into the course of transportation enterprise management not only enriches the breadth and depth of the course itself, promotes the development of the content of the course, but also guides students to pay attention to the industry and enterprises, and develops students’ professional abilities. It also cultivates application-oriented professionals for the society, and promotes local economic development, which has an important role and significance. 2.1 Analysis of Students The course is set up in the second semester of the third year of the college. Students have completed part of the professional courses, and also completed the school’s innovation and entrepreneurship courses. They have certain professional knowledge and entrepreneurship foundation. However, how to transform these knowledge into ability should be gradually transformed after docking with the industry. Before the beginning of the class, a survey was conducted on students in 2019. After analysis, it was found that students’ understanding of enterprise operation and management is superficial, and they have not yet understood the rules of enterprise economic
966
Y. Chen et al.
operation. In addition, some students have experience in innovation and entrepreneurship activities, but most of them have taken relevant courses such as entrepreneurship foundation, but still lack strong innovation and entrepreneurship awareness and clear future career and job development intentions. The survey data are shown in Table 1. Table 1. Statistics of Survey Data S/N Items
Attitude Affirm Negate
1
Have Taken Entrepreneurship Foundation and Other Related Courses 100%
0%
2
Participated in Innovation and Entrepreneurship Projects
30%
70%
3
Participated in Practical Activities inside and outside the School
70%
30%
4
Have a Strong Sense of Innovation and Entrepreneurship
30%
70%
5
Clear Future Career and Post Development Intention
10%
90%
6
Be Willing to Expand after Class
60%
40%
Therefore, it is necessary to integrate innovation and entrepreneurship teaching in the course of transportation enterprise management. First of all, it can promote students’ cognition and understanding of the industry and enterprises, and then it can guide students to clarify their career intentions and connect with career development. 2.2 The Promotion of Innovation and Entrepreneurship Teaching to Professional Courses Teaching The integration of innovation and entrepreneurship teaching can better clarify the ability goal orientation of professional curriculum teaching, and clearly cultivate the skills and qualities of students through career development. The cultivation of application-oriented professionals is mainly around employment. The ability requirements of applicationoriented talents are different from those of research-oriented and academic talents. Taking the transportation enterprise management course as an example, first of all, the course introduces the main line of “double entrepreneurship”. Whether it is employment or entrepreneurship, we should treat our work with the attitude of innovation and entrepreneurship, and the attitude determines our actions [7]; Secondly, strong innovation and entrepreneurship ability can make the work more successful [8]. Therefore, the ability training of the course will be carried out around the aspects of program design, publicity, communication and analysis, so as to clarify the ability and quality objectives of the course. Therefore, the integration of innovation and entrepreneurship education is a good promotion for the teaching of application-oriented professional courses. 2.3 The Course Meets the Needs of Enterprise Operation and Management The course of transportation enterprise management is a course that closely combines theory and practice. It connects with the posts of production, business, administration
Innovation and Entrepreneurship Teaching Design
967
and other departments of transportation enterprises. The teaching content of each unit is closely related to the work content of the post. In order to stimulate the students’ interest in learning, it is necessary to establish a link channel between them. Innovation and entrepreneurship teaching can just take on this channel task, taking the task/project as the carrier, Encourage students to combine theory with practice, pay attention to and analyze actively, deepen knowledge and transform ability by completing tasks/projects.
3 Innovation and Entrepreneurship Elements Contained in the Course To integrate innovation and entrepreneurship teaching into professional courses, it is necessary to excavate the innovation and entrepreneurship elements contained in the courses and integrate them into the teaching content and process of the courses, so as to achieve the purpose of integration of specialty and entrepreneurship. 3.1 Teaching Requirements for the Course The teaching objective of this course is to enable students to understand the enterprise management methods, processes and tools of passenger and freight transportation enterprises, transportation science and technology service enterprises, etc. in the field of domestic transportation comprehensively and systematically through the study of all the teaching contents specified in the syllabus, and master the strategic management, business analysis, customer management, brand marketing management, production and operation, business process The knowledge and methods of human resources, quality control, team management, material management, equipment management, information system and financial management will inspire innovative thinking and further strengthen the professional foundation and comprehensive ability of students majoring in transportation. 3.2 Main Teaching Contents of the Course The teaching content of the course mainly includes enterprise management theory, strategic management of transportation enterprises, business management of transportation enterprises, resource management of transportation enterprises and development management of transportation enterprises. 1) Enterprise management theory Combining the general enterprise management theory with the characteristics of transportation enterprises, and taking the five management functions of “planning, organization, command, coordination and control” of the process management theory of the process management school as the core, this paper analyzes the contents and methods of transportation enterprise management. 2) Strategic management of transportation enterprises Carry out enterprise strategic analysis of passenger and freight enterprises and transportation operation enterprises, list possible strategic options, make strategic choices and carry
968
Y. Chen et al.
out strategic implementation. Learn EFE matrix, IFE matrix, SWOT matrix and other strategic analysis tools, and learn the application of Porter’s competitive strategy selection model and Boston matrix’s diversification strategy selection model. Analyze the possible obstacles and possible measures for the implementation of enterprise strategy. Master the brand building process, brand equity model and brand management strategy of transportation enterprises. Use the “four levels” structure model of enterprise culture to construct the content level of transportation enterprise culture, master the implementation steps of enterprise culture, and refine the characteristics of transportation enterprise culture industry. Analyze the front and back service systems of passenger and freight transportation business. 3) Business management of transportation enterprises Use scientific methods to conduct transportation market survey and research, and find ways to solve the core business problems such as freight pricing, service satisfaction, transportation demand analysis, etc.; Understand the characteristics of production and operation of transportation enterprises, set production indicators, and conduct production and operation process management; Straighten out the passenger and freight transport business process, analyze the key points in the business process, optimize the process in combination with customer satisfaction, service efficiency, service quality, and other aspects, and use the evaluation index system to make a comparative analysis before and after optimization; Apply the “4P” theory model of enterprise marketing to establish the transportation marketing system, compile the marketing planning plan, control the marketing process, and evaluate the marketing work; Learn customer relationship hierarchy theory and three-level customer relationship marketing theory, establish appropriate customer relationships, understand what behaviors will affect customer loyalty, learn to use customer concentration index to analyze customer distribution, conduct targeted customer management, understand the key points of forming positive word-of-mouth communication and how to pay attention to customer word-of-mouth in the service process. 4) Resource management of transportation enterprises We need to use scientific management methods to effectively support business activities with human, financial, material, information and other resources in the enterprise. Specific learning contents include: talent selection and motivation, enterprise staffing method, Maslow’s demand hierarchy model, and two-factor motivation theory; The division of teams, site management, safety management and work responsibilities of team leaders, and how to select team leaders; Material management and production control JIT method, ABC inventory management method, ordering method, BOM list and other methods related to material management, MRP system cognition; Methods and management contents for selection, maintenance and renewal of transportation equipment; The functional structure, deployment, application and design of the transportation enterprise management information system, as well as the enterprise information construction;
Innovation and Entrepreneurship Teaching Design
969
Investment and financing management, current asset management and financial analysis of transportation enterprises, draw financial analysis index system, and recognize DuPont analysis method [9, 10]. 5) Transportation enterprise development management Learn service quality statistical analysis, service quality gap analysis, SERVQUAL service quality model, service quality management cost analysis, etc., and be able to scientifically formulate service quality evaluation index system; How to correctly formulate project objectives, decompose project tasks and manage progress; How to carry out innovative management in the operation process of transportation enterprises. 3.3 Analysis of Elements of Innovation and Entrepreneurship From the teaching objectives and contents of the course, it can be concluded that the innovation and entrepreneurship elements contained in the course mainly include the following categories. 1) Method class Including the management tools and methods used in the management process of transportation enterprises, which is the industry knowledge reserve for innovation and entrepreneurship. 2) Skills Including operational management skills such as scheme design, coordination and organization, leadership and command, process control and target management, as well as working skills for innovation and entrepreneurship. 3) Quality Including team communication, roadshow, analysis and judgment in the process of completing the project/task, which is a necessary quality for innovation and entrepreneurship. 4) Occupation Including corporate ethics, service awareness, concern for the environment and sustainable development, courage to endure hardship, self-discipline and other industrial corporate culture appeals, which are the industrial attributes of innovation and entrepreneurship.
4 Innovation and Entrepreneurship Teaching Design 4.1 Project Design After the students are divided into groups, each study group first selects a business direction, and after completing the study of management theory knowledge of each unit, take the innovation and entrepreneurship of “transportation enterprise management practice” as the main line, and use the knowledge learned to complete a task (project), which is jointly completed by the team. Each large class (80 min) is about 20 min for task practice. The project design is shown in Fig. 1.
970
Y. Chen et al.
Fig. 1. List of Innovation and Entrepreneurship Projects of the course
4.2 Comprehensive Practical Design By selecting a certain transportation business, from market research, strategic analysis and selection, to the formulation of marketing strategies, service process optimization and service quality control, an operation and management plan with distinct themes and detailed contents will be formed. In the process of completing the comprehensive practical task, the students have carried out an innovative thinking on enterprise operation and management and carried out a creative work [11]. 4.3 Teaching Design The teaching design is carried out by the combination of team-based learning and taskdriven method. The team-based learning requires to guide students to participate in the teaching process and play the interactive role between students, so as to help students acquire knowledge and skills [12]. Task-driven By setting up project development tasks
Innovation and Entrepreneurship Teaching Design
971
for comprehensive application of knowledge, effectively organizing teaching materials, promoting students’ autonomous learning in classroom teaching, and improving the application ability of knowledge learned [13]. Therefore, the combination of teambased learning and task-driven method to design the teaching organization process can well stimulate the creativity of students and cultivate the awareness of innovation and entrepreneurship. The relevant teaching process design is shown in Fig. 2.
Fig. 2. Teaching Process Organization
4.4 Assessment Design The course assessment is also integrated into the innovation and entrepreneurship teaching design. In terms of course assessment, three assessment units A, B and C are designed. Unit A is mainly for personal testing to assess students’ mastery of theoretical knowledge of enterprise management, integrate industry and enterprise knowledge into the assessment content, and increase students’ awareness and understanding of industry and enterprise; Unit B, mainly for team task assessment, takes innovative thinking and entrepreneurial practice as the main assessment basis, and evaluates students according to the completion of each class task practice, and integrates innovation and entrepreneurship teaching into professional teaching; Unit C is mainly for comprehensive assessment. Students are required to answer questions and complete analysis and design related to the course content within a limited time to evaluate the teaching effect of the integration of specialty and creativity [14]. The evaluation formula is as follows. S=
t α × (A + × B) + (1 − α) × C 2 T
(1)
In formula (1), S means the comprehensive evaluation score, A, B, C respectively refer to the assessment score of each unit in A, B, C, α refers to weight coefficient, usually 40% to 60%, t refers to individual learning engagement performance score, and T refers to the average score of learning performance of all members of the individual’s group.
972
Y. Chen et al.
For example, the weight coefficient set for the course is 50%, the assessment scores of each unit of A, B and C of a student are 90, 88 and 85 respectively, the individual learning performance score is 95 points, and the average group learning performance score is 90 points. According to formula (1), the student’s comprehensive assessment score is 88.72 points.
5 Conclusion 1) The paper analyzes and summarizes the current situation of the development of innovation and entrepreneurship education for college students, and points out that entrepreneurship is not the only purpose of innovation and entrepreneurship education, and promoting students’ professional development is the essence of innovation and entrepreneurship teaching. 2) The paper discusses the role and significance of integrating innovation and entrepreneurship education into the transportation enterprise management curriculum, and points out that professional courses and innovation and entrepreneurship education will promote each other, which is the carrier and channel for cultivating application-oriented professionals. 3) By analyzing the teaching content of the course of transportation enterprise management, the paper extracts the elements of innovation and entrepreneurship, and lays the foundation for the teaching design of innovation and entrepreneurship. 4) The thesis carries out a specific innovation and entrepreneurship teaching design in combination with the course of transportation enterprise management. Acknowledgment. This project is supported by The second batch of school-level teaching team in 2019 “Applied Effective Teaching Design Teaching Team” (2019XJJXTD10) and Nanning University’s second batch of teaching reform project of specialized innovation integration curriculum (2020XJZC04).
References 1. Qi, C.: Talking about the current situation of innovation and entrepreneurship education for college students at home and abroad. Modern Econ. Inf. 18, 394 (2016). (in Chinese) 2. Wang, C., Gong, C., Liang, Z.: The status quo and improvement measures of innovation and entrepreneurship education for college students in the new era. New West 04, 136–138 (2022). (in Chinese) 3. Wang, Y.: Analysis of the current situation of college students’ innovation and entrepreneurship education and research on countermeasures. Res. Pract. Innov. Entrepreneurship Theory 4(07), 182–184 (2021). (in Chinese) 4. Li, R., Chang, P.: Analysis and suggestions on the development of innovation and entrepreneurship education for college students. Inner Mongolia Sci. Technol. Econ. 01, 38–39 (2021). (in Chinese) 5. Cao, L., Yang, H.: Cultivation of artisan design professionals by industry masters as the main body of teaching guidance – exploration of the cultivation of application-oriented talents in design majors of Nanning University. Educ. Teach. Forum 33, 99–102 (2022). (in Chinese)
Innovation and Entrepreneurship Teaching Design
973
6. Ping, X.: Research on the status quo and countermeasures of innovation and entrepreneurship education for undergraduate students in application-oriented universities. Res. Pract. Innov. Entrepreneurship Theory 3(06), 74–75 (2020). (in Chinese) 7. Zheng, Y.: Research on innovation and entrepreneurship teaching system of engineering graphics course group. In: Proceedings of 4th International Conference on Social Science and Higher Education (ICSSHE 2018) (Advances in Social Science, Education and Humanities Research), (181), 492–494 (2018) 8. Yu, J.: Research on the construction of innovation and entrepreneurship teaching system based on computer science and technology. In: Proceedings of the 2018 8th International Conference on Management, Education and Information (MEICI 2018), pp. 927–931 (2018) 9. Liu, X.: Research on the application of humanistic management thought in modern enterprise management. Account. Corpor. Manage. 4(3), 87–94 (2022) 10. Aggarwal, A., Verma, R., Singh, A.: An efficient approach for resource allocations using hybrid scheduling and optimization in distributed system. Int. J. Educ. Manage. Eng. (IJEME) 8(3), 33–42 (2018) 11. Harijanto, B., Apriyani, M.E., Hamdana, E.N.: Design online learning system for kampus merdeka: a case study web programming course. Int. J. Educ. Manage. Eng. (IJEME) 11(6), 1–9 12. Zhao, H., Shen, Y.: Innovative exploration of peer learning to integrate international talent training in colleges and universities. Beijing Educ. (Higher Educ.) 10, 50–52 (2022). (in Chinese) 13. Ding, X.: Practice and exploration of task-driven flipped classroom in the course of “Web programming.” Sci. Technol. Wind 496(20), 121–124 (2022). (in Chinese) 14. Berrabah, F.Z., Belkacemi, C., Ghomari, L.Z.: Essential and new maintenance KPIs explained. Int. J. Educ. Manage. Eng. (IJEME), 12(6), 11–20 (2022)
The Construction of University Teachers’ Performance Management System Under the Background of Big Data Technology Fengcai Qin1 and Chun Jiang2(B) 1 Nanning University, Nanning 530299, Guangxi, China 2 College of Digital Economics, Nanning University, Nanning 530299, Guangxi, China
[email protected]
Abstract. Big data has been applied to student learning, teacher teaching, resource construction and university management. Teacher performance management is an important part of university management and an important means to improve the quality of education. However, there are many problems and defects in the performance management of university teachers. The performance evaluation of university teachers is mainly carried out by hand-filling and statistical summary. In the process of statistics, the audit of indexes is time-consuming and laborious, which is easy to produce data errors and low efficiency. Based on big data technology, through big data analysis to evaluate and optimize talent training, it is divided into three different important modules: the information-based performance evaluation of Talent Innovation Training, the information-based application performance evaluation of university scientific research capability and the information-based capability evaluation of university comprehensive services, and the establishment of data warehouse, to build an open network management system, to build an open network sharing platform, to strengthen the dynamic study of teaching management, to formulate various objectives and indicators of performance evaluation, to realize the flat, dynamic and integrity of performance evaluation management. Keywords: Data mining · Performance management · Big data
1 Introduction Globalization, competition, and reform calls offer several serious problems for highereducation institutions in terms of funding sources and allocations, teaching quality, and operational administration. It is vital to effectively analyze performances in order to maximize operational and management advantages [1]. The performance appraisal of university teachers is an important work of the school. Mastering the current situation of teachers, measuring teachers’ scientific research achievements and teaching workload, and supervising teachers’ teaching process are important means to reflect teachers’ teaching level and teaching quality [2]. Through the comprehensive evaluation of teachers’ education and teaching performance, scientific research performance, social service performance and teachers’ professional development, teachers can effectively guide and © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 974–982, 2023. https://doi.org/10.1007/978-3-031-36115-9_87
The Construction of University Teachers’ Performance Management System
975
stimulate the effect, and promote the improvement of teachers’ comprehensive quality and professional skills, which makes the ability of schools and teachers develop together [3]. Big data can collect a large number of different types of data, and seek the hidden relationship and value behind the data in deep mining and scientific analysis [4]. With the continuous development of information construction in universities, many universities have established and run various kinds of database systems, including all kinds of basic data needed for performance evaluation [5, 6]. How to effectively integrate the basic data and conduct in-depth analysis with the index system has become an urgent problem to be solved in the further deepening of university performance evaluation [7]. The design of evaluation index system, the collection of basic data and the construction of evaluation methods are the key factors of university performance evaluation [8]. The massive data in the information age provides a reference and basis for the development of education. Under the impact of the information age, the performance evaluation and information construction of education have become the driving force for the development of current higher education [9]. This study starts with the current situation and problems of University Teachers’ performance management, and uses the concept of big data for reference, analyzes the Enlightenment of big data thinking on University Teachers’ performance management, and puts forward the construction method of university performance management system based on data mining technology.
2 Problems of Performance Management in Colleges and Universities Under the Background of Big Data The fragmentation of data in the digital era is full of university teaching environments, and it also lays out the path for university teaching decision-making. Yet, how to mine the critical data required for university personnel training and management innovation using high-tech information data has become the emphasis and challenge of information technology usage in a big data environment. Teaching indicators are primarily used to assess instructors’ instructional burden. Computer technology is permeating university instructional administration. The use and practice of digital management significantly improves teaching quality. On the one hand, it is entirely reliant on information technology; on the other hand, it is continually supported by big data. A cloud computing system is a data information system that consists of virtualized network data storage, dispersed network data terminals, and integrated information technology. Its main advantage is that it can perform large-scale information data management as well as comprehensive evaluation and analysis on large talent training data models and university parameter indicators, allowing it to effectively build a virtual, high-scalability, high-speed, and highspeed cloud computing system. Reliable, low-cost information-integrated network-based performance evaluation system for teaching management. At present, the performance evaluation of university teachers is mainly carried out by manual filling and statistical summary. The statistics of basic data is cumbersome and complex, and the audit of indicators in the statistical process is time-consuming and laborious, which is easy to produce data errors and low efficiency. From the content of indicators, teaching indicators are mainly the workload of teachers’ teaching, and
976
F. Qin and C. Jiang
scientific research indicators are mainly the number of papers published, monographs, topics, etc. the determination of such indicators is too much emphasis on the results of indicators, lack of evaluation of specific work process. From the evaluation criteria of indicators, the quantity of some indicators cannot represent the quality of indicators, especially in academic indicators. Overemphasizing the quantity of annual achievements deviates from the essential characteristics of academic research. At the same time, to a certain extent, it will lead to increasingly strong internal friction and utilitarianism among teachers, seriously affecting the physical and mental health of teachers and the construction of university teachers Design. The university teaching management is complex, so whether it is performance evaluation or information transmission, it is always inseparable from the support of big data, especially the university work information transmission needs timeliness and dynamic, so as to carry out performance evaluation for the cultivation of educational technology talents. Figure 1 reflects the main structure of school public governance.
Fig. 1. The main structure of public governance
Although some colleges and universities have established information-based performance management platforms, due to the lack of teachers’ information literacy, the function setting of the management platform is not comprehensive enough, and it is difficult to fully play the role of the performance management platform, so the work efficiency has not improved greatly [10]. In practice, through the sharing of big data resources, data mining, analysis and statistics, the teaching management information in different links such as talent innovation, management innovation and performance evaluation is analyzed and quantified, and through the information docking between universities, the openness of the market and the effectiveness of information transmission are increased. University resource information and related teaching management information can be uploaded to the cloud platform through virtual servers to share data resources on campus and off campus, thus providing intellectual support for innovative training of talents.
The Construction of University Teachers’ Performance Management System
977
3 The Influence of Big Data on University Teachers’ Performance Evaluation Platform The quality of university teachers plays a vital role in the quality of education. Education level not only determines the level of training talents, but also relates to the improvement of national quality [11]. At present, the problems in the construction of university teachers. Schools need to formulate corresponding measures to maintain high-quality talents, stimulate teachers’ work enthusiasm, and improve teaching quality, scientific research level, social service level and their own professional development level [12]. From the content point of view, the index teaching indicators mainly include teachers’ teaching work, and the research measures mainly include the number of published papers, the number of books, and the problem of quantity. As a result, the university lacks the specific working process of evaluation, which cannot be reflected by the quality index. As time goes by, the accumulation of data increases the amount of basic data, and even establishes a big data system library for each college department, so as to more effectively quantitatively manage the performance evaluation of teachers and college departments. The purpose of performance management is to promote the development of teachers themselves. However, the most important thing in the university performance management system is to provide reference or play a role in the development of teachers, which leads to the form of performance evaluation. The main purpose of performance management is to promote the development of teachers, improve the efficiency of university management and promote the rapid development of universities. However, performance management in many universities does not help teachers’ development, and performance evaluation has become a superficial form.
4 Implementation of University Performance Evaluation Management Platform Based on Big Data 4.1 Architecture of University Performance Evaluation Management Platform The framework of a university educational technology talent training performance evaluation system should include information infrastructure construction, information application analysis, information resource utilization and sharing, and information organization structure management, all based on big data technology. Nevertheless, the informatization construction must always review and optimize the associated talent training mode through big data analysis in order to deploy resources wisely. The system’s business function has flaws of its own. When it comes to acquiring vendors, each system’s institutions are uneven. Third-party services are often offered by mature products rather than the school’s enhanced unified data standard, and some do not even provide data interfaces and table domains [13]. Of course, the development performance of the University’s educational technology talent training should always be consistent with the construction of informatization, focusing on the intelligent training of talents and the innovation of University’s scientific research strength, as well as the comprehensive strength research in the process of University’s talent training [14]. The performance assessment management platform based on big data analysis must link with the school’s more relevant
978
F. Qin and C. Jiang
business system in order to facilitate data sharing and sharing. The important connection is the successful integration of information and data resources. The relatively independent data statistics and analysis of business systems is a major barrier to big data, so we should integrate the information resource management system, particularly for existing data integration, to further accelerate quantitative evaluation based on big data analysis in university performance management. The university performance evaluation system is divided into three different important modules, the informatization performance evaluation of innovative cultivation of talents, the informatization application performance evaluation of university scientific research ability and the informatization ability evaluation of university comprehensive service. The specific process is shown in Fig. 2.
Fig. 2. Performance evaluation system
Because big data is open in exploring the internal laws of things, that is, it is not necessary to preset the conclusions in advance, and more unknown laws and multiple systems can be obtained by studying and analyzing the collected fine data. The performance evaluation analysis layer evaluates the performance of universities by applying big data mining algorithm, deeply analyzes the internal relationship of evaluation data. Performance evaluation and analysis layer can use data cloud storage and distributed management to integrate system architecture. In the age of big data, universities must integrate the information background development trend, promote the effectiveness and modernization of university teaching work through various technical means and methods, change the traditional teaching management practice mode through computers and data management systems, and realize the effective disclosure and sharing of data resources [15]. 4.2 Construction of University Performance Data Warehouse Therefore, the integration of information resources is a key step. The current university network system is uneven, and each subsystem is relatively independent, just an entrance to the employee account, which is lack of data sharing. In the long run, the implementation of quantitative assessment of teachers’ performance needs the accumulation and
The Construction of University Teachers’ Performance Management System
979
storage of basic data. With the further development of education informatization, big data has also been applied to university teaching management, but many universities are difficult to design big data independently, so the government needs to do a good job in the top-level design first, and then gradually implement the big data design of each university. The structural relationship between the independent perceived familiarity and the score of local samples and non-local samples is shown in Fig. 3.
Fig. 3. The relationship between local samples and non-local samples
University data integration layer is the basic data source of university performance evaluation, including heterogeneous data sources distributed in various systems. After establishing the starting mechanism of heterogeneous data sources, the data needed for
Fig. 4. Data mining process in financial analysis and management
980
F. Qin and C. Jiang
university performance evaluation can be incrementally integrated into the data warehouse to ensure the accuracy of basic data. The data mining process of performance management is shown in Fig. 4. Suppose the expected output is, and the global error between expected and actual output is defined as L: 1 2 zk − zk L= 2 m
(1)
k=1
Through the back propagation process, the error is expanded to the hidden layer as: t
2 m m 1 1 2 f (λk ) − zk = wjk yj + bk − zk L= f 2 2 k=1
k=1
(2)
k=1
Finally, the reverse transmission to the input layer is: ⎡ ⎛ ⎛ ⎞ ⎞ ⎤2 t
2 m m t n 1 1 ⎣ ⎝ ⎝ f f L= wjk yj + bk − zk = wjk wij ai + bj ⎠ + bk ⎠ − zk ⎦ 2 2 k=1
k=1
k=1
k=1
j=1
(3) The network error is a function of the weights wij and wjk. Therefore, the error E can be changed by changing the weight of the neuron, thus: wij = −ε
∂L (i = 1 . . . m, j = 1 . . . n) ∂wij
(4)
wjk = −ε
∂L (j = 1 . . . n, k = 1 . . . t) ∂wjk
(5)
where ε represents the rate. Teachers are the subject and object of performance management in universities. The main role is that teachers can participate in the formulation and implementation of performance management system, while the object role is that teachers are the managed objects. This dual identity makes teachers occupy a key position in the application of new technologies, so performance management based on big data puts forward requirements for teachers, and teachers must have informatization awareness, so as to stably carry out informatization performance management [16]. The cultivation of educational technology talents within universities requires a strong emphasis on scientific research. Therefore, it necessitates the integration of high-quality scientific research resources and the establishment of an open network management system to allow scientific researchers and experimental instruments to be shared on an open network platform, facilitating dynamic research on teaching management and developing various objectives and evaluation indicators for performance assessment, to realize the flattening, dynamic and integrity of performance evaluation management [17].
The Construction of University Teachers’ Performance Management System
981
5 Conclusion The performance evaluation and analysis system of university teachers needs to be continuously improved according to the practice results [18]. This paper introduces the importance of university performance management evaluation to university management, analyzes the impact of big data on university teachers’ performance management under the background of big data, and puts forward the step plan for building university teachers’ performance data warehouse, providing reference for future performance management of universities: 1) Collect basic data, including the performance evaluation of talent training informatization, the performance evaluation of university scientific research capacity informatization application and the evaluation of university comprehensive service informatization; 2) Integrate information resources and information of various business systems; 3) Gradually implement the top-level design and coordinate the informatization construction at the school level; 4) Enhance teachers’ information literacy, promote teachers to accept the performance management system and apply it in teaching and management. Acknowledgment. This project is supported by the 2022 private higher experts project of the 14th five-year plan of Guangxi educational science (2022ZJY3220) and the 2021 Professor Training project of Nanning University (2021JSGC15, 2021JSGC08).
References 1. Tsen, Y.-J.: An exploratory analysis in the construcion of college performance indices. Int. J. Organizat. Innov. 5(3), 98–132 (2013) 2. Yan, X.: University performance evaluation management platform based on big data. Microcomput. Appl. 033(012), 3–6 (2017) 3. PeiJiayin, X., Hanwen, L.Y.: Business curriculum reform and teaching practice based on OBE concept and data mining. Int. Public Relat. 106(10), 124–125 (2020) 4. Jian, S.: The design and implementation of hospital personnel salary management sysem. Electron. Technol. Software Eng. 175(05), 203–205 (2020) 5. Youxing, X., Huilin, H.: Evaluation of the business performance of listed companies in my country’s big data concept sector. Data Min. 007(001), 16–25 (2017) 6. Donghui, C.: Reform and innovation to promote performance management, and actual results are shown. Finan. Super. 484(22), 16–18 (2020) 7. Ma, L., Zhou, X.: Research on the evaluation mechanism of innovation and entrepreneurship team management based on data mining classification. Modern Electron. Technol. 042(011), 178–180+186 (2019) 8. Yi, Z.: Enterprise human resource performance management innovation in the era of big data. Shangxun 146(13): 100+102 (2018) 9. Dongying, C.: Research on the application of decision tree data mining technology in the performance evaluation of public utilities management. Digit. Des. 6(09), 265–266 (2017) 10. Amjad, M., Linda, N.J.: A web based automated tool for course teacher evaluation system (TTE). Int. J. Educ. Manage. Eng. (IJEME) 10(02), 11–19 (2020)
982
F. Qin and C. Jiang
11. Zhu, X., Zhu, Z.: An application of the wiles test in the formulation of education strategy for the newly-upgraded colleges of China. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 3(02), 15–21 (2011) 12. Dafeng, C., Haiyong, C.: Analysis of information system trend auditing under big data environment. Finan. Account. Monthly 861(17), 118–125 (2019) 13. Hongrun, G.: Research on human resource assessment management system based on data mining. Enterp. Reform Manage. 385(20), 57–59 (2020) 14. Zhao, L.: A preliminary study on enterprise human resource performance management in the big data era. China Manage. Inform. Technol. 22(08), 72–73 (2019) 15. Jingru, Z.: Research on enterprise human resource performance management innovation in the big data era. China Manage. Inf. Technol. 022(011), 70–72 (2019) 16. Xinxin, W.: Discussion on the construction of enterprise human resource performance management system in the era of big data. Times Finan. 786(32), 101–103 (2020) 17. Patil, M.M., Hiremath, B.N.: A systematic study of data wrangling. Int. J. Inf. Technol. Comput. Sci. (IJITCS) 10(01), 32–39 (2018) 18. Sharma, K., Marjit, U., Biswas, U.: PTSLGA: a provenance tracking system for linked data generating application. Int. J. Inf. Technol. Comput. Sci. (IJITCS) 7(04), 87–93 (2015)
Curriculum Evaluation Based on HEW Method Under the Guidance of OBE Concept Chen Chen1(B) and Simeng Fan2,3 1 School of Business Administration, Wuhan Business University, Wuhan 430056, China
[email protected]
2 Department of Education, Anyang University, Anyang 14028, South Korea 3 Department of Environmental Art and Design, Hubei Institute of Fine Arts,
Wuhan 430204, China
Abstract. OBE (Outcomes-Based Education) concept emphasizes learning achievement-oriented, and combining OBE concept to construct curriculum quality evaluation system is the basic activity of reverse design and forward implementation in OBE. The construction and evaluation of curriculum index system is a complex and systematic engineering. The HEW (Hierarchy Entropy Weight) Method evaluation method is designed by combining the hierarchical structure of AHP (Analytic Hierarchy Process) and EWM (Entropy Weight Method). A curriculum quality evaluation model suitable for the concept of OBE is constructed. Based on the three basic features of OBE concept, namely, reverse design, forward implementation and learning outcomes, an evaluation system with 27 indicators of 3 layers is constructed. EWM was used to convert experts’ opinions into evaluation weights of index. In the evaluation system, learning outcomes are critical index of curriculum evaluation, which includes goal achievement, achievement quality, and feedback. In the reverse design process, the level of teaching team, the rationality of teaching objectives and teaching organization are the most important concerns. In the forward implementation process, students’ attitude to learning is the most important factor. Under the student-centered concept, teacher needs to use various methods and means to stimulate students’ inner learning drive. Keywords: Curriculum evaluation · OBE · Entropy Weight Method
1 Introduction Curriculum evaluation is an important content to guarantee the quality of teaching implementation, which has been widely paid attention by teaching managers and teachers. Under the concept of OBE, the curriculum is based on learning results to complete reverse design and forward implementation, so it is more important to establish a scientific curriculum evaluation system. Redesigning the index system of curriculum quality evaluation according to the concept of OBE, and setting the scientific weight for evaluation index, thus carrying out objective evaluation of curriculum, is not only the basis of the reverse design of OBE, but also the key of the teaching design in the forward implementation process. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 983–994, 2023. https://doi.org/10.1007/978-3-031-36115-9_88
984
C. Chen and S. Fan
In the process of course evaluation, scientific and reasonable evaluation system is the most important thing. The evaluation system is like a commanding baton, which will affect the whole process of curriculum design and implementation under the concept of OBE. In the evaluation system, it is necessary to make subjective selection and judgment depending on the experts, but also to guarantee the objectivity of the system construction, so it is necessary to choose scientific methods to establish the curriculum evaluation system under the concept of OBE, which plays an important role in enriching the teaching theory and promoting the teaching practice. 1.1 OBE 1.1.1 Correlation Coefficient Derivation OBE was first put forward by Spady scholars in 1981 and explained as “the design of every activity in the education system should be based on the final learning result” [1]. OBE includes three core concepts: student-centered, reverse design and continuous improvement. In the modern information society with rapid change of talent demand, the role of curriculum is to create a formal knowledge about ideation and process of start-up, university curriculum invest more time to guide students, but the effect is not ideal [2]. Higher education needs to be more connected with social needs and training goals. Under such background, more and more educational scholars pay attention to the educational concept of OBE. The OBE philosophy emphasizes the focus on learners’ actual learning needs, clearly focus on expected learning outcomes, and focus on learning outcomes. Following the principle of reverse design and forward implementation, the course structure is constructed, the teaching is designed and implemented, and the dynamic adjustment is realized by evaluation, which ensures the achievement of expected learning results and the cultivation of students’ ability. His implementation consists of four steps: Defining - Realizing - Assessing - Using [3]. Assessing is a connecting step from the top to the bottom, from the reverse to the forward link, scholars have done a lot of research on curriculum evaluation. Osman discussed the perceptions of the undergraduate students from Civil and Structural Engineering Department who have undergone their industrial training program, in order to assurance to the effectiveness of OBE [4]. Alghamdi determine a learning strategy using an Outcomes-based Education approach for E-Business Strategies and redesign the curriculum using an Active Learning Approach [5]. Premalatha describes the framework of OBE and detailed survey on CO (Course Outcomes)-PO (Program Outcomes) mapping and its attainment models [6]. Under the background of OBE teaching concept, course assessment should pay attention to the core characteristics of OBE, combined with the teaching process of forward design and reverse implementation, design index of response evaluation system, and analyze the weight of the index to guide the process of continuous improvement. 1.1.2 EMW EMW is a theoretical mathematical method, which originates from thermodynamicsand is introduced by Shennong. It is a measure to describe the uncertainty of system states. From the point of view of information theory, information is a measure of the degree of
Curriculum Evaluation Based on HEW Method
985
order and entropy is a measure of the degree of disorder. According to this property, the smaller the information entropy, the smaller the degree of information disorder and the larger the value of useful information. The larger the index weight, the larger the entropy weight, and the opposite, the smaller the entropy weight. According to the viewpoint of information theory, the weight of the index is calculated according to the information amount carried by each index. The varying degree of each factor in the index depends on the information amount. The larger the variation degree of each factor, the larger the information amount carried by the index. That is, the greater the difference, the greater the weight of the counter. Conversely, the smaller the weight, the weight will contain the information itself characteristics, has a certain objectivity, such a calculation method applied in subjective evaluation method can make the evaluation have a certain objectivity. In the process of project evaluation and target decision-making, appraisers often consider the relative importance of evaluation indicators and reflect the weight of indicators. According to entropy, the quality of information obtained in decision-making process will directly affect the precision and reliability of decision-making. Entropy weight method is an objective value assignment method, based on information quality, the entropy weight of each index is calculated by information entropy, and then the weight of each index is calculated by entropy weight. Among all kinds of subjective evaluation methods for analyzing project scheme, entropy weight method has the characteristics of high precision, strong objectivity and good evaluation. It is widely used in many fields such as scheme selection [7], algorithm optimization [8], safety production [9]. 1.1.3 Curriculum Evaluation Hierarchical structure can help the decision-maker to make more reasonable evaluation under the framework of the hierarchy of specific problems, and make the evaluation problems of multiple indexes and multiple factors more systematic and scientific. The hierarchical index system is the choice of many scholars in the evaluation process [10–13]. At the same time, in the course quality evaluation, Liu chooses an innovative perspective that considers course evaluation as a multiple criteria decision-making problem, they proposed a model with AHP and FIS (fuzzy inference system) to measure the course effectiveness regarding various indicators [12]. Nielsen studied the relationship between teachers’ and students’ styles and course evaluation, and found that their styles, especially for complex content, are not compatible, which may cause teachers’ and students’ different evaluations of curriculum [14]. Irina Integrated Application of Multiple Criteria Decision Making (MCDM) and Analytical Hierarchy Process Fuzzy (AHPF) Method, design a new methodology for evaluating the quality of distance learning courses [15]. Maarten analyzed over 3000 courses taught in a large European university and found that the traditional use of student evaluations to evaluate courses may not be a good choice, they found that course evaluations is upward biased, and that correcting for selection bias has non-negligible effects on the average evaluation score and on the evaluationbased ranking of courses [16]. From the research of scholars, it is a complicated problem how to evaluate the curriculum from an effective angle, and different quality indicators
986
C. Chen and S. Fan
will analyze the curriculum quality from different perspectives such as the objective and personnel.
2 Hierarchy Entropy Weight Method Model 2.1 Hierarchy Entropy Weight Method Model Process The HEW model mainly consists of four steps. First of all, establish the hierarchical assessment model, and then introduce the entropy as the weight of the evaluation system when determining the weight of assessment criteria. The operation processes are shown in Fig. 1.
Fig. 1. HEWM model process
1) Considering the feasibility and systematicness of the evaluation index, the course assessment criteria are selected according to the essential characteristics of OBE. 2) Combined with the hierarchical model of AHP method, the indexes of assessment are hierarchically constructed. 3) Experts in the field are selected, the importance evaluation of the index weights is collected, and the index weights are determined by the information entropy of the evaluation, and the evaluation system is constructed. 4) The evaluation index system is used to evaluate the course. 2.2 Assessment Criteria for Curriculum Under OBE Concept The first step of constructing evaluation system is to select evaluation indicators. The existing mature evaluation structures such as CIPP (context, input, process, product) education evaluation indicators and hybrid learning evaluation cannot fully reflect the core characteristics of OBE, so we need to design a more targeted index system for OBE. The curriculum evaluation index for OBE should include the following three parts, reverse design, forward implementation evaluation and outcome and output evaluation.
Curriculum Evaluation Based on HEW Method
987
2.2.1 Reverse Design Index The core of the concept is to carry out the reverse design of the curriculum based on the objectives and expected learning outcomes. The whole environment and concept starting from teaching design must reflect the core of the reverse design based on learning outcomes. For example, objective of teaching design, teaching organization and arrangement, and learning resource design should support the curriculum design and implementation around OBE. 2.2.2 Forward Implementation Evaluation The forward implementation of hybrid curriculum should take teachers and students as the center, and carry out online and offline teaching activities according to the teaching scheme formulated in the reverse design stage, so as to realize the learning output. Therefore, we should pay attention not only to the process evaluation of students, but also to the performance of teaching process, such as teaching attitude, teaching activities and teaching means application. Teachers and students, as two main subjects in teaching activities, are bound to have interactive behavior. The essence of classroom teaching is the process of dialogue between teachers and students. Therefore, the frequency, time and depth of interaction between teachers and students in teaching should also be the important content of course design and evaluation. 2.2.3 Outcome and Output Evaluation The evaluation of learning output is based on the comprehensive evaluation of the expected and non-anticipated achievements of the course, and further analysis of the achievement of the curriculum objectives, so as to understand the learning effect of the curriculum comprehensively. It is used to reflect the development of knowledge, ability and comprehensive quality of students and the application of learning results in the curriculum of learning. Teachers reflect on the problems in the course design and implementation, and improve the curriculum continuously. 2.3 Hierarchical Assessment Criteria According to the evaluation index of OBE curriculum, combining systematicness and measurement, the index system is constructed according to the hierarchical structure, which includes three levels of index system, as shown in Table 1.
988
C. Chen and S. Fan Table 1. Assessment criteria for OBE curriculum
Level 1
Level 2
Level 3
F1 : Reverse design
S11 : Teaching Objectives
T111 : Feasibility T112 : Developmental T113 : High-order
S12 : Teaching programme
T121 : Design Concept T122 : Teaching Content T213 : Application of teaching methods T124 : Teaching Method
S13 : Learning Resources
T131 : Teaching Team T132 : Hardware Condition T133 : Software Resources
F2 : Forward implementation
S21 : Teaching Organization
T211 : Teaching Plan Execution T212 : Teaching Activity Organization T213 : Application of Teaching Methods T214 : Teaching Attitudes
S22 : Teaching atmosphere
T221 : Teacher-student Interaction T222 : Student–student Interaction
S23 : Learning Performance
T231 : Classroom Learning Attitude T232 : After-Class Time
F3 : Learning outcomes
S31 : Target achievement
T311 : Knowledge Acquisition T312 : Capability Development T313 : Value Improvement
S32 : Curriculum Satisfaction
T321 : Satisfaction with the Curriculum T322 : Satisfaction with Teachers T323 : Satisfaction with Self-improvement
S33 : Quality of results
T331 : Scientific T332 : Innovative T333 : completeness
2.4 EWM Process EWM combines the subjective evaluation of experts and the objective entropy method to determine the index weight. The main steps are as follows. Firstly, the ranking opinions of the expert group are collected, and the ranking data set (ai1 , ai2 , ..., ain ) of the expert group on the importance degreeof a certain indicator set is C = (c1 , c2 , ..., cn ) assumed to be a ranking matrix A = aij k×n of the expert group, where aij is an evaluation of the expert i on the indicator j. In order to eliminate the uncertainty of experts in the ranking process, the blindness analysis is carried out and the total cognition degree is calculated. The transformation entropy function is used as formula (1), where I is the importance ranking value given
Curriculum Evaluation Based on HEW Method
989
by the expert, m = j + 2 is the number of transformation parameters, and j is the actual maximum order number. ln(m − I ) (1) ln(m − 1) The values in the sorting matrix A = aij k×n are brought into the formula (1) to perform corresponding transformation, where the membership matrix B = bij k×n of the sorting matrix A is obtained. Assuming that each expert has the same discourse power, the average degree of µ(I ) =
k
bij
cognition of the index bj = k is assumed. Define the blindness of experts’ perceptions of indicators cj arising from uncertainty as formula (2) i=1
Qj =
[max(b1j , b2j , ..., bkj ) − bj ] + [bj − min(b1j , b2j , ..., bkj )] 2
(2)
The evaluation vectors X = (x1 , x2 , ..., xn ) of the experts on the indicators cj are obtained, where xj = bj 1 − Qj . Normalize the indicators to obtain the weight matrix W = (w1 , w2 , ..., wn ), where xj . wj = n xj
j=1
3 Curriculum Evaluation Under OBE Concept 3.1 Expert Selection Six teaching experts were invited to evaluate the importance of curriculum indicators. All of them had senior titles or positions and had been engaged in teaching or management for more than 15 years, including both teaching managers and first-line teachers, which ensured the comprehensiveness and objectivity of evaluation. If the number of indicators is too large, experts’ relative importance may be affected. According to the indicator system, the indicators at different levels are grouped into five groups: F1 –F3 (three indicators), S11 –S33 (nine indicators), and T111 –T133 (ten indicators). There are eight indexes for T211 to T232 and nine indexes for T311 to T333 . Experts are invited to rank the relative importance of the indicators in the group. If the indicators are not different, they can rank the indicators in the same order. 3.2 Weight of Assessment Criteria Base on EWM Collect the original evaluation and evaluation of experts and sort it out to form an importance ranking matrix. The importance of indicators is shown in the Table 2.
990
C. Chen and S. Fan Table 2. Assessment criteria weight and ranking table
Level 1
Level 2
F1 : Reverse S11 : Teaching design Objectives (Relative weight: (0.282) 0.330) S12 : Teaching programme (0.401)
S13 : Learning Resources (0.318)
F2 : Forward implementation (0.236)
S21 : Teaching Organization (0.456) (0.456)
S22 : Teaching atmosphere (0.270)
Level 3
relative weight
weighting
Ranking
T111 : Feasibility
0.460
0.0427
7
T112 : Developmental
0.307
0.0285
19
T113 : High-order
0.233
0.0217
26
T121 : Design Concept
0.253
0.0336
15
T122 : Teaching Content
0.258
0.0342
14
T213 : Application 0.306 of teaching methods
0.0406
11
T124 : Teaching Method
0.182
0.0241
22
T131 : Teaching Team
0.465
0.0488
3
T132 : Hardware Condition
0.304
0.0319
16
T133 : Software Resources
0.231
0.0242
21
T211 : Teaching Plan 0.276 Execution
0.0297
18
T212 : Teaching Activity Organization
0.362
0.0390
12
T213 : Application of Teaching Methods
0.360
0.0388
13
T214 : Teaching Attitudes
0.344
0.0220
25
T221 : Teacher-student Interaction
0.433
0.0277
20
T222 : Student–student Interaction
0.222
0.0142
27
(continued)
Curriculum Evaluation Based on HEW Method
991
Table 2. (continued) Level 1
F3 : Learning outcomes (0.433)
Level 2
Level 3
relative weight
weighting
Ranking
S23 : Learning Performance (0.274)
T231 : Classroom Learning Attitude
0.632
0.0408
10
T232 : After-Class Time
0.368
0.0238
23
S31 : Target achievement (0.478)
T311 : Knowledge Acquisition
0.374
0.0774
2
T312 : Capability Development
0.407
0.0845
1
T313 : Value Improvement
0.219
0.0454
5
T321 : Satisfaction 0.393 with the Curriculum
0.0463
4
T322 : Satisfaction with Teachers
0.254
0.0300
17
T323 : Satisfaction with Self-improvement
0.352
0.0415
8
S33 : Quality of T331 : Scientific 0.411 results T332 : Innovative 0.379 (0.250) T333 : completeness 0.210
0.0447
6
0.0412
9
0.0229
24
S32 : Curriculum Satisfaction (0.271)
3.3. Characteristics of Curriculum Evaluation Indexes Under OBE Concept
4 Examples of Curriculum Evaluation 4.1 Curriculum Evaluation The evaluation of the course is carried out in logistics management specialty of W university. W University was founded in 1985, is a local, applied undergraduate university in the middle of China. The W university has more than 10 thousand students, and the logistics management specialty, which is a first-class major in Hubei Province, with more than 500 students. The course of “Logistics System Analysis” was selected for evaluation. This course is the basic course of logistics management major. It is opened in the 5th semester, with 48 h and 3 credit points. In assessment, 2 teaching management personnel, 2 professional teachers and 5 students who have selected the course are invited. Each item in evaluation system, form T111 to T333 , is scored from 1 to 5 points, with a full score of 5. The arithmetic mean value is taken for the evaluation staff of the same type. The total score
992
C. Chen and S. Fan
of the three categories is 40% for teaching management staff, 40% for professional teachers and 20% for students respectively. The final evaluation score of Logistics System Analysis is 4.296, among which the average evaluation score of teaching managers is 4.317, the average evaluation score of professional teachers is 4.425 and the average evaluation score of students is 3.996. The course has been reformed by OBE, and the curriculum syllabus based on OBE concept has been developed, and the teaching is carried out according to the curriculum syllabus, and the quality evaluation of OBE concept has obtained a high score. At the same time, the evaluation result of the curriculum reform is excellent, which also proves the effectiveness of the evaluation result of the curriculum reform. 4.2 Characteristics of Curriculum Evaluation Under OBE Concept Learning results are the most important factor in evaluation, which means that the results are the core goal of OBE. The second important reverse design process and the lowest positive implementation process indicate that more energy should be invested in reverse design of courses under OBE, especially the design and presentation of learning results, which is the most important index to evaluate the quality of courses. It should be noted that in the learning outcomes part with the highest importance, the weight of each factor is increased, and the target achievement, including the knowledge attainment and ability development of students, are the most important factors affecting the curriculum quality, which fully reflects the core essence of OBE, which also indicates that teachers’ teaching is not only the completion of the whole teaching process, but also the significance and final achievement of the teaching activities. On the other hand, in the second important reverse design part, teaching team, teaching goal, teaching organization and teaching content are the key indicators that need to pay attention to, which indicates that the concept of OBE teaching design actually puts forward better requirements for the teaching team, and teachers need to fully understand its connotation from the teaching concept, and complete the deep-level transformation of teaching objective and teaching organization on this basis. At last, students’ attitude to learn, teacher’s activity organization and means application are the two most important indicators in the process of curriculum implementation. Student’s input is the most important part of the indicators, and teachers are more like the directors of teaching activities, so they should promote and stimulate independent learning ability of student, and use various means and methods to guide and promote their positive attitude to learn. It is worth noting that in the course of evaluation of Logistics System Analysis, students’ scores of T311 -T313 are obviously lower than those of teaching managers and professional teachers. Through the interviews with students, we know that students are actually quite satisfied with the overall course, but not satisfied with their own input, so they are not satisfied with the value enhancement and ability development they have gained. This dissatisfaction is due to their unsatisfactory to themselves. It can be seen that students are eager for self-improvement, which also promotes teachers to stimulate student’s self-dissatisfaction in the teaching process and turn them into positive learning attitude and behavior.
Curriculum Evaluation Based on HEW Method
993
5 Conclusion In order to better adapt to the scientific evaluation of courses under the concept of OBE education, a quality evaluation model of curriculum based on HEWM was constructed. Firstly, the evaluation Indicators were organized more clearly by using the hierarchical structure of AHP, and then experts were invited to perform the weight analysis on the indexes. The weights of the indexes are determined by EWM method, and the scientific evaluation of the curriculum is carried on. 1) An evaluation system based on the concept of OBE, which includes 27 specific indexes of 3 levels, is established. The index system reflects the core characteristics of OBE forward design, reverse implementation and emphasis on learning outcomes. 2) The HEW Method firstly establishes the hierarchical index content, and then converts the expert’s subjective opinion into objective entropy value by EWM in order to establish the scientific weight of indicators. 3) In the evaluation system, learning outcomes, as core of OBE concept, become critical content of evaluation, from the side of the side, it shows that evaluation indicators embody the core of OBE concept. 4) In the reverse design process of curriculum construction, teachers as the main designers, the level of teaching team, the rationality of teaching objectives and the organization of teaching is the most important index, while in the forward implementation process, the cooperation between students and teachers is emphasized, among which students’ attitude to learning is the most important factor, and teachers need to use various methods to stimulate intrinsic learning drive of student. Acknowledgments. This project is supported by the Ministry of Education Collaborative Education Project “Cultivation Effect of Logistics Management Professionals Based on OBE Concept” (220905522223325).
References 1. Spady William, G.: Outcome-based education: critical issues and answers. Arlington: American Association of School Administrators, p. 212 (1994) 2. Wiradinata, T., Antonio, T.: The role of curriculum and incubator towards new venture creation in information technology. Int. J. Educ. Manage. Eng. (IJEME) 9(5), 39–49 (2019) 3. Acharya, C.: Outcome-based education (OBE): a new paradigm for learning. Centre Dev. Teach. Learn. 7(3), 7–9 (2003) 4. Osman, S.A., Zaidi, M., Mat, K., et al.: Outcome based education (OBE) curriculum assessment for industrial training program: based on students’ perception. New mark. Res. J. 12(6), 454–463 (2009) 5. Alghamdi, T., Jamjoom, A.: Developing E-business strategies curriculum case study in the information systems department. Int. J, Mod. Educ. Comput. Sci. (IJMECS) 4(2), 1–7 (2012) 6. Premalatha, K.: Course and Program Outcomes assessment methods in outcome-based education: a review. J. Educ. 199(3), 111–127 (2019) 7. Bruno, M.S., Leoni, P.G., Lucila, M.S.C.: Performance evaluation of green suppliers using entropy-TOPSIS-F. J. Clean. Prod. 207(10), 498–507 (2019)
994
C. Chen and S. Fan
8. Susan, S., Ranjan, R., Taluja, U., Rai, S., Agarwal, P.: Neural net optimization by weightentropy monitoring. In: Verma, N., Ghosh, A. (eds.) ICCOML: International Conference on Computational Intelligence, pp. 201–213. Springer, Singapore (2019). https://doi.org/10. 1007/978-981-13-1135-2_16 9. Xiangxin, L., Kongsen, W., Liwen, L., Jing, X., Hongrui, Y., Chengyao, G.: Application of the entropy weight and TOPSIS method in safety evaluation of coal mines. Procedia Eng. 26, 2085–2096 (2011) 10. Asli, L., Gokturk, S.: Quality evaluation of graduates. Soc. Behav. Sci. 70(25), 1009–1015 (2013) 11. Jianting, S., Chunyou, D., Jiuyang, H.: Research and design of teaching evaluation system based on fuzzy model. Int. J. Educ. Manage. Eng. 2(10), 45–51 (2012) 12. Yan, L., Xin, Z.: Evaluating the undergraduate course based on a fuzzy AHP-FIS model. Int. J. Modern Educ. Comput. Sci. (IJMECS) 6(12), 55–66 (2020) 13. Suartini, N.K.Y., Divayana, D.G.H., Dewi, L.J.E.: Comparison analysis of AHP-SAW, AHPWP, AHP-TOPSIS methods in private tutor selection. Int. J. Modern Educ. Comput. Sci. (IJMECS) 1(15), 28–45 (2023) 14. Nielsen, T., Kreiner, S.: Course evaluation for the purpose of development: What can learning styles contribute? Stud. Educ. Eval. 54(9), 58–70 (2017) 15. Irina, V., Romualdas, K.: Methodology for evaluating the quality of distance learning courses in consecutive stages. Soc. Beh. Sci. 191(7), 1583–1589 (2015) 16. Maarten, G., Anna, S.: Measuring teaching quality in higher education: assessing selection bias in course evaluations. Res. Higher Educ. 58(7), 341–364 (2017)
The Relevance of a Systematic Approach to the Use of Information Technologies in the Educational Process Nataliya Mutovkina1(B) and Olga Smirnova2 1 Department of Management and Social Communications, Tver State Technical University,
Tver 170012, Russia [email protected] 2 Institute of Economics and Management, Tver State University, Tver 170021, Russia
Abstract. The object of research in this work is modern information technologies and software, which is advisable to use for solving different tasks in the educational process. The authors focus on the positive aspects of the use of information technologies and consider them an integral part of the educational process. Digital competencies gained by students in the process of computer-based learning are necessary for their professional activities. The better students’ information technology skills are developed, the more competitive they will be in the labor market. The article presents an overview of information technologies and software that are successfully used in teaching students. The paper provides examples of the use of modern information technologies to solve the problems of information retrieval, data analysis, and visualization. It is established that information technologies and software should be applied, considering the continuity of the educational material, as well as the interdisciplinary nature of the educational process. The novelty of the research lies in the proposal of a systematic approach to the use of information technologies in the educational process. Under the systematic approach, information technologies should ensure not only the continuity of educational material but also the continuity of each other. Keywords: Educational process · Information technology · Software · System approach · Analysis tasks · Visualization · Digitalization
1 Introduction Using information technologies and computer software in the educational process is an integral part of the digitalization of the education system. Some 15 − 20 years ago, the educational process in most Russian educational organizations was implemented only through classroom communication between teachers and students. The major tools of the teacher were a wooden board and chalk, and students made notes of lecture material and solved problems in notebooks. Now traditional teaching methods are gradually being replaced by interactive, remote methods, and on-line technologies. This does not mean that the blackboards and notebooks have sunk into oblivion. But now students are © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 995–1005, 2023. https://doi.org/10.1007/978-3-031-36115-9_89
996
N. Mutovkina and O. Smirnova
increasingly replacing ordinary paper notebooks with laptops or tablets. Some students even take notes on smartphones. Interactive whiteboards, computers, projectors, and other technical devices came to the teachers’ aid. Each teacher can broadcast educational material using presentations displayed via a laptop and a projector on a special screen. Over the past five years, distance learning has gained undeniable importance. Information technologies implemented in various software environments are actively used by students to solve educational tasks in various directions. Information technologies are simply irreplaceable for processing large amounts of data, their analysis and visualization, in solving optimization and forecasting problems. The results of solving educational tasks are also drawn up by students as reports and presentations, and then they are accepted for defense. Thus, modern information technologies in the educational process are used not only as teaching tools, and organization of the educational process but also as tools used by students to solve educational problems. However, many questions arise in this area. For example, choosing software products for solving educational tasks issues of continuity of information technologies in the educational process; the need to consider the interdisciplinary nature of educational material when using information technologies, and others. The primary task that the authors of this study solved can be planned: what should the complex information technologies used in the educational process be in order to ensure a high level of learning efficiency? The existing limitations should be considered: the cost of software; the complexity of mastering individual software products; the time required for training to work in separate software environments. The purpose of the work is to study the possibilities of a systematic approach to the use of modern information technologies and software in the educational process. 1.1 Literature Review The relevance of the use of information technologies in the educational process is because of the social need to improve the quality of education, as well as the practical need to use modern computer programs. The main purpose of using information technologies in the educational process is, first, to strengthen the intellectual capabilities of students in the information society, as well as individualization and improvement of the quality of education. This is stated in the works of many scientists, for example, in [1−5] and others. In the article [6], the authors point out the main positive effects of the use of information technologies in the educational process. They include high-speed processing of large amounts of information, extensive visualization capabilities, multiple reproducibilities of educational material, time savings, and increased learning efficiency. Some researchers are considering the possibilities of information technology for teaching a specific course, for example, mathematics [7]. The paper [8] considers the role of information technologies in creating business incubators and innovative enterprises based on educational organizations. In Indonesia, as in many other countries, information technology is used as a tool to activate and promote entrepreneurship and contribute to a smooth transition from training to real business.
The Relevance of a Systematic Approach
997
In the economic sphere, the creation of digital companies is of increasing interest [9]. But the creation and operation of a digital company is impossible without a workforce with digital competencies. In some works, the authors talk about the experience of using certain information technologies and software in the educational process. So, the article [4] tells about the InetRetriever tool, which students can use to get any necessary information from the Internet in real time. Students can also use InetRetriever to implement, execute and test their projects using real data. The article [1] demonstrates the technology of using the DirectumRX software product in the educational process of studying the disciplines “Content Management in an organization” and “Document management in an organization”. Software products such as Tableau [10], Microsoft Excel, STATISTICA [11], programs for building a decision tree: Deductor, See5, WizWhy [12], GIS technologies [13] and other software can be used for analysis and forecasting. There are choosing software tools. However, acceptable alternatives may be significantly limited. 1.2 Directions of Digitalization of the Educational Process The main directions for digitalization of the educational process are the creation of an interactive learning environment in each educational organization with free access to scientific libraries and databases; introducing modern information technologies and software into the educational process to solve educational tasks that turn into professional tasks; the creation of digital educational organizations in which students will receive an education without leaving them at home. Information technologies make it possible to implement such teaching methods as modeling and simulation work [6]. These methods are becoming increasingly common in the formulation and solution of educational tasks. For example, future economists solve tasks ranging from optimization and management of the company’s activities to analysis of the country’s foreign economic activity. Information technologies are used already in the first year of training. And here, a systematic approach is important. With this approach, the teacher ceases to be a translator of information, but becomes an intermediary between the student and the means of information technology, instilling skills in independent work. The analysis of thematic publications and the results of the experience of using information technologies in educational organizations has shown that the greatest efficiency of the educational process can be got through the systematic use of information technologies in an optimally selected combination. When forming an information technology system in the educational process, it is necessary to consider its specifics, direction and planned learning outcomes, the level of the material and technical base of the educational organization and the level of training of the teaching staff.
998
N. Mutovkina and O. Smirnova
2 A Systematic Approach to the Use of Information Technologies in the Educational Process 2.1 The Essence of the System Approach Using a systematic approach in teaching involves the relationship between the components of the educational process, each of which can function with maximum efficiency, relying on internal connections in the educational system. The content of the studied material is one of the structural components of the educational process, the development of which is associated with the selected methods, forms, and means of teaching. The teacher manages this system, also being its component. The effectiveness of the educational system depends on which technologies the teacher will use [14]. A systematic approach to the use of information technologies in the learning process should not be replaced by inefficient “digitized” pedagogical practice, when a teacher withdraws from the educational process, for example, by transferring an electronic version of his lectures to students for independent study. Here, the student closes himself in his computerized environment, to the detriment of interactive and group forms of work, which reduces the effectiveness and the strategy of educational activity does not change. A systematic approach that meets the needs of the information society assumes that university teachers and students become active “subjects of higher education” at the same time. The teacher acts as the subject of the organization of the educational process at the university, and the student is the subject of the educational process. In forming universal and professional competencies, a student in modern conditions becomes not only an object of training and education but also an initiative subject of educational activity. It is assumed that the manifestation of purposeful activity in personal formation and self-realization through a set of actions, and implementation ensures the ascent to the next stage of professional development [2]. 2.2 Modern Information Technologies in the Educational Process According to the system approach, information technologies in the educational process can be defined as a set of methods, hardware, and software for collecting, organizing, storing, processing, transmitting, and presenting information that expands students’ knowledge and develops their capabilities to manage socio-economic, technical and other processes in the future. Each information technology can be considered as a system comprising the elements shown in Fig. 1.
The Relevance of a Systematic Approach
999
Fig. 1. Components of information educational technology
The hardware environment is a set of technical means used (computer, projector, peripherals, and so on). The software environment includes basic and application software. The subject environment is the content of a specific subject area, science, technology, and knowledge. The methodological environment is formed by the instructions for the use of the elements listed above. Currently, the following information technologies can be successfully applied in the educational process. 2.2.1 Working with Web Resources Students are invited to analyze Web resources on the themes offered by the teacher and those resources that the students themselves will find. At the beginning of the lesson, the teacher briefly talks about the problem and the possibilities of its solution, and makes a statement about the problem. Then students, using specific Web resources, carry out a thematic search, analysis and processing of information under the task. At the end of the lesson, students summarize the results got from several Web sources, and evaluate the effectiveness of Web resources. Such classes contribute to the development of students’ critical thinking. A variety of Web resources that can be used in the educational process are virtual excursions. Students can visit thematic exhibitions, museums, and research institutes without leaving the classroom, and get acquainted with the objects of their study at no costs. On-line boards can be used for students to work in groups, for brainstorming, and for students to develop joint solutions. The digital approach to brainstorming allows you to better organize a discussion, discuss problems, and record and visualize intermediate and final results. Web resources for organizing surveys allow the teacher to find out the opinions of students regarding the organization of the educational process, the availability of presentation of educational material. To organize on-line surveys, you can use Socrative, Google Forms, SurveyMonkey, Poll Everywhere and several other services. Based on the results he got, the teacher can react quickly and improve the quality of classes.
1000
N. Mutovkina and O. Smirnova
To keep students up to date with upcoming events, it is advisable to create an on-line calendar for the group. This calendar will mark the most important events, upcoming events of the student group, topics of upcoming classes, tasks of the group, and individual students. This disciplines students, allows them to always be aware of the events of the group. To create an on-line calendar, such programs as Google Calendar, Yandex. Calendar and others are suitable. The on-line calendar is also convenient for teachers, as it allows them to correlate planned events with their capabilities and adjust the plan. The changes made immediately become known to all other participants in the educational process. 2.2.2 Creating Presentations with Multimedia Elements Lectures are conducted with the help of presentations of educational material. Presentations can be developed in software environments, for example, Microsoft PowerPoint, Photo Show PRO, Apple Keynote, Kingsoft Presentation, Impress, Hippani Animator, Google Slides, Piktochart, Zoho Show and others. The primary requirement for the software used is its accessibility for all participants in the educational process. The presentation file should be opened both on the teacher’s computer and on the students’ computers. To effectively hold the attention of students and activate their interest in the topic, ordinary presentations comprising text and some diagrams no longer help. Presentations containing various media elements that can be turned on at the right moment are much more useful. These include images, podcast clips, pictograms, interactive graphics, sound effects, websites and others. Adding high-quality media objects makes lectures more attractive. The same applies to presentations of student reports prepared as part of practical classes or scientific conferences. 2.2.3 Educational Video This information technology refers more to passive learning tools. However, it is very effective for mastering educational material if it is discussed at the end of watching the video. There are extensive collections of educational videos on the Internet. Many of them are well structured, and have an internal search, which allows you to quickly find the video. Well-chosen educational videos can positively affect the development of a student’s competencies, memory, critical thinking, and ability to solve problems. 2.2.4 Conference Communication Technologies These information technologies have become relevant to the COVID-19 pandemic. With the weakening of the pandemic, interest in them has not disappeared but even intensified. The educational process has gone beyond the classroom. Now teachers and students often use services such as Microsoft Teams, Skype, Zoom. Besides conducting classes within the curriculum, videoconferences can be organized to communicate with experienced teachers, and experts from other countries,
The Relevance of a Systematic Approach
1001
to conduct joint work with students from other educational organizations. Such communication allows students to get acquainted with new ideas, the opinion of eyewitnesses. It is advisable to invite students to prepare questions on the videoconference in advance. This way they will feel like full-fledged participants of the video meeting and not passive viewers. 2.2.5 Virtual Simulators Virtual simulators are indispensable for training specialists in medical, engineering, and natural science fields. However, more and more opportunities are opening up for their use in the training of economists [15], sociologists, psychologists, and teachers. For example, the TeachLive virtual simulator, developed at the University of Central Florida, allows you to virtually immerse yourself in the reality of teaching a lesson, form practical skills related to lesson organization, and managerial decision-making in various pedagogical situations [16]. 2.2.6 Use of Students’ Mobile Devices Smartphones and tablets are now available to every student. It’s no secret that many teachers see a serious problem because students are often distracted by these devices during classes. Indeed, this interferes with the conduct of classes, the development of educational material. However, students’ addiction to mobile devices can be turned to good by delivering educational content via mobile devices. It is necessary to get used to this reality and use mobile devices for educational purposes. There are several educational platforms using students’ mobile devices. These include Edap, Skill Pill, Bridge, Kitaboo Insight, SoloLearn, Quizlet and other information resources. 2.2.7 Information Technologies of Knowledge Assessment Testing systems are mainly used to assess students’ knowledge, which are embedded in the electronic information and educational environment of the university, for example, based on Moodle. Their major advantage is fast, convenient, impartial and automated processing of students’ answers. But there is also a drawback, comprising the fact that the answer system does not allow the student to show his creative abilities. Interactive knowledge monitoring systems, for example, PROClass, Mentimeter, act as an alternative to such test technologies [17].
3 Application of Information Technologies in the Course “Economic Theory” The systematic application of these technologies, software and hardware makes it possible to form an integral educational complex that considers the cognitive characteristics of the student, because of the systematic accumulation and processing of data on the cognitive activity of students. The following are some examples of the use of information technologies listed in the second section of the article in the course “Economic Theory”.
1002
N. Mutovkina and O. Smirnova
3.1 Use of Internet Resources The following Web resources can be used in teaching economic theory: 1) Search engines in economics (for search, collection, extraction, processing, verification, analysis and storage of information and digital content to solve tasks). 2) Reference and legal databases (Guarantor, Consultant plus) for the search, analysis of regulatory documents in economics. 3) Cloud services (Google Drive, Yandex. Disk, Dropbox, etc.) for storing and sharing information. 4) Communication services (MS Team, Zoom, Skype, etc.). 5) Collaboration platforms (Trello, Miro). An important addition to the educational process toward “Economics” is the analysis of information from the official websites of state authorities, the Central Bank of Russia and the application of this information in solving specific tasks. 3.2 Use of Data Analysis and Visualization Technologies When conducting practical exercises, for example, when performing analytical tasks, it is advisable to supplement them with various visualization tools by preparing a dashboard [18]. Figure 2 shows a screenshot of the money supply analysis task in Google Data Studio. This dashboard is interactive.
Fig. 2. The structure of the money supply in the Russian Federation
Using dashboards and visualization tools is relevant in all areas of professional activity, including personnel management, but in each of them, they will take a unique form adapted to the tasks of a particular business or project. As part of the independent preparation for the business games, it is proposed to use the Miro platform. The Miro platform has a wide functionality with which students can perform their part of the team task both jointly and individually. In addition, the business game involves preparing the performance of teams and here it is possible to use the MS Teams platform, which has already become familiar during the distance learning period,
The Relevance of a Systematic Approach
1003
Fig. 3. “Market VS State” on the Miro virtual board
on which each team prepares a speech with the defense of its own position. Figure 3 shows a version of the task for the business game “Market VS State”. These tools are aimed at developing visualization and teamwork skills (soft-skills) and increasing its effectiveness. 3.3 Use of Knowledge Assessment Technologies When conducting the test asks, it is advisable to use Google forms as a tool. This makes it possible to keep the test execution time under control, does not require printing out assignments on paper, and also reduces the complexity of checking assignments by the teacher and generates a summary statement based on the results of passing the test. In order to conduct a survey, vote or receive feedback from the audience in real time, both at lectures and in practical classes, it is possible to use the service Mentimeter. This allows you to determine the mood of the audience, get a slice of residual knowledge, involve students in an active discussion and conduct a live dialogue with the audience, which makes classes more communicative and digital at the same time.
4 Summary and Conclusion The educational process should be built in such a way that the following principles are fulfilled: 1) All educational tasks should, if possible, be solved with the help of modern software in order to save time and develop skills.
1004
N. Mutovkina and O. Smirnova
2) The software used should be relevant and used in professional activities. 3) The tasks solved by students in the framework of studying individual academic disciplines should be built according to the scheme “from simple to complex”. 4) The results of solving some problems can act as initial data for formalization and solving others. This means the implementation of the so-called “end-to-end” approach to the solution, as well as ensuring continuity and “inheritance” of educational material. Information and educational technologies are based on the theory of modern pedagogical theories that emphasize the development of an active role of students in learning. The rational use of modern information technologies can arouse students’ interest in learning. However, too much information technology should not be used to avoid duplication of information and complication in the educational process. It is enough to choose 2−3 information technologies to solve each type of educational task. When choosing, you should be guided by the principles of accessibility of these technologies, ease of use, and sufficiency of functionality. It is planned to continue the development of modern information technologies and software products in the educational process. Of particular interest is the study of the reaction of students and teachers to introduce digital tools into the educational process. It is planned to develop a method for evaluating the effectiveness of digital tools in the educational process, considering the specifics of the direction and profile of students’ education.
References 1. Alekseeva, T.V., Gubina, L.V.: Application of modern software products in the field of education. Baltic Hum. J. 3(32), 24–28 (2020) 2. Shibaeva, N.A., Voronkova, L.V.: Digital technologies application in higher education as a social innovation of the modern information society. Drucker’s Bulletin 2, 70–80 (2020) 3. Dabas, N.: Role of computer and information technology in education system. Int. J. Eng. Techniques 4(1), 570–574 (2018) 4. Mohamed, N., Al-Jaroodi, J., Jawhar, I.: Enhancing the project-based learning experience through the use of live WEB data. IJMECS 4(11), 33–43 (2012) 5. Wang, J., Liang, H.: Discussion on domestic universities construction of digital teaching platform. IJEME 2(6), 1–6 (2012) 6. Li, W., Zhou, R., Deng, P., Fang, Q., Zhang, P.: Construction of case teaching model for Management specialty supported by information technology. IJEME 2(9), 44–48 (2012) 7. Chang-Xing, L.: Research and reflections on college mathematics teaching based on information educational technology. IJEME 1(2), 43–47 (2011) 8. Wiradinata, T., Antonio, T.: The role of curriculum and incubator towards new venture creation in information technology. Int. J. of Educ. Manage. Eng. (IJEME), 9(5), 39–49 (2019) 9. Al-Samawi, Y.: Digital firm: requirements, recommendations, and evaluation the success in digitization. Int. J. of Inf. Tech. Comput. Sci. (IJITCS) 11(1), 39–49 (2019S) 10. Bibhudutta, J.: An approach for forecast prediction in data analytics field by Tableau software. Int. J. of Inf. Eng. Electr. Bus. (IJIEEB) 11(1), 19–26 (2019) 11. Yakovlev, V.B.: Econometrics in Excel and Statistica: A Textbook. KnoRus, Moscow (2020)
The Relevance of a Systematic Approach
1005
12. Gabdulin, R.R., Lyaskovskaya, E.A., Korovin, A.M., Rets, E.A.: Forecasting demand in the market of road construction equipment using data mining. Bulletin of the South Ural State University. Ser. Computer Technologies, Automatic Control, Radio Electronics 22(3), 117– 131 (2022) 13. Shumakova, A.S., Vdovin, S.A., Ubozhenko, E.V.: Application of ArcGIS and ArcView software products to solve applied economic problems. Regul. Land Property Relations in Russia: Legal and Geospatial Support, Real Estate Valuat. Ecol., Technol. Solut. 3, 213–218 (2021) 14. Fedulova, M.A., Karpov, A.A.: System approach in the design of educational process with the application of information technologies. In: The Collection: Science. Informatization. Technologies. Education. Materials of the XIII International Scientific and Practical Conference, PP. 416–419 (2020) 15. Reutov, V.Ye., Reutova, V.V., Kravchenko, L.A., Troyan, I.A.: Business simulation as an interactive method for training economists. Finance, Banks, Investments 1, 162–171 (2021) 16. UCF Center for Research in Education Simulation Technology. University of Central Florida. https://sites.google.com/view/teachlive/home?pli=1. Accessed: 08 Jan 2023 17. Kravchenko, L.: Online resources that will decorate any activity. Mentimeter. https://novator. team/post/869. Accessed 09 Jan 2023 18. Smirnova, O.V.: To the question of updating the content of economic disciplines using digital technologies. Bulletin Tver State University. Series: Econ. Manage. 2(54), 249–257 (2021)
Construction and Practice of “CAD/CAM Foundation” Course Based on Learning Outcome Ming Chang1(B) , Wei Feng1 , Zhenhua Yao1 , and Qilai Zhou1,2 1 School of Materials Science and Engineering, Wuhan University of Technology,
Wuhan 430070, China [email protected] 2 Faculty of Science, Shizuoka University, Shizuoka 422-8529, Japan
Abstract. This paper introduces the course design and reform of “CAD/CAM Foundation” based on learning outcome. This paper analyzes the challenges brought by the change of learning environment, industry development, students’ learning intentions and habits, and the change of teaching philosophy under the current engineering education background. The design of the course starts from the course objectives. By refining the course objectives, ILOs (Intended Learning Outcomes) were set for the teaching content of each chapter and corresponding practical tasks, assignments, and interactive sessions were designed to examine ILO achievement. Through the quantitative evaluation of ILO, a detailed learning outcome evaluation mechanism was established, the process assessment was enhanced, and the comprehensive evaluation of knowledge objectives and capability objectives was realized. At the same time, it also introduces the teaching strategies and practical experience combining with students’ learning characteristics. Through the outcome-oriented curriculum design and reform, the learning outcome has been improved, and the instructional design and product evaluation of the curriculum can be continuously improved in this way. Keywords: CAD/CAM · Curricula construction · Learning outcome
1 Introduction The ability to apply computer technology is an essential part of university education. With the rapid development and popularization of computer technology, the requirement of students’ computer ability in application field, the character changes of pupils and the development of teaching concept, innovation in CAD/CAM course are required urgently, including the teaching objectives, course content, teaching organization, teaching strategies, etc. Nowadays, few industries can operate without the aid of computer. More and more software were developed to meet the requirement of specific application or used as common platform software. Obviously, it is impossible to embrace all the software operation © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 1006–1015, 2023. https://doi.org/10.1007/978-3-031-36115-9_90
Construction and Practice of “CAD/CAM Foundation” Course
1007
or theories of algorithm in CAD/CAM course. The aim of learning CAD/CAM technology is no longer to handle software but to have a clear understanding of CAD/CAM technology, and utilize CAD/CAM technology to analyze and solve problems [1]. CAD/CAM software should be defined as tools to realize the ideas of user. Students should have the ability to choose proper tools and grasp the usage of tools rapidly. Knowledge competence is not equivalent to competitive competence [2]. Curriculum objectives have become not only knowledge objectives, but also competence objectives [3]. And in some way, the latter will be increasingly important. As to the pupils in nowadays, they are named as the original inhabit of internet. They grew up with the company of electronic products and the internet. The habit of get information via electronic products from internet was deeply ingrained [4, 5]. The rich of resources on internet provide the facility to quick study for students; meanwhile, inappropriate searching keyword and Lack of a complete knowledge framework may leads to misunderstanding. Furthermore, students become less patient to listen to a long speech; especially under the condition they can get the answer quickly via searching. They prefer to logical think rather than boring memory and deduction. These changes are conducive to the realization of curriculum objectives, but put forward higher requirements for curriculum organization. In addition, along with the spread of computer application in education, the uses of course website become normal state. Affected by the COVID epidemic in recent years, appropriate course construction and adaptive teaching organization become more and more important [6, 7]. Furthermore, the ideas of teaching also changed dramatically in recent years [8, 9]. With the gradual implementation of engineering education certification in china, the OBE (Outcomes-based Education) concept has been familiar to educators. And as the core of OBE concept, production-oriented, learning-centered and continuous improvement has been widely adopted in university education [10–12]. According to the curriculum objectives and the characteristics of the curriculum, production-oriented education can be realized in different path, such as project-based teaching, online learning and teaching, and even game-based teaching methods [13–16]. In the practice of OBE concept, the evaluation of learning production is the key point, and it is the foundation for continuous curriculum improvement. Some quantitative calculation methods of curriculum objectives were proposed [17, 18]. Obviously, the rationality of evaluation approach and criterion are the key to realize effective evaluation. Major of the authors attended engineering education certification in two years. The course “CAD/CAM basis” had been reconstructed and designed based on production-oriented approach. In this paper, the practice in the reform of “CAD/CAM basic” course was introduced.
2 Background and Course Analysis 2.1 Course Objective As a course to establish students’ basic knowledge of CAD/CAM, “CAD/CAM basic” is offered to sophomore students. After revisions in recent years, the curriculum objectives have been set as: a) Able to understand and deal with the relationship between human and computer; has clearly understanding of the purpose and significance of CAD/CAM application in professional field. b) Master the basic principles of CAD/CAM in the
1008
M. Chang et al.
course and have the ability to apply these basic principles. c) Able to use CAD/CAM technique in the analysis of engineering application; has a certain sense of innovation and innovation ability. It can be seen that multiple objectives such as engineering ethics, knowledge ability and innovation ability were included. 2.2 Course Content The main content of the course are listed in Table 1. It can be seen from the table that the course mainly contains basic theories of CAD/CAM, the application methods of basic principles and the practical application of CAD/CAM software. Table 1. Main content of the course Chapter
Main knowledge points
Basic understanding of CAD/CAM system Basic concepts, Man-machine relationship, System function, The relationship between operating system, supporting software and application software, Development of application software Engineering data processing
Data storage, Data classification and processing method,
Graphic image processing
Basis of graphic transformation, Image processing
Geometric modeling
Geometric modeling method, Curves and surfaces,
Introduction of modeling software
Basic modeling function, Secondary development
CAM and its applications
CAM basis, Example of CAM application
2.3 Learning Basis and Learning Effect of Previous Sessions Before the course, a questionnaire survey about learning basis and learning willingness was conducted. The result was summarized in Table 2. It can be seen that most students have more or less access to CAD/CAM software. But few students can handle one or two CAD/CAM software deftly although most of them had tried to teach themselves. They are eager to master CAD/CAM technology, but they confuse mastering CAD/CAM technology with mastering software usage. Meanwhile, quick access to information via the internet did not help them much in their self-study. Boring commands, complex operations and learning without specific goals lead to poor learning results.
Construction and Practice of “CAD/CAM Foundation” Course
1009
Table 2. Learning basis and learning willingness (survey results) Question
Majority answer
Do you ever have access to CAD/CAM software more or less?
Yes
Can you handle one or more CAD/CAM software expertly?
Cannot
Have you ever tried to learn CAD/CAM software through self-study?
Yes
What approaches have you used to learn CAD/CAM techniques?
Textbook or video tutorials
What do you want to learn from course?
Usage of one or two software
Another survey was conducted on the performance of previous students in subsequent studies. Most of them have a good grasp of CAD/CAM basic knowledge and can use certain CAD/CAM software to complete certain tasks. But they didn’t do as well when confronted with software they hadn’t encountered before. This indicates that their understanding of CAD/CAM technology was not sufficient which results in the insufficient of learning ability.
3 Practice of Curriculum Reform In production oriented learning, students are the main body of learning, so the design and implementation of learning should be carried out on this basis. The focus of this reform is to combine curriculum objectives with students’ learning characteristics, fully mobilize their initiative, provide appropriate learning materials, conduct appropriate teaching organization, and carry out assessment in a reasonable way. 3.1 Curriculum Planning Based on Outcome Objectives The course was reorganized in accordance with curriculum objectives. To ensure the objective evaluation of learning production, the curriculum planning was conducted logic is shown in Fig. 1.
Fig. 1. The logic of curriculum planning
1010
M. Chang et al.
The relationship between the main content of the course and the course objectives were established via setting the intended learning outcomes (ILOs) of main content. The ILOs were written to describe the content to be learned, the level to be achieved and descriptions enacted to outcome [19]. The main content was divided into four categories. The ILOs setting and the course objective supported are listed in Table 3. The course objective can be subdivided into several ILOs according to the classification of content. In this way, the course objectives were broken down into smaller objectives, and the instructional design became more targeted. Table 3. Main content, ILOs and Course objective Main content
ILOs
Course objective
Cognitive information
Understanding of basic concept, consistence, importance or meaning
a)
Foundations (mathematical, computer skills and engineering knowledge)
Have a ground foundation to support the application of CAD/CAM
b)
Methodologies related to CAD/CAM
Understand how CAD/CAM works; Grasp certain Methodologies in CAD/CAM
b) & c)
Experiment section
To complete specific tasks; Extended learning and design
b) & c)
According to ILOs, teaching materials were reasonably organized. In addition to the teaching materials of knowledge content, the extended reading materials, data reference guidance, basic knowledge-related exercise materials, and experimental guidance materials were organized according to the needs of ILOs. For example, experiments are responsible to the realization of course objects b) and c). Step by step approach has been adopted in the design of experiment [20], to gradually realize the purpose from the practice of theoretical content to independent learning, practice ability and creative thinking. Every experiment task contains three parts and materials prepared are listed in Table 4. Table 4. Teaching material organization of experiment section Experiment content
Materials prepared
Course objective
Part1
Well prepared programming files, Guidance video of software operation
b)
Part2
Guidance of information acquire, Supplementary materials prepared for inquiry purposes
b) & c) (continued)
Construction and Practice of “CAD/CAM Foundation” Course
1011
Table 4. (continued) Experiment content
Materials prepared
Course objective
Part3
Examples Recommendation of design target
b) & c)
Part 1 is the verification part, which mainly realizes the basic methods taught in the course through software. The part was designed to be easy to complete, which helps to increase students’ interest and serves as basis for students to make their own design. So, well prepared programming files and guidance video of software operation were provided. The part 2 is advanced part, which can be realized by modify the given program to meet certain requirement. In this part, learners need to find and use methods which were not given in the example. Supplementary materials and guidance of information acquire were provided. The part 3 is a free design part and some recommended topics and examples were offered. The implementation methods were divided as class lecture and interactive sessions, group discussion, after class reading, software practice, and so on. The adoption of implementation methods depend on concrete knowledge type and the learning characteristics of students. Along with the implementations, the evaluations of ILOs achievement were also designed to quantify the degree of achievement. For example, the results of the experiment were quantified from four aspects: design idea, implementation degree, discussion of results, and information acquisition and application. For other approach, the quantitative evaluations were also defined in the same way. After such setting, the evaluations of learning outcome can be more detailed and accurate. 3.2 Mobilize the Initiative of Students Making students clearly recognize the outcome goal of learning is the basis for realizing learning-centered. The course objectives and ILOs of each chapter should be posted to students in class and course website. So does to the evaluation method. In this way, clear learning goals were established for students, as well as how to get a better score. Besides, a clear framework of knowledge should also be presented to students. This helps students to establish a complete knowledge system, master the correlation of knowledge, and then be able to analyze problems from the perspective of systems engineering. In practice, the knowledge frame was presented in the way of mind map. In the learning process, the students’ interest relate to challenge and motivation. Proper balance of challenge and motivation will keep students interested in learning. Too difficult challenges will greatly increase the probability of students to give up, while too easy will become boring and lose interest. In classroom teaching, teaching time should be allocated according to the type of content. Contents that need logical understanding should be focused, and specific implementation steps or procedures should be minimized. For example, when introducing the data processing method, the focus should be on the basic principle and characteristics of the method, while the program implementation
1012
M. Chang et al.
of the method can be minimized or omitted. This is in line with the characteristics of students’ learning interests. Too much detailed teaching will only make students feel bored and lose interest. Appropriate interaction such as questions and quizzes were adopted. With the convenience of the course website, these interactions can take place in real time to all or to individuals. In the experiment teaching, well prepared experimental guidance materials can help students to get the goal in few minutes and it is helpful to arouse the interest of students. At the same time, it also reduces the obstacles in program operation, command use, especially when using unfamiliar software. In the second and third parts of the experiments, it is necessary to give priority to several students who are progressing faster. It turns out that students are willing to mentor and share with each other when there are gaps in their progress. Especially in the third part, discussing and sharing experiences will further stimulate students’ creative desire. In the study of Wang [21], this kind of learning environment would adjust students’ learning approaches and study behaviors, regardless of individual differences. Besides, it is also helpful to review students’ works. In the course, comments were made by teachers and other students. For students, this process can not only give them a sense of achievement but also gain new harvest [22]. 3.3 Outcome Evaluation In order to reasonably evaluate the learning outcome, a variety of evaluation approaches was adopted. As mentioned above, this reform sets ILO for every teaching links in the curriculum, and each ILO can be quantified based on the assessment of corresponding tasks. The data sources of evaluation included the learning performance in classroom and on course website, the completion of course assignments, the completion of experiment tasks and score of final exams, etc. The evaluation of learning process was strengthened and the criteria for evaluating were refined. The changes of evaluation are listed in Table 5. Table 5. Changes of evaluation Items
Portion (previous)
Portion (present)
Details (previous)
Details (present)
Final exam
70%
50%
Knowledge mastery
Knowledge mastery, Ideas for application
Experiment
20%
30%
Process and Result
Design, Process, Discuss, Information acquire and usage
Learning process
10%
20%
Assignment
Assignment, quizzes, Learning performance
In this way, the achievement of course objectives is quantitatively calculated and documented, which serves as the basis for course effectiveness analysis and continuous improvement.
Construction and Practice of “CAD/CAM Foundation” Course
1013
4 Results and Discussion After two years’ curriculum construction and practice, the mode of teaching and evaluation of the curriculum has been fixed. The results of the achievement evaluation were used in continues improvement. The achievement of each course objectives and the achievement of each student were calculated and graphed. The results of last term are shown in Fig. 2. The achievement degree of each course goal exceeds 0.8, reaching the preset goal. It can also be clearly seen from the figure that the achievement of course objectives of some individual students is lower than 0.8. By analyzing these scores, why and what need to be improved can be found out.
Fig. 2. Course objectives achievement
From the perspective of students’ performance in the learning process, time spending on the course website increased about 20%, classroom teaching is more active than before, and students show more creative performance in the course, especially in the experimental section.
5 Conclusion To realize the transformation of the curriculum to be outcome-oriented, learning-centered and continuous improvement, “CAD/CAM basis” course was redesigned. After two-year practice, following conclusions can be drawn: 1) The refinement of curriculum outcome objectives into specific ILOs is conducive to the effective quantification of curriculum objectives.
1014
M. Chang et al.
2) Organizing teaching resources around ILOs is conducive to the achievement of outcome objectives. 3) Curriculum planning aiming at learning production is conducive to continuous improvement of curriculum.
References 1. Ye, X., Wei, P., Chen, Z., Cai, Y.: Today’s students, tomorrow’s engineers: an industrial perspective on CAD education. Comput. Aided Des. 36(14), 1451–1460 (2004) 2. Hyland, T.: Competence, knowledge and education. J. Philos. Educ. 27(1), 57–68 (1993) 3. Chen, J., Liu, D., Li, X., Zeng, D.: Curriculum system of construction management specialty in universities and colleges based on core knowledge and competence. J. Arch. Educ. Inst. Higher Learn. 23(01), 41–45 (2014). (in Chinese) 4. Bere, A., Deng, H., Tay, R.: Investigating the impact of eLearning using LMS on the performance of teaching and learning in higher education. In: 2018IEEE Conference on e-Learning, e-Management and e-Services (IC3e), Langkawi, Malaysia, pp. 6–10 (2018) 5. Shivanibindal, S.V.: Impact of ICT on teaching and learning. ZENITH Int. J. Multidisciplinary Res. 3(1), 262–273 (2013) 6. Jandri´c, P., et al.: Teaching in the age of Covid-19. Postdigital Sci. Educ. 2, 106–1230 (2020) 7. Sobaih, A.E.E., Hasanein, A., Elshaer, I.A.: Higher education in and after COVID-19: The impact of using social network applications for E-Learning on students’ academic performance. Sustainability. 14(9), 5195 (2022) 8. Zhu, Y.: The Supply-side reform of higher education from the perspective of government function transformation. J. Higher Educ. 37(08), 16–21 (2016). (in Chinese) 9. Mihaela, A., George, M.T., Ligia, V.V.: The Exposure of Chinese higher education to the development of international education system. Ovidius University Annals, Economic Sciences Series, xiii(1), 353–358 (2013) 10. Wang, J., Zhu, Z., Li, M.: Result-oriented: From certification concept to teaching mode. China Univ. Teach. 6, 77–82 (2017). (in Chinese) 11. Xiaoying Lin, Sheng Xu, Zhipeng Liu. An analysis of curriculum teaching reform based on the background of engineering education certification[J]. Journal of Higher Education, 2022, 8(6): 136–138,142. (in Chinese) 12. Wang, J., Zhang, Q., Li, C.: Research and discussion on teaching reform of modern control theory based on OBE concept. Univ. Educ. 8, 81–83 (2022) 13. Sola Guirado, R.R., Vacas, G.G., Alabanda, Ó.R.: Teaching CAD/CAM/CAE tools with project-based learning in virtual distance education. Educ. Inf. Technol. 27: 5051–5073 (2022) 14. Berselli, G., Bilancia, P., Luzi, L.: Project-based learning of advanced CAD/CAE tools in engineering education. Int. J. Int. Des. Manuf. (IJIDM) 14, 1071–1083 (2020) 15. Akhtar, S., Warburton, S., Xu, W.: The use of an online learning and teaching system for monitoring computer aided design student participation and predicting student success. Int. J. Technol. Des. Educ. 27, 251–270 (2017) 16. Goli, A., et al.: Architectural design game: a serious game approach to promote teaching and learning using multimodal interfaces. Educ. Inf. Technol. 27, 11467–11498 (2022) 17. Wang, Z., Ma, Y., Gao, Y.: Calculation method of evaluation target value for the achievement degree of curriculum objectives: for engineering education professional certification. Educ. Teach. Forum 21, 41–44 (2021) 18. Liu, Z., Feng, G., Zhao, H.: Achievement evaluation of practical courses for engineering education certification: practical courses of computer science and technology. Educ. Teach. Forum 37, 108–109 (2020)
Construction and Practice of “CAD/CAM Foundation” Course
1015
19. Biggs J. Teaching for Quality Learning at University [M]. SRHE and Open University press, 2003, 2nd edition 20. Yixian, D., Tian, Q., Xuan, D., He, K.: CAD/CAM courses integration of theoretical teaching and practical training. Procedia. Soc. Behav. Sci. 116(1), 4297–4300 (2014) 21. Wang, X., Yelin, S., Cheung, S., Wong, E., Kwong, T.: An exploration of Biggs’ constructive alignment in course design and its impact on students’ learning approaches. Assess. Eval. High. Educ. 38(4), 477–491 (2013) 22. Gelmez, K., Arkan, S.: Aligning a CAD course constructively: telling-to-peer and writingto-peer activities for efficient use of CAD in design curricula. Int. J. Technol. Des. Educ. 32, 1813–1835 (2022)
Research and Practice of Ideological and Political Education in the Context of Moral Education and Cultivating People Geng E. Zhang1,3 and Liuqing Lu2(B) 1 Faculty of Intelligent Manufacturing, Nanning University, Nanning 530000, China 2 Personnel Office of Nanning University, Nanning 530000, China
[email protected] 3 Faculty of Engineering, Universiti Malaysia Sabah, 88400 Kota Kinabalu, Sabah, Malaysia
Abstract. Curriculum ideological and political construction is an important guarantee for the implementation of the fundamental task of strengthening moral education and cultivating people, “for whom to cultivate people, what people to cultivate, how to cultivate people” is always the fundamental problem of education. How to integrate professional education with moral character education to achieve ideological and political education of all members and the whole process. By consulting literature and under the guidance of curriculum moral character construction, this paper studies and practices the reform of curriculum ideological and political teaching on the teaching system, teaching content, teaching methods and teaching evaluation. After a semester of teaching practice, the students and peer teachers’ evaluation forms are collected and analyzed, and the average score is 95 points, indicating that the moral character reform of the curriculum has achieved a certain effect, which can play a very good supporting role in cultivating students in moral education. Keywords: Curriculum ideological and political education · Strengthen moral education and cultivate people · Teaching reform
1 Introduction Since the curriculum moral character education was proposed in 2016, it has gone through six years. The state attaches great importance to it, all levels and departments take joint actions, colleges and universities focus on the implementation, constantly promote the teaching reform of curriculum ideological and political education. On August 14, 2019, the General Office of the CPC Central Committee and The General Office of the State Council printed and distributed Several Opinions on Deepening the Reform and Innovation of Ideological and Political Theory Courses in Schools in the New Era. Its guiding ideology is to accelerate the modernization of education, build an educational powerhouse, run an education that the people are satisfied with, and strive to cultivate new generation to take on the great task of national rejuvenation, train socialist © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 1016–1027, 2023. https://doi.org/10.1007/978-3-031-36115-9_91
Research and Practice of Ideological and Political Education
1017
builders and successors with all-round development of moral, intellectual, physical, aesthetical and labor education. On April 25, 2022, the CPC Secretary General Xi Jinping made an inspection of Renmin University of China and delivered an important speech, stressing that “for whom, who and how to cultivate people” is always the fundamental issue of education. We will carry out the fundamental task of strengthening moral education and cultivating people, and cultivate new generation to take on the great task of national rejuvenation for the party and the country. Textbook [2020] 6 the notice of Propaganda Department of the CPC Central Committee and Ministry of Education on Printing and Distributing the Implementation Plan for the Reform and Innovation of Ideological and Political Theory Courses in Schools in the New Era. The university stage focuses on enhancing students’ mission, further enhancing the pertinence and effectiveness of teaching according to the requirements of different types of schools and different levels of talent training, consciously practicing socialist core values. The notice of Guidance Outline for the Construction of Ideological and Political Courses in Colleges and Universities No. 3 of Higher Education Office [2020] emphasizes that professional courses are the basic carrier of the construction of moral character courses. It is necessary to combine the education of Marxist standpoint, viewpoint and method with the cultivation of scientific spirit in the course teaching, so as to improve students’ ability of correctly understanding, analyzing and solving problems. For engineering courses, emphasis should be placed on strengthening engineering ethics education, cultivating students’ craftsman spirit of striving for excellence, and inspiring students’ feelings and mission of serving the country through science and technology. Under the guidance of various documents, the related research on curriculum ideology and politics has been deepened continuously, more and more colleges and universities have participated in the research boom of curriculum ideology and politics, and the research results have increased exponentially. Table 1. The distribution statistics of curriculum ideological and political research topics from 2017 to 2022 Year of publication
Theoretical research
Practical research
2017
10
2
2018
12
2019 2020 2021
School
Majors and courses
Teaching
Text-book
Teacher
Resources
Total
15
3
1
26
15
25
18
6
3
93
69
35
91
51
1
15
7
266
119
61
156
92
4
21
4
457
12 4
35
Table 1 shows the distribution statistics of subjects of moral character studies in the curriculum from 2017 to 2022 [1]. It can be seen from Table 1 that shows on morality studies in the curriculum are mainly focused on majors and courses. The majority of colleges and universities continue to implement the concept of strengthening moral education and cultivating people, and actively carry out morality reforms and studies
1018
G. E. Zhang and L. Lu
in professional courses, with 156 papers increased by 156 times from none in 2017 to 156 in 2021, effectively solve the organic integration of profession and ideological and political areas.
Fig. 1. Statistics of academic ideological and political papers from 2017 to 2022(Unit: papers)
Figure 1 shows the published length of periodical papers on moral character themes from 2017 to 2022 according to the statistics of China National Knowledge Network. In 2019, under the guidance of the Ministry of Education’s notice on the Guidelines for the Construction of Ideological and Political Construction in Higher Education Curriculum, 1314 journal papers on the subject of curriculum moral character education were published, and 7728 journal papers were published in 2022, with an increase of 488%. There are 34 references in the search terms of course ideology and politics + automotive electrical appliances on CNKI. There is only one paper on the moral character research of Automotive Electrical Appliances. In the paper Exploration on the Course Ideological and Political Thinking of Automotive Electrical Appliances in Higher Vocational Colleges from the Perspective of Three Education [2]. Li Jing discussed the combination of techniques and methods in the course of Automotive Electrical Appliances in higher vocational colleges, but did not mention the course morality research on the course of Automotive Electrical Appliances in ordinary undergraduate colleges. This paper studies and practices how to carry out moral character teaching in the course of Automobile Electrical Appliances in application-oriented colleges and universities.
2 Problems in Curriculum Ideology and Politics Yang Mei et al. put forward the following problems of moral character education in college curriculum [3]. First, teachers of professional courses in colleges and universities do not have strong ideological and political awareness of the curriculum, they lack a comprehensive knowledge and understanding of moral character thinking in the curriculum. Professional teachers think that morality education in the curriculum is the responsibility of ideological and political teachers, which is a burden to themselves. They do not pay enough attention to curriculum moral character education and lack the initiative and enthusiasm to carry out curriculum ideological and political reform. Second, the overall consciousness of moral character construction in college curriculum is not strong, the boundaries of moral character construction in different professional
Research and Practice of Ideological and Political Education
1019
courses are blurred, lack of personality, cannot reflect the professional characteristics, cannot carry out curriculum ideological and political research in specific courses under the awareness of professional integrity, most of the research reflect as generic research. Curriculum moral character education can be divided into “explicit ideological and political education” and “implicit ideological and political education”. Wang Yiqing et al. pointed out that “implicit ideological and political education” has some problems, such as unclear principles and concepts, lack of close integration with majors, lack of abundant educational resources, and lack of ideological and political literacy of education subjects, and the paper proposed corresponding solutions [4]. Dong Hui et al. elaborated the difficulties of moral character teaching from the perspective of teachers’ ability, and pointed out that there were problems to be solved in the aspects of teachers’ consciousness, knowledge level, teaching ability, teaching wisdom and morality experience in curriculum ideological and political construction [5]. Liang Jin et al. analyzed the practical predicament of integrating curriculum ideology and politics into teaching in colleges and universities in the new era from the three aspects of colleges, secondary colleges and professional teachers [6]. This paper also pointed out that the main problems hindering the integration of curriculum moral character education into teaching are the insufficient supply of supporting policies in colleges, the insufficient play of the main front role of secondary colleges, and the lack of curriculum ideology and politics ideas, abilities and feelings of college teachers. This paper also put forward the corresponding solution strategy. Sun Gang et al. pointed out that teachers and students generally lack a deep understanding of the curriculum moral character training mode, the relevant mechanism is not perfect, the overall design and implementation ability is not mature, the endogenous motivation of professional teachers is insufficient, evaluation standards are not clear and other factors affect the construction of professional curriculum ideological and political teaching, and put forward corresponding solutions [7].
3 Research on the Path of Ideological and Political Teaching Reform of Automobile Electric Appliances Course Major courses should reflect the characteristics of their major, as a core course of transportation related majors [2], the moral character education reform of the curriculum should organically integrate the national education objectives, college education objectives, professional education objectives, curriculum teaching objectives and teaching unit objectives. In addition, the national, college and professional objectives should be deepened into the curriculum and detailed to each teaching unit. Reflect the curriculum ideological and political education in every teaching unit, so as to enrich the students’ quality education invisibly. Diversified reform should be carried out from personnel training program, curriculum teaching system, teaching content, teaching method and teaching evaluation, so that explicit moral character education can be carried out simultaneously with implicit moral character education, constantly improve morality education, and fully implement the idea of strengthening moral education and cultivating people.
1020
G. E. Zhang and L. Lu
3.1 Reform of ideological and Political Teaching System of Automobile Electric Appliances Course Fang Caihong et al. proposed that the construction of curriculum moral character system in colleges and universities should follow three principles and three dimensions, from the three principles of combining theory and practice, explicit education and implicit education, overall promotion and key breakthrough, and from the three dimensions of educational operation mechanism, educational guarantee mechanism and educational evaluation mechanism to build a perfect system of curriculum morality education [8]. Constantly excavate the history, culture and industry spirit of different majors and refine them into the core value system of majors, so as to integrate them into the curriculum teaching system. According to the characteristics and advantages of transportation majors, we should scientifically and reasonably expand the breadth, depth and temperature of professional morality education, and build a curriculum moral character system with comprehensive coverage, rich types, progressive levels and mutual support, as shown in Fig. 2. In view of the course learning objectives into task design, the task-driven main line online and offline mixed teaching mode is used to implement the teaching process, the curriculum moral character education is integrated into it, it can be OBE-oriented and combined with the curriculum objectives to carry out diversified assessment. assessment learning assessment Curriculum objectives
A taskOBE -
driven
oriente
online and
d
offline
assess
mixed
ment
teaching model
Task presentation
Process demonstration
Knowledge, technology,
Two
emotional
pairs
integration
and
of spiraling Project implementation
Project acceptance
curriculum ideological
one though t test
and political
course
system
objecti ves
Summary evaluation
Diversified assessment
Dual base test
Dual skill test
I&S assessment
Fig. 2. Ideological and political teaching system of knowledge, technology, emotional integration
3.2 Ideological and Political Teaching Content Reform of Automobile Electric Appliances Course Automobile electrical equipment is one of the standards to measure the advanced level of automobiles. The electronic control technology of automobiles is gradually developed with the development of electronic technology and the improvement of automobile
Research and Practice of Ideological and Political Education
1021
related regulations (fuel consumption regulations, emission regulations and safety regulations). This course mainly introduces the structure, composition, working principle, maintenance experiment and fault diagnosis method of automotive power system, starting system, ignition system, instrument system, lighting and signal system and auxiliary electrical system. The specific teaching content mainly includes the structure, type, working principle, working characteristics, capacity and influencing factors, charging and fault of battery; structure, type, working principle, characteristics of silicon rectifier generator; principle and classification of regulator; protection circuit of automotive power system; structure, working principle, characteristics and models of DC series motor; starter transmission mechanism; starter drive protection circuit; requirements of automotive ignition system; composition and working principle of traditional ignition system; analysis of the working process of ignition system, factors affecting secondary voltage; structure, composition, working principle of instrument system, lighting and signal system, auxiliary electrical system; composition and characteristics of automotive electrical circuit; distribution devices of automotive electrical circuit; expression and analysis methods of automotive circuit diagram; typical car circuit analysis and other contents. After summarizing the content of the course, it mainly includes three parts: the structure, working principle and maintenance of automotive electrical components. How to integrate moral education elements into each teaching content is shown in Table 2, the morality elements of the curriculum are reflected in the selection and arrangement of teaching content [9]. 3.3 Reform of Ideological and Political Teaching Methods of Automobile Electric Appliances Course Teaching methods directly affect the degree of realization of teaching objectives and excellent grade of teaching effect. In the process of teaching implementation, it is necessary to pay attention to the diversification of teaching methods and effective interaction between teaching and learning, adhere to the student-centered, teacher-led. The organic integration of various teaching methods reflects the diversity and flexibility of teaching methods, as well as two-way interaction between teachers and students [10]. The reform of ideological and political teaching method of Automobile Electric Appliances course mainly includes two aspects: 1) BOPPPS effective teaching model is adopted for teaching design. The BOPPPS teaching model mainly includes six elements, among which the most critical one is the design of learning objectives, which can be divided into cognitive objectives, skill objectives and emotional objectives. Integrating moral education elements into emotional objectives can reflect curriculum moral character education. 2) In the P2 participatory teaching link of BOPPPS teaching model, a variety of effective teaching methods are used to change students’ passive learning into active learning. Through a variety of effective teaching methods such as scene teaching, group discussion, smart site, fish tank teaching method, carousel, peer teaching method, students’ participation can be improved, teacher-student interaction and student-student interaction can be strengthened, students’ interest in learning can be improved, so that can
1022
G. E. Zhang and L. Lu Table 2. Ideological and Political Teaching Contents
Serial number
Course content
Ideological and political elements
1
Introduction
From the development of automobile electronic technology leads to the development and application of our country electronic technology, cultivate students’ spirit of scientific and technological power
2
Maintenance, use and overhaul of storage battery, generator, starter and electrical system
Moral education elements are added to the contents related to the use, maintenance and overhaul of various electrical systems to train the students of automobile service engineering, cultivate the spirit of great craftsman, the consciousness of safe operation based on professional ethics, and the spirit of team cooperation
3
Structure and principle of storage battery, generator, starter and electrical system
Cultivate students’ rigorous logical analysis ability and scientific research spirit, and cultivate students’ awareness of environmental protection and ecological civilization based on automobile electrical appliances
4
Complete vehicle circuit repair
Combined with various models, the vehicle line overhaul, interspersed comparison of various countries of automobile enterprises in the development of automotive electrical system, put forward the advantages of the development of automotive electrical in our country, as well as the need to improve the place, encourage students to become strong in science and technology
achieve effective teaching. Cultivate students’ sense of teamwork, dialectical thinking, spirit of craftsman, spirit of science and technology, and awareness of safety and environmental protection. 3.4 Ideological and Political Teaching Evaluation Reform of Automobile Electric Appliances Course The subject of curriculum moral character teaching evaluation is both the teacher and the student, and the evaluation of the effect of curriculum moral character teaching should be multi-dimensional, multi-level and comprehensive [11]. Teaching evaluation should be carried out in the whole process of learning and teaching, closely combined with cognitive goals, using more authentic evaluation methods in the evaluation, taking students as the center, experiencing participatory evaluation based on problems, and giving timely feedback to the evaluation results. The quality evaluation system of curriculum ideological and political “cloud teaching” based on 4E theory is constructed from the four dimensions of economy, efficiency, participation and effectiveness, which improves the depth and breadth of theoretical research on curriculum moral character quality evaluation [12]. The “collaborative evaluation” of professional course teachers and ideological and political counselors has certain effectiveness in solving the problem of morality evaluation of professional courses [13]. The moral character teaching evaluation of Automobile Electric Appliances course should reflect the elements of safe operation, the elements of team cooperation and
Research and Practice of Ideological and Political Education
1023
craftsman spirit. In the group discussion and report, safe operation accounted for 10%, team cooperation accounted for 20% and craftsman spirit of great power accounted for 20%. Morality elements should be deeply integrated into the teaching evaluation.
4 Effect of Ideological and Political Teaching Reform of Automobile Electric Appliances Course There are various ways to evaluate the effect of teaching reform, including questionnaires, quality evaluation forms, multiple choice questions and learning experiences. In order to fully understand the ideological and political reform of Automobile Electric Appliances course, this paper adopts the form of quality evaluation table for data collection and analysis. After the end of the course, students and fellow teachers will evaluate according to the lecture content. Table 3 shows the statistical table of teaching quality evaluation in the first semester from 2022–2023. There are 20 items related to ideology and politics in the quality evaluation form, the maximum score is 5 points for very good, the minimum score is 1 point for not very good, and the total score is 100 points. Table 3. Evaluation table of ideological and political teaching quality Serial number
Evaluation content
5 points
1
This course takes 5 moral cultivation as the central link, educating people in the whole process and in all aspects
2
It can improve 5 students’ professional skills and professionalism
3
The moral character 5 content integrated into the course keeps up with the trend of The Times and can be attractive to students
4
The course content 5 design conforms to the requirements and rules of science and engineering education
4 points
3 points
2 points
1 point
(continued)
1024
G. E. Zhang and L. Lu Table 3. (continued)
Serial number
Evaluation content
5 points
4 points
5
Teachers in the course of moral character construction has a certain research result
6
The content of 5 professional knowledge has synergistic effect with the content of moral character theory
7
Moral character 5 material content is novel, teaching courseware, cases, exercises and other teaching materials are rich
8
It can integrate 5 knowledge imparting, ability cultivation and quality education
9
It can integrate moral character elements into the teaching content invisibly
10
It can enhance the 5 affinity and pertinence of moral character education
11
When it comes to the teaching of moral character content, students can keep a good mental state
4
12
Students take the initiative to ask questions to teachers or discuss moral character content
4
3 points
2 points
1 point
4
5
(continued)
Research and Practice of Ideological and Political Education
1025
Table 3. (continued) Serial number
Evaluation content
5 points
4 points
13
Curriculum 5 assessment forms are diverse and not simple
14
Establish a teaching feedback mechanism and have regular face-to-face communication with students or student representatives
15
Homework or 5 examination questions have both professional assessment and moral character assessment
16
Can use a variety of modern teaching methods to carry out curriculum teaching
17
Be able to master the knowledge points required by this course and apply them skillfully
4
18
It has a high recognition of the necessity of moral character courses in science and engineering
4
19
The course cultivates 5 students’ practical ability and innovation ability
20
Students can put the moral character requirements in the curriculum into practice
5
5
5
3 points
2 points
1 point
1026
G. E. Zhang and L. Lu
As can be seen from Table 3, the average score is 95, indicating that the moral character reform of the curriculum has achieved certain effects and can play a good supporting role in cultivating students in moral education.
5 Conclusion Ideological and political education plays a vital role in the talent training of contemporary colleges and universities. Education and politics is not the same as ideology and politics. It refers to students who receive moral education in the study of professional courses and learn how to be a man before learning. There is a Chinese saying in China that "ten years to grow trees, and a hundred years to cultivate people". Moral education is not only the content of learning in the political course, but also should be permeated into each course, so that students can subtly learn knowledge, understand the truth, set up the correct socialist core values, and become the successors of the new era. This paper analyzes and research the documents related to morality education. Found that few studies were conducted on the Automotive Appliances course. For this purpose, this paper carries out the research and practice of curriculum ideological and political reform from the teaching reform path, teaching system, teaching content, teaching method and teaching evaluation, Put forward the curriculum system of the deep integration of knowledge, technology and situation. The teaching content was designed using the BOPPPS teaching model. The morality content throughout, and designed a complete set of course evaluation forms. Some results have been achieved. It is a reference for the ideological and political reform of other courses. Acknowledgment. This paper is supported by: (1) The second batch of Nanning University “Curriculum Ideological and Political Education” demonstration course construction Project, No. 2020SZSFK19; (2) Diversified “Collaborative Ideological and Political Education” teaching team, No. 2022SZJXTD03.
References 1. Lu, D.: The research progress, difficulty focus and future trend of curriculum ideology and politics in the new era. J. Xinjiang Normal Univ. (Philos. Soc. Sci. Ed.) 03, 1–16 (2022) 2. Li, J., Xie, D.: The ideological and political exploration of automobile electrical appliance course in higher vocational colleges from the perspective of San Quan education. Time Motor 08, 62–63 (2022) 3. Yang, M., Zhzo, S., Li, J.: Current situation, problems and countermeasures of ideological and political development in college curriculum – based on visualization map and Cite Space. J. Shaoyang Univ. (Soc. Sci. Ed.) 21(02), 95–99 (2022) 4. Wang, Y., Li, L., Hao, L.: Problems and improvement strategies of “recessive education” in ideological and political teaching in colleges and universities. Educ. Theory Pract. 40(3), 37–39 (2020) 5. Dong, H., Du, J.: The difficulties of ideological and political promotion of curriculum and its solutions. Ideol. Theoret. Educ. 5, 70–74 (2021)
Research and Practice of Ideological and Political Education
1027
6. Ling, J.: The realistic dilemma and mechanism innovation of college curriculum ideological and political integration into professional teaching in the new era. Occupation (04), 33–36 (2022) 7. Sun, G., Li, H., Fang, Y., Zhao, C.: Restrictive factors and effective ways of ideological and political thinking in applied university curriculum. J. Changchun Normal Univ. 41(02), 93–96 (2022) 8. Fang, C.: Research on the construction of ideological and political system of college curriculum under the background of strengthening moral education and cultivating people. J. Soc. Sci. Jiamusi Univ. 40(01), 233–235+239 (2022) 9. Zhang, J., Mao, X.: Exploration and practice of implementing curriculum ideology and Politics in the teaching of Automobile Electrical and Electronic Control Technology. Time Motor 01, 99–100 (2022) 10. Meng, Z., Li, L.: Some problems and improvement paths in ideological and political teaching practice of curriculum. Chin. Univ. Teach. 03, 51–57 (2022) 11. Jun, C.: Thinking on Constructing Ideological and political evaluation system of college curriculum. J. Chongqing Univ. Sci. Technol. (Soc. Sci. Ed.) 03, 106–112 (2022) 12. Hongmei, M.: Construction of curriculum ideological and political “cloud teaching” quality evaluation system based on 4E theory. J. Soc. Sci. Jiamusi Univ. 40(02), 224–227 (2022) 13. Tianqi, Z.: Research on the effectiveness of the “Collaborative Evaluation” mechanism of ideological and political effects of professional courses. J. Higher Educ. 8(04), 168–171 (2022)
A Quantitative Study on the Categorized Management of Teachers’ Staffing in Colleges and Universities Zhiyu Cui(B) Personnel Department, Wuhan University of Technology, Wuhan 430072, China [email protected]
Abstract. The reasonable allocation of university faculty has always been one of the important contents of university management. Based on the literature review of university faculty staffing management, this paper clarifies the principles and basis that should be followed in university staffing. By using the mathematical statistical methods such as exponential smoothing, which changes with time, the student equivalent of a university is predicted, and the number of teachers and the corresponding number of teaching assistants are allocated accordingly. Through dynamic management of university staffing, adjustment of the position proportion of various types of personnel, and analysis of the reasonable degree of the allocation of the number of personnel in each post, in order to achieve the purpose of making full use of staffing benefits. The paper expounds the need of social and economic development and the importance of building a high-level university faculty, and puts forward corresponding suggestions for the future staffing management, aiming at creating conditions for promoting the characteristic development of university management. Keywords: Universities · Stuffing management · Cubic exponential smoothing method · Quantitative study
1 Introduction Higher education is related to the development of the national economy. The quality of education directly determines the future of the country [1, 2]. With the further deepening of the reform of the higher education system, every university regards continuous optimization of the structure, stable improvement of quality and overall improvement of efficiency as the goals of reform and pursuit [3]. To achieve the reform objectives and promote the reform of university management system, one of the contents is to do a good job in university staffing and post management [4]. A research team has conducted a questionnaire survey and evaluation on the progress of university reform based on expert consultation. In one survey, a total of 230 questionnaires were distributed in colleges and universities across the country, and 203 valid questionnaires were recovered. Only 3.4% of the respondents believed that the reform effect was very significant, and 17.2% of the respondents commented that the result was very insignificant, while the rest of © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 1028–1038, 2023. https://doi.org/10.1007/978-3-031-36115-9_92
A Quantitative Study on the Categorized Management
1029
the respondents commented that the reform effect was not high. Among the eight evaluation indicators listed, the reform of institutions and staffing system of colleges and universities is considered as “not significant” and “not significant”. Such investigation and evaluation results have caused many scholars’ deep reflection [5–7]. It can be seen that it is imperative to implement the staffing and post setting of institutions of higher education. A new mechanism should be established to not only dilute and simplify the establishment of staffing, but also strengthen the self-restraint and self-control of staffing management. The construction of teachers is the most important guarantee for the quality of higher education [8–10]. Headcount management is an effective way to reasonably allocate human resources in colleges and universities. It can effectively control the total number of personnel, scientifically plan the proportion of personnel structure, dynamically adjust the structure of staff and workers, and has a strong guiding role in the overall development of colleges and universities. Scientific and reasonable staffing verification and post setting management is the key to building a reasonably structured, capable, efficient and dynamic talent team, and also the key to promoting the discipline construction of colleges and universities, which has a guiding role for the scientific development of the university [11]. The establishment of posts is not simply an increase or decrease in the number, but a reorganization and optimization of resources. From the perspective of economics, the staffing and posts of colleges and universities have economic characteristics [12, 13]. By strengthening the accounting of personnel costs, we can achieve the best school-running benefits with the minimum employment costs. Scientific staffing management and strict on-demand post setting can control and adjust the proportion and quantity of all kinds of personnel, can eliminate the disorder in management, and is also the key to improve the scientificity of university recruitment and achieve the organizational purpose with minimum consumption [14]. The establishment of institutions of colleges and universities involves all aspects of university work, mainly including the establishment of internal institutions and the determination of functions, the establishment and staffing of disciplines and professional posts, as well as the efficiency, effect and benefit in actual activities. In accordance with the spirit of the Guiding Opinions of the CPC Central Committee and the State Council on Promoting the Reform of Public Institutions by Classification (2011) and the Several Opinions of the Ministry of Education and other five departments on deepening the reform of streamlining administration, delegating powers, delegating powers and optimizing services in the field of higher education (2017), in order to promote the reform of the employment system of colleges and universities and further promote the scientific and comprehensive development of colleges and universities, in-depth discussion is made on the methods of staffing and posts suitable for the situation of colleges and universities and highlighting the characteristics of each college, Undoubtedly, it has important practical significance.
2 Research Methods The research methods used in this paper mainly include empirical research, comparative analysis and mathematical statistics.
1030
Z. Cui
2.1 Methods of Empirical Research Headcount management is a highly practical administrative practice. For the history and current situation of headcount management in colleges and universities, including the situation of institutions and personnel, the documents of headcount management, the headcount needs of departments and secondary units, etc., it is necessary to obtain detailed information and relatively accurate data through interviews, symposiums, questionnaires and other empirical research methods. 2.2 Methods of Comparative Analysis Human resource allocation and benefit control in colleges and universities are issues of universal concern to global higher education, and also the focus of practice and exploration in domestic colleges and universities in recent years [15–17]. Through document retrieval and data analysis, we can understand the historical evolution, current situation, management systems and methods of university headcount management in various countries, which can broaden our horizons and provide reference for the research of university headcount management [18–20]. At the same time, by comparing and analyzing their practices in organization setting, post division, staffing verification standards, and fund management, we explore the innovative thinking of staffing management. 2.3 Methods of Mathematical Statistics The logical starting point of staffing management is the objective evaluation and reasonable setting of workload, and how to measure the workload of teaching, scientific research and school affairs is a very complex problem. As the tested objects, such as teachers, their work will change randomly due to various internal or external conditions, and their work efficiency will also be randomly distributed. Therefore, mathematical statistics methods such as regression analysis, statistical mean, simple proportion, etc. are widely used in the analysis of research variables [21, 22]. The widely collected data can be used to study the fractional function and numerical characteristics of random variables as well as the relationship between various types of compilation, and then infer regular conclusions. This has been proved to be effective in the management of education around the world [23–25].
3 Prediction Results The student equivalent data of each college in recent years are shown in Table 1. It can be seen from the table that the change of student equivalent of each college in consecutive years presents a quadratic curve trend, so the exponential smoothing method can be used to predict. Among them, the cubic exponential smoothing method uses the weighted average of the observation values of each period in time order as the prediction value, showing the characteristics of the impact of the future value of historical data changing with time. This method is more suitable for the research content of this paper.
A Quantitative Study on the Categorized Management
1031
The prediction model of the cubic exponential smoothing method is:
yt+T = at + bt T + ct T 2 , T = 1, 2 . . .
(1)
where, ⎧ (1) (2) (3) ⎪ ⎪ ⎨ at = 3St − 3St + St (1) (2) α bt = [(6 − 5α)St − 2(5 − 4α)St + (4 − 3α)St(3) ] 2(1−α)2 ⎪ ⎪ ⎩ ct = α 2 2 [St(1) − 2St(2) + St(3) ]
(2)
2(1−α)
(1)
(2)
(3)
While, St , St , St is the first, second and third exponential smoothing value separately, and its calculation formula is: ⎧ (1) (1) ⎪ ⎨ St = αyt + (1 − α)St−1 (2) (1) (2) (3) St = αSt + (1 − α)St−1 ⎪ ⎩ (3) (2) St = αSt(2) + (1 − α)St−1
Table 1. Prediction of student equivalent trend in 2023 based on cubic exponential smoothing method College
Historical data of student equivalent
Exponential smoothing
Forecast trend
2016
2017
2018
2019
2020
2021
2022
Optimal value of α
2023
A
4420
4397
3869
3846
3814
3838
4062
0.990000
3837
B
2974
3028
3025
3051
3036
3058
4362
0.876947
3058
C
2196
2179
2111
3811
3828
3853
3899
0.971271
3853
D
3953
3854
3801
2923
3133
3283
3397
0.85211
3281
E
1408
1463
1440
2201
2172
2174
2511
0.939704
2175
F
2040
2087
2126
1755
1843
2027
3185
0.014609
2025
G
1732
1739
1736
3447
3058
3066
2051
0.77394
3085
H
1227
1221
1236
333
769
874
1902
0.45272
738
K
776
749
692
1288
1656
889
1137
0.346483
1249
Considering the known student data, the smoothing coefficient α Search and optimize within its value range. In the process of optimization, the mean square error of the predicted student equivalent is used as the optimization objective function. The expression of the objective function is as follows: n 1 (Xi − Yi )2 (4) σ= n i=1
1032
Z. Cui
In formula (4), Xi is the original student equivalent, Yi is the predicted student equivalent, and n is the number of original data. When the value of the smoothing coefficient α changes, there is a continuous functional relationship between the predicted student equivalent result error σ and the smoothing coefficient α, which can be expressed as σ = f (α)
(5)
The optimal smoothing coefficient α can be obtained by solving the following equation: f (α) = 0
(6)
In accordance with the formulas (1)–(6), the cubic exponential smoothing prediction equation is established, and the optimal values of the smoothing coefficients of each college are listed in Table 1. The student equivalent prediction results based on the optimal smoothing coefficients are displayed in Fig. 1.
Fig. 1. Predicted student equivalent of each college of a university in 2022
It can be clearly presented from the above figure that the predicted student equivalent of College A and College C in 2023 is relatively large. Combining the actual student equivalent data of these two colleges from 2016 to 2022, it can be found that when the smoothing coefficient α When it is close to 1, the weight of student equivalent in recent years in the predicted value is larger; The best smoothing coefficient of F College α When it is close to 0, the weight of student equivalent in the distant years in the predicted value is larger; When the best smoothing coefficient α When it is close to 0.5 (e.g., H College), the student equivalent in the middle year of the predicted value has a larger weight. Based on the above forecast data, in line with the “Regulations on the Management of Headcount of General Institutions of Higher Education (Draft)” jointly issued by the
A Quantitative Study on the Categorized Management
1033
Office of the Central Organization Headcount Committee, the Ministry of Education and the Ministry of Finance, the staffing of institutions of colleges and universities is divided into the staffing of teachers, the position of teaching and auxiliary staff, the personnel of full-time scientific research, the member of staff performing various management responsibilities, and the staffing of affiliated units including logistics and undertaking public welfare community services. It is also emphasized that all kinds of personnel in colleges and universities should form a reasonable proportion of the structure, among which teaching and research personnel and teaching and auxiliary personnel should account for more than 80% of the total number of college personnel. At the same time, it is stipulated that the total number of faculty and teachers should be determined considering the ratio of students to teachers. On the basis of the relevant standards of the Ministry of Education, combined with the actual situation of school management, based on the student equivalent data, the teacher preparation plan of each teaching unit can be calculated, as shown in Table 2. Table 2. Number of teachers according to the requirements of different student-to-teacher ratios College
Student equivalent
14: 1
16: 1
18: 1
20: 1
Number of teachers
Number of teachers
Number of teachers
Number of teachers
A
3837
274
240
213
192
B
3058
218
191
170
153
C
3853
275
241
214
193
D
3281
234
205
182
164
E
2175
155
136
121
109
F
2025
145
127
113
101
G
3085
220
193
171
154
H
738
53
46
41
37
K
1249
89
78
69
62
23301
1664
1456
1295
1165
Total
Taking the student equivalent predicted in Fig. 1 as the base, the teacher staffing plan of each college calculated according to the ratio of students to teachers is indicated in Fig. 2. Therefore, it can be seen that the number of teachers in each college in Fig. 2 is consistent with the trend of student equivalent in Fig. 1. Since the Ministry of Education has made clear quantitative requirements for human resources investment in the process of running colleges and universities in the undergraduate teaching evaluation program, defined the connotation and scope of the student-teacher ratio, and new innovations have also emerged in the staffing management of colleges and universities, making the human resources investment of colleges and universities meet the minimum standards to ensure the quality of education and teaching.
1034
Z. Cui
The data predicted by the exponential smoothing method lacks the ability to identify the turning point, i.e., if the student equivalent of a certain college has an obvious turning trend, the three-time exponential smoothing method cannot accurately predict the success, but this can be compensated by the survey prediction method or expert prediction method.
Fig. 2. The number of teachers predicted according to the standard student-teacher ratio of 14:1 required by the Ministry of Education
Moreover, the algorithm has strong adaptability and can automatically identify data patterns and adjust the prediction model. For the current situation that the ratio of students to teachers is too high in most colleges and universities, and the shortage of staff in some colleges and universities is serious, we should expand the team of full-time teachers, full-time counselors, and full-time scientific researchers through the introduction and appointment system, so that the ratio of students to teachers can be reduced from the current 20:1 to about 18:1 as soon as possible, laying the groundwork for the realization of the standard of 14:1 issued by the Ministry of Education in the next stage. The number of teachers in each college predicted in the light of the standard student-teacher ratio of 14:1 of the Ministry of Education is shown in Fig. 2.
4 Analysis and Suggestions With the continuous development of the reform of higher education, the reform of the staffing management in colleges and universities has also accelerated. The purpose of the establishment management of university teachers is to improve educational efficiency and meet the requirements of scientific management of education and teaching. At present, the common pursuit of all major universities in the reform of teacher staffing management is to improve the overall level of teachers and promote the healthy development of education. Among them, the main measures include the following three aspects.
A Quantitative Study on the Categorized Management
1035
4.1 Entry and Exit Mechanism of Teachers Based on the cubic exponential smoothing algorithm, this paper can predict the number of student equivalents of each college in the future, and then predict the number of teachers. In fact, in order to alleviate the shortage of teachers caused by the expansion of undergraduate and graduate enrollment, major universities have increased the introduction of teachers, and the number of new teachers has increased year by year. At the same time, the entry of non-teachers should be strictly controlled. The vacancy of non-teachers should be solved by internal adjustment as much as possible. For other urgently needed professional technicians and managers, they should be recruited to the public through personnel agency. There will be no new headcount for work posts, and try to gradually reduce such headcount through labor outsourcing and retirement. Through the implementation of the above measures, gradually adjust the proportion of students and teachers to make it moderate and reasonable. At the same time, the employment system of combining fixed and mobile headcount is implemented. The fixed headcount is steadily rising. The number of mobile headcount is increased, a relaxed employment atmosphere is created, and the employment environment is improved. Attention is paid to “career retention”, “emotional retention”, and “treatment retention”. Use scientific methods to manage teachers, fully mobilize and explore the enthusiasm and potential of teachers, form a good academic atmosphere and an academic environment of fair competition, explore and establish a relatively stable backbone layer and an orderly flow layer management mechanism. 4.2 Enrichment of Staffing Methods At present, the way of staffing in colleges and universities is relatively single, lacking scientific and reasonable staffing methods. Although the cubic exponential smoothing method in this paper can be used to predict the student equivalent, the relatively rough method of determining the ratio of students to teachers has gradually become inapplicable to the current level of university management and development. Various specific and quantifiable factors, such as teaching duration and quality, teaching content and effect, scientific research level and achievements, should be appropriately included in the process of making a comprehensive assessment. In addition, on the basis of the approval scheme of other types of post staffing approved by the executive unit, we began to make preparations for the major adjustment of the next stage of staffing, such as the fine-tuning of organizational structure, the sorting and optimization of post responsibilities, publicity and learning, skill training, and internal rotation, to improve the structure of university faculty. At this stage, the number of overstaffed positions is mainly frozen without adding new employees, so that the number of management, work attendance and other positions can be controlled, and retirement and job transfer can be added to maintain the state of natural reduction. 4.3 Establish and Improve Relevant Supporting Measures Some colleges and universities have made some innovative attempts to explore the management of teacher staffing, such as gradually abolishing the career staffing, implementing the employment system of labor contracts, or the personnel agency system. These
1036
Z. Cui
measures are opportunities and great challenges for current colleges and universities. The channels of talent flow are smooth to a higher degree, and excellent talents, especially high-level talents, are easily attracted to these colleges and universities. At the same time, the pre-employment and long-term employment system of “go up or down” makes the faculty lose its due stability, and reduces the sense of belonging and happiness of college teachers who take the mission of moral education as the core, In the long run, it will affect the realization and healthy development of the school’s talent training objectives. In fact, the management of teachers’ staffing should be reflected in the qualification, employment, assessment, evaluation and other links. By clarifying the professionalism of teachers, teachers can comprehensively promote their confidence in their own work and enhance their sense of responsibility and enterprising spirit. At the same time, colleges and universities should establish an open and transparent evaluation mechanism in line with their own development strategies to reduce the structural contradictions between the school staff, which can not only achieve the reasonable allocation of school resources, but also make talents in line with the sustainable development of the school stand out.
5 Conclusion Scientific and reasonable staffing verification and post setting management in colleges and universities is the key to building a reasonably structured, capable and energetic talent team in colleges and universities, and also the key to promoting the discipline construction in colleges and universities, which is of great significance to the overall development of colleges and universities. This paper uses the cubic exponential smoothing algorithm to predict the equivalent of future students, and then determines the number of teachers according to the ratio of students to teachers, which better realizes the combination of qualitative and quantitative, and improves the prediction accuracy of the number of posts in colleges and universities. On this basis, several suggestions and measures for the management of university teachers’ staffing are put forward. On the basis of in-depth analysis of the current situation of university teachers’ development and talent training objectives, the scientific allocation of teachers is carried out to provide assistance for the reform of university teachers’ staffing management system and the development of their own quality. Post staffing is a very complex system engineering, which is closely related to the future human resource allocation of the school and the vital interests of the existing staff. Scientific and reasonable post setting and effective post staffing management are of great significance to innovation of personnel management system in colleges and universities, optimization of human resource allocation, and mobilization of enthusiasm, initiative and creativity of faculty and staff. Therefore, in line with the idea of taking into account the current situation and long-term development of the team, the new round of post staffing optimization should be practical and realistic, and at the same time, the transition should be stable and gradually in place. Adopt a combination of qualitative and quantitative methods to implement staffing according to posts, optimize and adjust the structure of human resources based on the principle of “adjusting the internal structure and reaching consensus”, so that the structure of human resources tends to be more optimized and
A Quantitative Study on the Categorized Management
1037
reasonable, which not only promotes the development of key disciplines in colleges and universities, ensures the “double strength” of teaching and scientific research, but also achieves the goal of streamlining and improving management level.
References 1. Zhiqin, L., Jianguo, F., Fang, W., Xin, D.: Study on higher education service quality based on student perception. Int. J. Educ. Manage. Eng. 2(4), 22–27 (2012) 2. Singh, V., Dwivedi, S.K.: Two way question classification in higher education domain. Int. J. Mod. Educ. Comput. Sci. 7(9), 59–65 (2015) 3. Beltadze, G.N.: Game theory - basis of higher education and teaching organization. Int. J. Mod. Educ. Comput. Sci. 8(6), 41–49 (2016) 4. Hemei, L.X.: Research on the issue of post and staffing of management posts in colleges and universities. Yunmeng Acad. J. 36(2), 88–91 (2015). (in Chinese) 5. Lan, L.: Problems and countermeasures of post management of university teachers. Educ. Guide 11, 39–41 (2010). (in Chinese) 6. Wang, B., Zheng, Y., Jing, Y., et al.: On the reform and management of the establishment of institutions in colleges and universities. J. Handan Vocational Tech. College 34(4), 59–61 (2021). (in Chinese) 7. Cunhong, Y.: Promote the reform of teacher staffing policy and release the vitality of education human resources. Jiangxi Educ. 31, 14–15 (2019). (in Chinese) 8. Shi, L., Yao, Y., Xia, X., Xie, X., Lanlan, Wu.: Practical teaching staff construction under new situation. Int. J. Educ. Manage. Eng. 1(2), 57–61 (2011) 9. Meng, X.: Survey of research quality and ability of vocational school teachers. Int. J. Educ. Manage. Eng. 2(10), 9–16 (2012) 10. Kumar, S.: A fuzzy based comprehensive study of factors affecting teacher’s performance in higher technical education. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 5(3), 26–32 (2013) 11. Luis, R.: Understanding tenure reform: an examination of sense-making among school administrators and teachers. Teach. Coll. Rec. 122(11), 1–42 (2020) 12. Jiagan, D.: The characteristics of teacher staffing management from the perspective of human resources allocation in Japan. Comp. Educ. Res. 1, 56–58 (1999). (in Chinese) 13. Danan, W.: Discussion on the performance management of university teachers under the post appointment system. Liaoning Educ. Res. 2, 89–91 (2006). (in Chinese) 14. Maddin, B.W., Mahlerwein, R.L.: Empowering educators through team-based staffing models. Phi Delta Kappan 104(1), 33–37 (2022) 15. Wang, J., Pu, S., Wang, H., et al.: A tentative discussion on the classified management of university teachers’ posts – The experience of foreign first-class universities and the practice of Chinese universities. Journal of Sichuan Univ. (Philosophy and Social Sciences Edition) 2, 127–136 (2014). (in Chinese) 16. Paradise, L., Zhu, Y., Zhang, Y., et al.: Value identification of classified management of university teachers. Higher Eng. Educ. Res. 5, 59–64 (2015) 17. Wang, J., Pu, S.: Analysis on the classified management of university teachers: based on the perspective of modern university personnel management system reform. China Univ. Teacher Res. 1, 1–6 (2014). (in Chinese) 18. Jianmin, H.: An analysis of the space for improving the efficiency of human resources management in colleges and universities – The development of university staffing function from the perspective of the evaluation scheme’s student-teacher ratio index. China Higher Educ. Res. 1, 56–59 (2009). (in Chinese)
1038
Z. Cui
19. Lianzhang, Z.: Construction of human resource management system for teachers of Xi’an Foreign Affairs University. Lanzhou Jiaotong University, Lanzhou (2014). (in Chinese) 20. Gong, Y., Liu, D., Liu, W., et al.: Farewell to the old personnel management system - the record of Tsinghua University’s implementation of post appointment and post allowance system. China Higher Educ. 2, 4–7 (2000) 21. Lanping, L.: Quantitative evaluation of university teachers’ post and staffing. J. Gansu Union Univ. (Nat. Sci.) 26(2), 30–32 (2012). (in Chinese) 22. Zhu, X., Jiang, B., Chang, W., et al.: Discussion on staffing method of a large military hospital. J. Hosp. Manage. People’s Liberation Army 4, 373–375 (2005). (in Chinese) 23. Kotok, S., Knight, D.S.: Revolving doors: cross-country comparisons of the relationship between math and science teacher staffing and student achievement. Leadersh. Policy Sch. 21(2), 345–360 (2022) 24. Christopher, R., Tuan, N.: Recent trends in the characteristics of new teachers, the schools in which they teach, and their turnover rates. Teach. Coll. Rec. 122(7), 1–36 (2020) 25. Rasheed-Karim, W.: The Influence of policy on emotional labour and burnout among further and adult education teachers in the U.K. Int. J. Emerg. Technol. Lear. (iJET) 15(24), 232 (2020)
Course Outcomes and Program Outcomes Evaluation with the Recommendation System for the Students Khandaker Mohammad Mohi Uddin1(B) , Elias Ur Rahman1 , Prantho kumar Das1 , Md. Mamun Ar Rashid1 , and Samrat Kumar Dey2 1 Department of Computer Science and Engineering, Dhaka International University,
Dhaka 1205, Bangladesh [email protected] 2 School of Science and Technology, Bangladesh Open University, Gazipur 1705, Bangladesh
Abstract. Different kinds of educational systems have been applied in learning and teaching to maximize the slope of the learning curve of the students. One of those educational systems is the OBE system or Outcome-based education. Outcome-based methods have been adopted in education systems around the world, at multiple levels. In this education system, students and teachers have a clear view of what needs to be accomplished by the end of the course. However, in this education system, as many of the students are trying to get the same outcome, weak students are getting behind to achieve their goals. Hence creating inequity, the proposed work uses the prior student data to apply the Course Outcomes and Program Outcomes (CO-PO) attainment model and recommends the students to focus on their individual weaknesses. So as the students are studying through their courses, their results are recorded. Then the data of those students are used to train the application in our proposed work. Consequently, the application can identify pupils who need to learn their needed topics. As a result of this proposed method, students with different abilities can not only rely on the outcome, but also have to develop their required subjects to thrive in their learning process. Keywords: OBE · CO-PO · attainment model · Recommend system · Outcome
1 Introduction Goals are the focal point of each educational system element according to the educational paradigm known as outcomes-based education (OBE). By the end of the learning session, every student should have been able to complete the task [1]. There is no single specified technique of instruction or evaluation used in OBE; rather, all of the courses, opportunities, and exams should help students achieve the predetermined outcomes. The role of the faculty person may shift to include that of an instructor, trainer, facilitator, or mentor depending on the desired outcomes [2, 3]. Outcome-based methods have been adopted in education systems around the world, at multiple levels. Australia and South Africa adopted OBE policies from the 1990s to the mid-2000s but were abandoned in © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 1039–1053, 2023. https://doi.org/10.1007/978-3-031-36115-9_93
1040
K. M. M. Uddin et al.
the face of substantial community opposition [4, 5]. The United States has had an OBE program in place since 1994 that has been adapted over the years [6]. In 2005, Hong Kong adopted an outcome-based approach for its universities [7]. Malaysia implemented OBE in all of its public school systems in 2008 [8]. The European Union has proposed an education shift to focus on outcomes, across the EU. In an international effort to accept OBE, The Washington Accord was created in 1989; it is an agreement to accept undergraduate engineering degrees that were obtained using OBE methods [9]. As of 2017, the full signatories are Australia, Canada, Taiwan, Hong Kong, India, Ireland, Japan, Korea, Malaysia, New Zealand, Russia, Singapore, South Africa, Sri Lanka, Turkey, the United Kingdom, Pakistan, China, and the United States. The outcome-based education system embodies the commonsense thinking and practice of effective instructional design and delivery that can be recognized in highperforming learning systems across our society. In this approach, the outcomes of the students’ labor are more precisely targeted. Due to the fact that all students must be prepared for the demands of continuous learning and improvement in the Information Age job market, outcome-based models address a clear need in our society for learning systems that promote rather than limit the learning potential of all students. Although in OBE, the individuals who implemented the system in place decide the results. Even while the same results were claimed to have been attained, across programs or even amongst instructors, they could be perceived differently, resulting in a disparity in education. Assessments may become overly mechanical in assessing if an outcome has been met by just checking to determine if the student has learned the material. The assessment may not be primarily concerned with how well a student can use and apply knowledge in various contexts [10]. Students may never be taught how to apply the knowledge they have learned as a result of the emphasis on determining whether the outcome has been reached. Our research focuses on this concern with the outcome-based education system. The information from a class from a department’s course at a university is considered to be evaluated in this research. Our application will then use this data to determine which students need to increase their knowledge in their particular classes. As a consequence of this assignment, every student will be able to pinpoint their areas of weakness. All underachieving students will be helped to improve their courses by being made aware of their limitations.
2 Related Works In a study, Md Nujid et al. [11] evaluated the results of the Geotechnic course in the Bachelor of Engineering (Hons.) Civil program at the school of the department in a university. This study evaluates the result of the students and their survey. It shows that not everyone is correctly evaluated in OBE. The average COPOs percentages are 67 and 81. Lavanya et al. [12] in a study defined the true use of CO-POs in an Outcome-based education system. In this study, they also did the survey from the students and then they used that survey answers to find the attainment of the CO. To attain the PO they took the same approach. This study finds the CO and PO scores of each student and then they attain them. If the scores are below the threshold then the required step must be taken.
Course Outcomes and Program Outcomes
1041
This approach shows how the attainment of the CO and PO can help weak students to become better. Mawandiya et al. [13], in their study, went for the same approach. But in this study, the scoring of the CO and PO is very comprehensive. This study found that the CO attainment level for each student and for the entire class can be estimated giving ample scope for multiple possibilities of corrective actions. In a study, Sahar et al. [14] created a project named Quality Assurance System (QAS). This project is a web-based project where students, the dean of the faculty, and QAS Coordinators will have different views on the website. From this project, the students of the university can see their educational reports. Devasis Pradhan [15], in his study, showed how the Outcome-based education system can change the learning curve of the students. He showed that outcomes-based education can create students more skilled when susceptible to the real world. Shaikh et al. [16] created an application using Python programming language. This application is designed with staff and admin credentials where the admin can upload, edit and delete student lists for the current academic year and staff has the option of inserting data such as CO-PO matrix, target values, and direct and indirect assessment values. After accepting all values, the system generates CO and PO attainment sheet which shows whether the attainment has been achieved or not. This in turn can also help in designing the curriculum. Khwaja [17] built a web-based tool for assessing the CO-POs of the students. ASP.Net templates provided by Visual Studio are used for this project. With this application, the assessment of the results of the students can be easily done and the students of the institution can see their results. The UGC (University Grants Commission) has now mandated OBE-based education in Bangladeshi institutions. The main problem is that, if we try to accomplish this manually, it is a very challenging procedure to align the program outcomes with the course outcomes. The method by which this manual procedure can be automated using a special application is the theme of our research. We develop our online application, which utilizes the proposed optimum algorithms, to address this research problem.
3 Methodology Admin, Teacher, Respective Authority (Department Head or Dean), and Student are the four users of this system. Admins can edit courses, teachers, and CO/PO in addition to assigning CO/PO, COs, and teachers. The instructor will issue grades based on CO/PO, and the suggested method will then calculate and report the outcome. The outcome, which is known as a progressive report, will be produced for both the entire batch and each individual student. Our system’s main goal is to give each student and the entire batch a progress report so that the appropriate authorities can care for each student and the entire batch. The outcome appears as a pie chart. Students can only view their scores; whereas department heads and faculty deans can access batch-wise results individually. The created findings will all be available for download at the end, and each user will be able to obtain their individual progress report as a document file. The work flow of the proposed system is shown in Fig. 1.
1042
K. M. M. Uddin et al.
Fig. 1. Workflow diagram of the proposed system
The following task is carried out by users in the suggested system. • Admins can access the system by logging in and control all of its features. • Except for the outcome, the admin can add, modify, and delete all of the information and users. • The teacher will issue grades based on CO/PO. • The system will compute grades and produce a pie chart of the overall results. • Both batch-wise and individual outcomes can be observed by the appropriate authorities. • Students are only able to see their results and the system notifies them to improve their POs which are very important in the next semester. Every sort of user that uses our system must first log in. If they enter the proper username and password in the login section, they can access the system; if not, they are returned to the login page. The flowchart shows that the teacher can access their profile, see the courses they have been assigned, and can only give grades. Assigning courses, course teachers, batches, and CO/PO, on the other hand, is up to the admin. Algorithm 1 depicts an admin’s action in the suggested system. If an administrator tries to log in three times without success, their account will be deactivated.
Course Outcomes and Program Outcomes
1043
Algorithm 01: Algorithm for an admin 1. Begin 2. While i is equal to 1 to 3 do a. Input username of email_address and password b. If username of email_address and password match with the database i. Input choice ii. If choice is batch 1. assign_batch() 2. edit_batch_info() iii. Else if choice is course 1. assign_course() 2. edit_course() iv. Else if choice is teacher 1. add_teacher() 2. assign_teacher() v. Else if choice is CO-PO 1. add_CO_PO() 2. edit_CO_PO() 3. assessing_CO_PO() c. Else i. Enter username or email_address and password 3. End while 4. If the value of i is 4 a. Block_the_user() 5. End
3.1 Course Outcome (CO) The course outcomes effectively outline how each student would benefit from the course uniquely upon completion and how the same knowledge may be applied in the workplace and in academics after graduation. According to the curriculum, the university selects the many courses that students must complete within four years. Each course’s syllabus provides references to specific books, and before deciding on the course’s outcomes, the instructors study the prefaces of those books. The program outcomes are then matched to the course outcomes. That mapping establishes the connection between the program result and the course outcome. The four levels of the mapping are High (H), Medium (M), Low (L), and blank. Each CO is assigned to each PO independently if a course has three to twelve program outcomes. The CO determines what students should be able to do after they achieve the learning goal of a particular course. For each course, 3(three) to 4(four) CO are being used in our system. Every course has a unique CO. For instance, students who complete the CSE-407 (Artificial Intelligence & Neural Networks) course will be able to: • tackle significant challenges in the real world, integrate and apply networking analysis and artificial intelligence principles. • determine and examine various problems, then seek answers.
1044
K. M. M. Uddin et al.
• develop artificial intelligence-based solutions to a variety of issues. • prepare for local, regional, and global hackathons and project-based competitions. 3.2 Program Outcomes (PO) The POs in our system incorporate the knowledge, effectiveness, and abilities that graduates of the Computer Science and Engineering Program must have. The BSc in CSE program’s alumni will be qualified to: • Build on the understanding of problem-solving, teamwork, and communication skills obtained via the curriculum to advance as a computer science and engineering professional. • By continuing education, including graduate studies, practical qualifications, and licensure, engage in lifelong learning activities that advance their growth both professionally and personally. • Show your commitment to social justice, ethics, and leadership via your personal and professional contributions to the society. For our system, 12 (twelve) POs are predetermined [18–20]. The list of POs for our suggested system can be seen in Table 1. Table 1. List of POs in the proposed system PO 1
Engineering Knowledge
PO 2
Problem Analysis
PO 3
Design/development of solutions
PO 4
Investigation
PO 5
Modern tool usage
PO 6
The engineer and society
PO 7
Environment and Sustainability
PO 8
Ethics
PO 9
Individual work and teamwork
PO 10
Communication
PO 11
Project Management and Finance
PO 12
Life-long learning
Course Outcomes and Program Outcomes
1045
3.3 CO/PO Evaluation Process The system measures the percentage, pie chart, and bar graphs for the students in a course after receiving the total marks for the relevant CO. Course teacher will give the marks to the students for a specific course’s COs. Then the system calculates the percentage of the obtained marks of corresponding COs using an optimal algorithm. Though these CO’s are aligned with predefined POs, the system will evaluate the POs for the students after getting the marks of COs.
4 Design and Implementation We used a variety of tools to create our web application, choosing the ones we were most accustomed to working with. The development tools used to build this application are PHP (LARAVEL), HTML, TAILWIND CSS, VUE.JS, JAVASCRIPT, and MYSQL. 4.1 Modules of the System This app has four different types of users. Admins will be supreme users, teachers will only be capable of completing duties that have been allocated to them, and the other users will only be able to observe them. Admin, Teacher, Head of the Department (HoD)/Dean and student are the four different sorts of users of the proposed system. 4.2 Module Description Admin: The administrator will be able to perform superuser functions. In addition to assigning users and departments, he can also allocate courses to teachers and batch, semester, and co-op courses. Besides adding, updating, and deleting them, an admin can make whatever adjustments he wants. Teacher: The teacher is one of the few who makes use of this system. The teacher will be crucial in carrying out numerous system functions. They must first log onto this system using their name and an authorized password. After successfully signing in, he will be directed to the instructor panel page, as shown in Fig. 2. There, he will only see the courses he took and have the ability to input his grades for those courses (course, midterm, final, lab).
1046
K. M. M. Uddin et al.
Fig. 2. Teacher’s panel of the proposed application
When a teacher visits the instructor panel page, he can edit a student’s grades by selecting the “my course” option, then going to their batch course. The grades for every student in that batch will then be shown to him, along with an assign update option next to each student’s number that may be changed by clicking on it. One CO will satisfy one PO (program outcome), which is a set requirement, and each course will have four COs (course outcomes). The teacher can choose which CO in that course receives certain assignments or exam questions. Dean: Both the individual batch results and the total results will only be visible to DEAN. So that he may quickly assess each student’s performance on the CO-PO for each course and identify the poorer individuals. After successfully logging in, he will be taken to the dean panel shown in Fig. 3. He can view the course results for each batch both individually and collectively using the bar and pie charts in his dashboard.
Fig. 3. DEAN/Head of the department panel
Student: The CO-PO scores for the courses that make up their batch must be viewed by students. After successfully logging in, he will be directed to the student panel. He can see a bar chart that displays the course results for every student in his batch. He can also see a bar chart of the course results for the entire members of his batch.
Course Outcomes and Program Outcomes
1047
5 Result and Discussion The dean panel offers four choices. He has access to both the overall batch results and the weak student outcomes for each course in each batch. When he chooses the individual option for a course, the CO-PO result for that course will be displayed if he clicks on the result option next to them, as seen in Fig. 4.
Fig. 4. Individual CO-PO result
1048
K. M. M. Uddin et al.
As shown in Fig. 5, the CO-PO results for a course are presented even before the dean chooses the entire batch results option for that course.
Fig. 5. CO-PO result of a batch
Dean logs into the system, selects “poor student” from the menu bar, and then selects a department, batch, semester, and course before locating the batch’s worst student. Figure 6 contains a list of weak students. After successfully login in, the student will be sent to the student panel. Students can view a bar chart that displays each course’s outcomes, completed POs, and the percentage achieved. Student’s completed POs are shown as Fig. 7. Individual CO-PO result is shown as Fig. 8.
Course Outcomes and Program Outcomes
Fig. 6. Weak students’ list
1049
1050
K. M. M. Uddin et al.
Fig. 7. Student Completed PO
Fig. 8. Individual CO-PO Result
Course Outcomes and Program Outcomes
1051
6 Recommendation for the Students If a student receives a poor CO-PO score, he will be made aware of this CO PO, and moving forward, he must perform better in this course. Algorithm 2 demonstrates the logic behind recommending the courses the student should take to get excellent grades the upcoming semester and create standard POs.
Algorithm 02: Recommend the students based on POs percentages 1. Start 2. From session get stu_id, roll, semester_id 3. for i: 1 →12 // Total number of PO is 12 4. PO_wise_obtain_percentage( i, stu_id, roll) 5. for i: 1 →12 6. if PO_wise_obtain_percentage( i, stu_id, roll) is less than 40% 7. Notification_forPO (i) 8. end if 9. end for 10. end for 11. End Figure 9 Shows how notifications will be sent to students’ notification panels advising them to improve their CO-PO. They must raise their score after receiving the notifications.
Fig. 9. Notifications to a weak student to make good POs scores in the upcoming semester
1052
K. M. M. Uddin et al.
7 Conclusion The effective implementation of outcome-based education depends on the assessment of program outcomes (POs), and one of the most important instruments for achieving this is the analysis of course outcomes (COs). The proposed study applies the CO-PO attainment paradigm to previous student data and advises students to concentrate on their unique areas of weakness. The analysis helps the student identify their areas of weakness before they follow through with assignments, complete extra coursework, or enroll in other classes to strengthen their understanding and grade. This recommendation is put into practice for the particular course outcome that the student struggles with. Generally, this activity will aid in assessing the institution’s teaching and learning process, which will then aid in estimating the successes of several batches of a certain program. A mobile application will be developed in the future to improve the system’s usability.
References 1. Spady, W. G.: Outcome-Based Education: Critical Issues and Answers. American Association of School Administrators, ERIC, (1994) 2. Donnelly, K.: Australia’s adoption of outcomes based education: a critique. Issues Educ. Res. 17(2), 183–206 (2007) 3. Allais, S.M.: Education service delivery: the disastrous case of outcomes-based qualifications frameworks. Prog. Dev. Stud. 7(1), 65–78 (2007) 4. Austin, T.: Goals 2000: the Clinton administration education program. Retrieved. April. 4, 2005 (2005) 5. Borkar, H. G., Lakhandur, D. B.: OUTCOME-BASED EDUCATION 6. Kennedy, K.J.: Conceptualising quality improvement in higher education: policy, theory and practice for outcomes based learning in Hong Kong. J. High. Educ. Policy Manag. 33(3), 205–218 (2011) 7. Mohayidin, M.G., et al.: Implementation of outcome-based education in Universiti Putra Malaysia: a focus on students’ learning outcomes. Int. Educ. Stud. 1(4), 147–160 (2008) 8. Antunes, F.: Economising education: from the silent revolution to rethinking education. a new moment of Europeanisation of education. Eur. Educ. Res. J. 15(4), 410–427 (2016) 9. "Washington Accord”. International Engineering Alliance. Archived from the original on 26 January 2012. Retrieved 2 February 2012 10. Tam, M.: Outcomes-based approach to quality assessment and curriculum improvement in higher education. Qual. Assur. Educ. 22(2), 158–168 (2014) 11. Nujid, M.M., Tholibon, D.A.: Evaluation on academic performance of students in teaching and learning in engineering course. Asean J. Eng. Educ. 6(1), 33–39 (2022) 12. Lavanya, C., Murthy, J. N.: Assessment and attainment of course outcomes and program outcomes. J. Eng. Educ. Transformations, 35(4), 158–168 (2022) 13. Mawandiya, B. K., et al.: A new comprehensive methodology for evaluation of course outcomes and programme outcomes. J. Eng. Educ. Transformations, 36(1) (2022) 14. El_Rahman, S. A., Al-Twaim, B. A.: Development of Quality Assurance System for Academic Programs and Courses Reports. Int. J. Mod. Educ. Comput. Sci. 6, 30–36 (2015) 15. Pradhan, D.: Effectiveness of outcome based education (OBE) toward empowering the students performance in an engineering course. J. Adv. Educ. Philos. 5(2), 58–65 (2021) 16. Shaikh, H.M., Kumar, A.: Implementing an automated application for attainment calculations of program outcomes in outcome based education. J. Positive Sch. Psychol. 6(2), 6006–6016 (2022)
Course Outcomes and Program Outcomes
1053
17. Khwaja, A.A.: A web-based program outcome assessment tool. In: 2018 21st Saudi Computer Society National Computer Conference (NCC), pp. 1–6. IEEE, April (2018) 18. Chandna, V. K.: Innovative methodology for the assessment of Programme Outcomes. In: 2014 IEEE International Conference on MOOC, Innovation and Technology in Education (MITE), pp. 27–31. IEEE, December (2014) 19. Rajak, A., Shrivastava, A.K., Bhardwaj, S., Tripathi, A.K.: Assessment and attainment of program educational objectives for post graduate courses. Int. J. Mod. Educa. Comput. Sci. 11(2), 26–32 (2019) 20. Liu, Y., Zhang, X.: Evaluating the undergraduate course based on a Fuzzy AHP-FIS model. Int. J. Mod. Educ. Comput. Sci 12(6), 55–66 (2020)
Methodology of Teaching Educational Disciplines to Second (Master’s) Level Graduates of the “Computer Science” Educational Program Ihor Kozubtsov1(B) , Lesia Kozubtsova1 , Olha Myronenko1 , and Olha Nezhyva2 1 Kruty Heroes Military Institute of Telecommunications and Information Technology,
Kyiv 01011, Ukraine [email protected] 2 Taras Shevchenko National University of Kyiv, Kyiv 04053, Ukraine
Abstract. Purpose and objectives of the article. To substantiate the unified methodology of teaching the discipline of educational components to applicants of the second (master’s) level of higher education of the educational program “Computer Science” of the field of knowledge 12 Information Technologies of the specialty 122 Computer Science of full-time and part-time forms of study. To achieve this goal, the following tasks are set: 1. Analyze the current state of research and publications; 2. To substantiate the unified methodology of teaching the discipline of educational components to applicants of the second (master’s) level of higher education of the educational program “Computer Science” of the field of knowledge 12 Information Technologies of the specialty 122 Computer Science of full-time and part-time forms of education on the basis of a computer game. Research result. The unified methodology of teaching the discipline of educational components to applicants of the second (master’s) level of higher education of the educational program “Computer Science” of the field of knowledge 12 Information Technologies of the specialty 122 Computer Science of full-time and part-time forms of education on the basis of the teacher’s game is justified. Through the application of the reverse transformation of a student into a teacher, the educational and developmental goal of students acquiring primary educational quasi-professional teaching experience has been achieved. The scientific novelty lies in the fact that the first developed a unified methodology for teaching the discipline of educational components to applicants of the second (master’s) level of higher education of the educational program “computer science” on the basis of a game (gamified) approach interested in acquiring educational quasi-professional experience in teaching. Thus, the method proposed by the authors for teaching educational disciplines of educational components to students of the second (master’s) level of higher education of the educational program “Computer Science” of the field of knowledge 12 Information Technologies of the specialty 122 Computer Science provides a practical implementation of the theory of “anticipatory learning”. As a result of its application, a large reserve of time budget allocated for independent work will allow students to deeply study the educational material, prepare for a lecture class and, as a final result, acquire educational quasi-professional experience for future activities. In the case of applying © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 1054–1067, 2023. https://doi.org/10.1007/978-3-031-36115-9_94
Methodology of Teaching Educational Disciplines
1055
the methodology based on gamification, it allows to create such an informational and educational environment that contributes to the independent, active pursuit of higher education students to acquire knowledge, professional skills and abilities, such as critical thinking, the ability to make managerial (managerial) decisions, work in a team, be ready to cooperate; helps reveal abilities and motivates selfeducation. The methodology reveals excellent properties in the case of a mixed application of distance education and advance training, which will not only ensure the implementation of the curriculum for mastering the educational component, but also provide maximum opportunities for students to preserve life and health in war conditions. Keywords: teaching methods · academic discipline · methodology of scientific research · applicant · student · computer science
1 Introduction 1.1 Problem Statement According to the conditions of intensive growth in the amount of scientific and technical information, rapid change and updating of the system of scientific knowledge, there is a need for a qualitatively new theoretical training of future highly qualified specialists (master’s) level of higher education of the educational program “Computer Science” in the field of knowledge 12 Information technologies, specialty 122 Computer sciences. These qualified specialists will be capable of independent creative work, implementation of knowledge-intensive technologies in production and adaptation to the conditions of market relations. Knowledge of the methodology of scientific research, methods and organization of scientific research activity will help masters of the specialty 122 Computer sciences to easily get involved in professional activities, to translate scientific knowledge into a practical plane and will contribute to the development of rational and creative thinking. Scientific activity in higher educational institutions is an integral part of the educational process and is carried out with the aim of integrating scientific, educational and industrial activities in the higher education system. The Law of Ukraine “On Higher Education” (2014) [1] defines the main tasks of scientific activity in higher educational institutions, including: • organic unity of the content of education and programs of scientific activity; • direct participation of the subjects of the educational process in scientific and research works conducted in the higher educational institution; • organization of scientific, scientific-practical, scientific-methodical seminars, conferences, competitions of scientific research, coursework, diploma and other works of participants in the educational process. The COVID 19 pandemic created an unprecedented challenge to the education system, intensively prompted the use of distance education in institutions of higher education (HEIs) across the country, and became more active and the subject of widespread use [2]. The existing traditional methods of teaching do not allow providing high-quality
1056
I. Kozubtsov et al.
training. Rote teaching-learning methods are not as effective as far as the curriculum design in the computing field is concerned [3]. In this regard, there was a need to substantiate a unified author’s method of teaching the educational discipline of educational components to students of the second (master’s) level of higher education of the educational program “Computer Science” for the implementation of effective distance learning [4]. 1.2 Literature Review. Analysis of Recent Research and Publications Students’ and cadets’ motivation for traditional teaching methods began to fade paradoxically with the advent of computer technology. Game methods of teaching adults and gamification technologies in education came to the fore. These ideas were reflected in the publications of both foreign and domestic researchers. We will conduct a historical outline of the main works. In order to maximize learning in a constrained amount of time, the paper [5] investigates the application of active learning pedagogy. The potential for boosting student engagement and learning with active learning is quite promising. It is not a novel idea and has been encouraged and promoted since the 1980s. Active learning is used in classrooms by many professors due to its numerous advantages. Faculty are urged to examine their own pedagogical approaches and work to enhance them in order to better engage students and pique and hold their interest. Active learning has been used as a tool to pique students’ interest and ultimately boost learning, but it has rarely been used to directly affect learning in terms of time. The active learning approach used in this research is based on traditional pedagogies that were created using a variety of psychological theories of learning, motivation, and engagement. A survey of the students conducted following the use of several instances of this active learning technique showed an improvement in student learning. As the teaching and learning system is not a collection of knowledge that is packed in the mind, the development of the world today forces scholars in the area of education to examine the teaching techniques and tactics. The usage of educational teaching games is one of the most current developments. Games boost student motivation and assure engagement with instructional information, which provides a pleasant and engaging way to accomplish the intended goals. The writers of the article [6] make an effort to highlight the need of using games to enhance learning in practical ways and raise the bar of the educational process in a collaborative setting for both instructors and students. The suggested approach in this research was evaluated using a survey technique, and the findings are quite positive for academic professionals. The prototype of the development of the methodology of teaching certain disciplines in a game form was the justification of the methodology of teaching electrical engineering disciplines by the method of a virtual computer game [7], which was further developed in the concept of independent training of cadets of the Ground Forces on educational and training tools by the method of a game on a virtual computer [8]. Further practical steps in the gamification of teaching based on game methods and strategies of teaching and education can be found in the work of K. Kapp [9].
Methodology of Teaching Educational Disciplines
1057
V. Bugaeva rightly notes that gamification is an educational technology which is rapidly developing, having a huge potential to positively influence the effectiveness of the educational process [10, p.135]. From the point of view of our research, it is relevant for the preparation of future highly qualified specialists (master’s) level of higher education of the educational program “Computer Science” of the field of knowledge 12 Information Technologies of the specialty 122 Computer Science to consider gamification: as a way of forming the active professional behavior of future specialists in the IT industry [10]; as formal and informal space [11]; as an innovative pedagogical educational technology [12, 13] and learning technology in educational activity [14]. This confirms the opinion of N. Volkov that gamification is one of the trends of modern higher education [15]. 1.3 Highlighting Aspects that are not Sufficiently Studied The analysis of previous studies revealed that in modern pedagogy in terms of the development of this topic, there is no unified method of teaching educational disciplines of educational components to students of the second (master’s) level of higher education of the educational program “Computer Science” of the discipline. Based on this, the authors have chosen this current direction of research. 1.4 Purpose of the Article The purpose of the article is to substantiate the methodology of teaching the discipline of educational components to students of the second (master’s) level of higher education of the educational program “Computer Science” of the field of knowledge 12 Information Technologies of the specialty 122 Computer Science of full-time and part-time forms of education. 1.5 Research Objectives (Goals) To achieve the goal, the following tasks are set: 1. To analyze the current state of research and publications. 2. To justify the methodology of teaching educational disciplines of educational components to students of the second (master’s) level of higher education of the “Computer Science” educational program of the field of knowledge 12 Information technologies of the specialty 122 Computer sciences of full-time and part-time forms of education based on the teacher’s educational game.
2 Research Methods 2.1 Research Tools Basic research tools include methods of theoretical analysis and generalization of scientific literature on the topic of the research; generalization to formulate conclusions and recommendations for effectiveness.
1058
I. Kozubtsov et al.
2.2 Reliability and Accuracy of Results Reliability of the results of the study is ensured by the correctness of the use of mathematical apparatus and research methods. 2.3 Methodological Basis of the Study The object of scientific and theoretical research is not just a single phenomenon, a specific situation, but a whole class of similar phenomena and situations, their totality. The methodological basis of the research is the ideas of L. Vyhotskyi, P. Halperin, Y. Babanskyi, S. Rubinshtein, (pedagogical psychology), V. Bespalko, Y. Mashbits, (cybernetic approach in pedagogy, programmed learning and automated training systems). Thus, the basic theory for effective learning is proposed by the theory of “advanced learning” (M. Nechkina, 1984 [16]; S. Lisenkova, 1988 [17]) or the “inverted class” theory (J. Bergmann, A. Sams, 2012 [18]). If blended learning is backed by quality material, it may be successfully implemented. The topic, performance tasks, discussion forums, and quiz questions all make up good content if they are presented in an engaging and organized way. The ability of pupils in the cognitive, emotional, and psychomotor domains must also be evaluated using high-quality material [19].
3 Research Results 3.1 Theoretical Foundations of the Construction of Teaching Methods Scientific and research activities in higher education institutions of Ukraine are carried out on the basis of the current laws of Ukraine “On higher education” [1], “On scientific and scientific and technical activity” [20], statutes of higher education institutions and provide for the wide involvement of students to research work, enrichment of their knowledge with new scientific data, development of abilities for creative thinking, scientific analysis of phenomena, processes, which is a fundamentally important task not only for the department of “Computer Sciences”. The methodology of teaching the educational discipline of educational components to students of the second (master’s) level of higher education of the educational program “computer science” is based on the understanding of the students of higher education of the concept of “methodology” as a study of the organization of activities. The innovators of this point of view are leading scientists such as A. Novikov [21], L. Marakhovskyi, B. Sus, S. Zabara, I. Kozubtsov, I. Gevko, A. Stepaniuk, H. Tereshchuk, Yu. Khlaponin, etc. The combination of forms of conducting classes is a mandatory element of the successful assimilation of the educational material of the educational components of the students’ discipline of the second (master’s) level of higher education of the educational program “Computer Science” (the field of knowledge 12 Information Technologies of the specialty 122 Computer Science of full-time and part-time forms of education). It should be noted that currently the choice of pedagogical learning technologies is a key problem for the teacher – the subject of the educational process [22]. At the same
Methodology of Teaching Educational Disciplines
1059
time, the lack of recommendations regarding formalization in their selection expands the lecturer’s degree of freedom in creative search and experimentation. As can be clearly seen from Table 1 the following learning methods are the most accepted for the formation of knowledge: didactic games; practical training; teaching others (mutual learning) and independent work. Table 1. Comparative characteristics of different teaching methods Teaching methods
Solved tasks form knowledge
gaining skills thinking memory language experience
(lecture)
5%
++
–
–
–
++
reading
10% +
++
+
+
–
+
+
+
++
–
Practical Working with a 10% book (reading)
+
+
+
+
+
Educational discussions
50%
++
–
++
+
++
++
Didactic games
70%
++
–
++
+
++
++
Practical training
75%
+
++
++
+
–
++
Teaching others is the application of what has been learned
90%
++
–
++
+
++
++
80%
++
++
++
+
+
++
–
+
+
++
Verbal
Visually
listening
20%
Work with multimedia (audiovisual)
20%
Viewing the drawing
30%
Demonstration 30% Video viewing
Independent work Oral and written control
develop
50%
Note: ++ - solve very well; + - solve partially; – - solve poorly
The author’s method of teaching the educational disciplines of the educational components provides for the creation of positive pedagogical conditions for students of the second (master’s) level of higher education of the educational program “Computer Science” (the field of knowledge 12 Information Technologies of the specialty 122 Computer
1060
I. Kozubtsov et al.
Sciences of full-time and part-time forms of education) to encourage them to acquire the primary practical experience in the process of educational quasi-professional activity. Educational quasi-professional activity enriches students with the ability to direct them to self-development in the direction of future professional activity, which includes: management (managerial) activity; scientific (research) activity; pedagogical (teaching) activity. Educational quasi-professional management (managerial) activity is formed in students through a responsible attitude to the self-organization of independent work. Independent work is the main means of mastering educational material in the time free from standardized educational classes, that is, lectures and practical classes (auditory work). During independent work, students should pay attention to: work on processing and studying the recommended literature; preparation for discussions and other tasks proposed by the teacher; work on an abstract (educational article, theses, report); work on an individual research project, etc. The methods of conducting classes in the educational disciplines of the educational components with indications of approximate time are given in Table 2. Table 2. Methods of conducting classes Type of lesson
The structure of the lesson Introductory part
Main part
Final part
t’
t’
t’
method
main
additional
Methodical support
method
Lecture
10
C
EL
S, Cl, D
5
Practical class
20
C
E
IW
10
C
Methodical development. Software tasks
Independent work
–
–
IW
–
–
–
Methodical development. Tasks in the IW
–
The text of the lecture
3.2 Peculiarities of the Methodology, Application of Teaching Methods and Tools Features and methods of conducting classes with indications of estimated time are given in Table 2. The table uses the following abbreviations: C – conversation; EL – educational lecture; S – story; Cl – clarification; D – demonstration; DS – display; E – exercise; IW – independent work. The introductory lecture is conducted by the leading lecturer, one of the professors. The goal, task structure of the educational discipline is explained to the students at this lecture. Besides, it is justified in the professional need for its study by future teachers and scientists in the field of Information technologies.
Methodology of Teaching Educational Disciplines
1061
The lecturer must practically show the method of conducting the lecture, demonstrate at the highest level his/her own pedagogical skill, methodological culture, which should be reflected in the students as a certain standard to which one must go and try to exceed over time by forming his/her own pedagogical skill and methodological culture. The professional activity of students as future scientific and pedagogical workers begins with the scientific understanding of the object, subject and purpose of the work and the final result. Besides, the very course of work is a project. For students of the specialty 122 Computer science, in most cases, the presentation is interesting in a game form with an illustrative example in the form of an algorithm of the reporting stages of the academic discipline [23]. After a few lectures (these are usually 4–6 lectures depending on the methodology), it is suggested to actively involve students in the role of future lecturers (teachers). This pedagogical method gives students the opportunity to overcome their own fear of the audience, to try to realize themselves in a quasi-professional educational activity. In this case, we can talk about the pedagogical component. To do this, the “full-time lecturer” on the eve of the lecture sets students tasks for independent work through the head of the study group, determines the topics of the lesson, educational questions that must be worked out by students. Students who are preparing for lectures, study information search, analysis, validation, generally gain practical experience in scientific activities, etc. The technology of preparing and conducting lectures by students was worked out in the context of a scientific and methodical seminar at the department [24]. Incoming reports, the lecturer observers the practical phase of the development of professional experience, methodological culture of the student. The lecturer’s interaction with students in the process of lecture-scientific-methodical support is based on the model of subject-object interaction. According to the educational component of the scientific and methodological support of students, the theoretical basis is Disterweg’s fundamental idea. Disterweg notes that a bad teacher presents the truth, a good one teaches to find it [25, p. 161]. This idea emphasizes the special importance of creating such an educational atmosphere in which the student himself/herself will begin to find answers to questions. To realize this truth, the principle of independent work of students in the technology of scientific and methodological support should be followed, and the interaction between the lecturer and students according to the algorithm [26], directing its further development and interpreting to solve the task of acquiring quasi-practical experience. If necessary, the full-time lecturer manages and makes corrections in the pedagogical process. Moreover, this lecturer determines the rotation of the speaker to another student with the continuation of the report on the essence of the educational issue. Thus, the principle of attentiveness of listeners is realized. 3.3 Discussion of Research Results The lecturer’s work methods and students’ activities are based on the use of pedagogical technologies such as project; programmed training; problem-based learning by transforming ideas [27–30].
1062
I. Kozubtsov et al.
The method of teaching educational disciplines of educational components involves assessing the level of formation of the applicant (master’s degree) as a future teacher in the field of Information technologies (Computer sciences). Approximate evaluation points are given in Table 3. Table 3. Assessment of the level of teaching by the recipient Characteristics
Implementation in the activities of the lecturer
Formulation of the topic and definition of the sufficiently clear and understandable for the goal students Somewhat blurred They remained unclear to the students Plan and structure
The structure is clear, meaningful parts are highlighted and interconnected The general plan is defined, but the transitions from one semantic part to another remain unclear
Content
It is difficult to follow the development of the topic, the main ideas were expressed inconsistently
The ability to arouse interest in the topic
Theoretical positions were argued, supported by facts and examples Scientific, evidential, but very difficult to understand Very popular, empirical material prevailed The meaning of the topic is revealed convincingly, the material was connected with the personal experience of the student Only the need to study the topic was indicated, examples were used, there was no reliance on the personal experience of the students The importance of studying the topic was not motivated in any way, the material was not supported by examples
Problem statement
The lecturer drew attention to contradictions, formulated problematic issues, and encouraged the audience to discuss The lecturer formulated problematic questions and answered them himself/herself The lecturer expressed the theoretical material as something universally recognized, which does not require proof (continued)
Methodology of Teaching Educational Disciplines
1063
Table 3. (continued) Characteristics
Implementation in the activities of the lecturer
Contact with the audience
The contact was complete, all students worked, the lecturer took into account “feedback” Sometimes the lecturer lost contact with the audience and the students started to get distracted The lecturer failed to establish contact with the audience and take into account “feedback”
Culture of the lecturer’s language
Clear diction, optimal pace of speech, emotional presentation The diction and pace of speech are normal, but the emotional component was missing There were problems with diction, language pace, emotionality of presentation
The behavior of the lecturer
He/she held himself /herself confidently, reasoned freely on the topic, did not use the lecture notes He/she held himself/herself confidently, but he/she did not possess the skill of fluent speech, he/she relied on the lecture notes He/she kept himself/herself rigid, practically did not tear himself/herself away from the text of the lecture
Use of visual aids
Graphic methods of presenting the material and other visual aids were actively used A blackboard was occasionally used There were no visual aids
Good conclusion
The conclusion logically concluded and summarized what was presented The conclusion was unclear There was no conclusion
A large reserve of time budget allocated for independent work will allow students to deeply study the educational material, prepare for the lecture and, as a final result, acquire educational quasi-professional experience for future activities [16–18]. This is confirmed by the results of the study. Student success in studies depending on the time spent on extracurricular activities [31]. We definitely agree with the opinion of the authors that to improve teaching and learning, it is necessary to use assistive technologies [32]. For example, the use of gamification, but in such a way that the method of learning through the game does not turn into a pure game [9, 11, 12, 22].
1064
I. Kozubtsov et al.
4 Summary and Conclusion Thus, the method proposed by the authors for teaching educational disciplines of educational components to students of the second (master’s) level of higher education of the educational program “Computer Science” of the field of knowledge 12 Information Technologies of the specialty 122 Computer Science provides a practical implementation of the theory of “anticipatory learning”. As a result of its application, a large reserve of time budget allocated for independent work will allow students to deeply study the educational material, prepare for a lecture class and, as a final result, acquire educational quasi-professional experience for future activities. In the case of applying the methodology based on gamification, it allows to create such an informational and educational environment that contributes to the independent, active pursuit of higher education students to acquire knowledge, professional skills and abilities, such as critical thinking, the ability to make managerial (managerial) decisions, work in a team, be ready to cooperate; helps reveal abilities and motivates self-education. The methodology reveals excellent properties in the case of a mixed application of distance education and advance training, which will not only ensure the implementation of the curriculum for mastering the educational component, but also provide maximum opportunities for students to preserve life and health in war conditions. 4.1 Expanding the Boundaries of the Scientific Field The scientific result obtained by the authors expands the boundaries of pedagogical sciences in the part related to the scientific specialty “teaching theory and methodology”. 4.2 Scientific Novelty. Scientific Justification For the first time, a methodology was developed for teaching educational disciplines of educational components to students of higher (master’s) education majoring in “Computer Science” educational program of the field of knowledge 12 Information technologies major 122 Computer science full-time and part-time forms of study based on a game approach, interested in acquiring educational quasi-professional experience pedagogical activity for with the possibility of its application in war conditions due to mixed education. 4.3 Practical Use The proposed technique is fully ready for practical application as advanced anticipatory learning. Since the education system of Ukraine from February 24, 2022, like the entire country, functions in extreme conditions, it is necessary to adapt to the reality of war, during which it is necessary to continue the educational process of training students of higher education for the needs of the national economy. Thus, the use of distance education and advanced training will not only ensure the implementation of the curriculum for mastering the educational component, but also provide maximum opportunities for students to preserve life and health in war conditions.
Methodology of Teaching Educational Disciplines
1065
For the first time, lecturer I. Kozubtsov at the Department of “Computer Sciences” of the Lutsk National Technical University developed a methodology for teaching educational disciplines of educational components to students of the (master’s) level of higher education of the “Computer Science” educational program of the field of knowledge 12 Information technologies of the specialty 122 Computer sciences of full-time and parttime forms of education based on a game approach, interested in acquiring educational quasi-professional experience in teaching activities. 4.4 Prospects for Further Research and Study The research is expected to bring more academic and applicable value. The methodology needs further improvement if it is necessary to conduct laboratory or practical classes on the use of stationary equipment of higher educational institutions and it is not possible to practice it at home. Acknowledgment. The authors would like to express their respect to the organizers of the 3rd International Conference on Artificial Intelligence and Logistics Engineering (ICAILE2023) (March 11–March 12, 2023, Wuhan, China) and at a tragic time for Ukrainian researchers, Modern Education and Computer Science Press provided a grant to publish their scientific achievements free of charge. We are sincerely grateful.
References 1. Law of Ukraine. “On higher education” No.1556-VII (2014). http://zakon4.rada.gov.ua/laws/ show/1556-18 (in Ukrainian) 2. Schwab, K., Malleret, T.: COVID-19: The Great Reset. Edition 1.0. Switzerland. Cologny/Geneva: Forum publishing World Economic Forum. 2020 3. Churi, P., Rao, N.T.: Teaching cyber security course in the classrooms of NMIMS university. Int. J. Mod. Educ. Comput. Sci. 13(4), 1–15 (2021) 4. Kozubtsova, L.M., Kozubtsov, I.M.: On the problem of organizing effective distance learning. The First International Scientific and Practical Conference “Social aspects of Military Professional Activity of the Security and Defense Sector: Challenges of our Time”: collection of abstracts, (Kharkiv, 20.05). National Academy of the National Guard of Ukraine, pp. 284–286 (2021). (in Ukrainian) 5. Khan, A.A., Madden, J.: Speed learning: maximizing student learning and engagement in a limited amount of time. Int. J. Mod. Educ. Comput. Sci. (IJMECS). 8(7), 22–30 (2016). https://doi.org/10.5815/ijmecs.2016.07.03 6. Albilali, A.A., Qureshi, R.J.: Proposal to teach software development using gaming technique. Int. J. Mod. Educ. Comput. Sci. (IJMECS). 8(8), 21–27 (2016). https://doi.org/10.5815/ijm ecs.2016.08.03 7. Kozubtsov, I.N.: Teaching electrical engineering disciplines by virtual computer game method. In: Electrical Technologies, Electric Drive and Electrical Equipment of Enterprises: Collection of Scientific Papers of the Second All-Russian Scientific and Technical Conference, vol. 2, pp. 107–110. Ufa: USNTU Publishing House (2009). (in Russian) 8. Kozubtsov, I.M.: The concept of independent training of cadets of the ground forces on training facilities by playing on a virtual computer. In: Prospects for the Development of Weapons and Military Equipment of the Ground Forces. Second All-Ukrainian Scientific and Technical Conference, p. 77 (2009). (Lviv, April 28-29)
1066
I. Kozubtsov et al.
9. Kapp, K.: The Gamification of Learning and Instruction Game-Based Methods and Strategies for Training and Education. Pfeiffer, San Francisco, USA (2012) 10. Bugaeva, V.: Gamification as a way of forming active professional behavior of future IT industry specialists. Pedagogy Psychol. 56, 129–135 (2018) 11. Tkachenko, O.: Gamification of education: formal and informal space. Top. Issues Humanit. 11, 303–309 (2015) 12. Petrenko, S.: Gamification as an innovative educational technology. Innov. educ. 2(7), 177– 185 (2018) 13. Nezhivaya, O.: Innovative technologies in the educational process. In: Innovative Trends in Training Specialists in a Multicultural and Multilingual Globalized World: a Collection of Abstracts of Reports of the V All-Ukrainian Scientific Society-Practical Conference (Kiev, April 07). Kiev: KNUTD, pp. 74–77 (2020). (in Ukrainian) 14. Noskov, E.A.: Learning technologies and gamification in educational activities. Yaroslavl Pedagogical Bull. 6, 138–142 (2018). (in Russian) 15. Volkova, N.P.: Gamification as one of the trends of modern higher education. In: Modern Higher Education: Problems and Prospects: VI All-Ukrainian Scientific and Practical Conference of Students, Postgraduates and Scientists: Abstracts of Reports (Dnipro, 22.03). 2018: 33–35 (in Ukrainian) 16. Nechkina, M.: Increasing the effectiveness of a lesson. Communist. 2, 51 (1984) 17. Lysenkova, S.M.: The Method of Anticipatory Learning: a Book for Teachers: From Work Experience. Enlightenment, Moscow (1988) 18. Bergmann, J., Sams, A.: Flip Your Classroom: Reach Every Student in Every Class Every Day. International Society for Technology in Education, Washington, DC (2012) 19. Putu Wisna Ariawan, I., Divayana, D.G.H., Wayan Arta Suyasa, P.: Development of blended learning content based on Tri Kaya Parisudha-superitem in Kelase platform. Int. J. Mod. Educ. Comput. Sci. 14(1), 30–43 (2022) 20. Law of Ukraine “On scientific and technical activities”, No. 848-VIII (2015). http://zakon3. rada.gov.ua/laws/show/848-viii (in Ukrainian) 21. Novikov, A.M., Novikov, D.A.: Methodology. SINTEG, Moscow (2007). (in Russian) 22. Lishchina, V., Kozubtsov, I., Kozubtsova, L.: Choice of pedagogical training technologies as a key problem teacher – subject of the educational process. In: International Scientific and Methodological Conference “Innovative Technologies in Military Education”, (Odessa, June 25). Odessa: Military Academy, pp. 225–226 (2021). (in Ukrainian) 23. Kozubtsov, I.M.: Method of virtual cognitive presentation of reporting stages of an academic discipline to cadets. In: Fifth Scientific and Technical Conference “Priority Areas for the Development of Telecommunications Systems and Special-purpose Networks”, (Kiev, October 20–21). VITI NTUU “KPI”, pp. 144–147 (2010). (in Ukrainian) 24. Kozubtsov, I.N., Kozubsova, L.M.: Pedagogicheskaya technology organization nauchnomethodicheskogo workshop chair. Pedagogical skills. Theoretical and scientific-methodical Journal. O‘zbekiston Respublikasi Buxoro davlat universiteti. 1, 24–30 (2016) 25. Disterveg, A.: A guide to the education of German teachers. In: Selected Pedagogical Essays, pp. 136–203 (1956) 26. Mayer, R.V.: Cybernetic pedagogy: simulation modeling of the learning process. Glazov: GGPI, p. 138 (2013). (in Russian) 27. Trubavina, I., Kaplun, S.: Subjectivity of students as a pedagogical condition for the formation of their cognitive independence in learning. In: Fundamental and Applied Research: Modern Scientific and Practical Solutions and Approaches: Proceedings of the Fifth International Scientific and Practical Conference. National Academy of Sciences of Azerbaijan, vol. 5, pp. 288–292 (2019)
Methodology of Teaching Educational Disciplines
1067
28. Trubavina, I., Kaplun, S.: Complex of conditions for applying training as a form of organizing training for students of higher educational institutions of Ukraine. Problèmes et perspectives d’introduction de la recherche cognique innovante: collection de papiers cogniques ´ “OGO” avec des matériaux de la conférence diagnosique et pratique internationale, (Bruxelles, novembre 29). Plateforme scientifique européenne, vol. 5, pp. 21–23 (2019) 29. Trubavina, I.: Formation of skills of independent work among students by methods of problembased learning in a modern university. Bulletin of Luhansk Taras Shevchenko National University. Pedagogical Sciences 7(312), 2: 116–125 (2017). (in Ukrainian) 30. Nezhyva, O.: The aspects of smart education in the world. Khazar J. Humanit. Soc. Sci. 24(3), 62–72 (2021) 31. Sharma, N., Appukutti, S., Garg, U., Mukherjee, J., Mishra, S.: Analysis of student’s academic performance based on their time spent on extra-curricular activities using machine learning techniques. Int. J. Mod. Educ. Comput. Sci. (IJMECS). 15(1), 46–57 (2023). https://doi.org/ 10.5815/ijmecs.2023.01.04 32. Adebayo, E.O., Ayorinde, I.T.: Efficacy of assistive technology for improved teaching and learning in computer science. Int. J. Educ. Manage. Eng. 12(5), 9–17 (2022). https://doi.org/ 10.5815/ijeme.2022.05.02
Professional Training of Lecturers of Higher Educational Institutions Based on the Cyberontological Approach and Gamification Oleksii Silko, Lesia Kozubtsova, Ihor Kozubtsov(B) , and Oleksii Beskrovnyi Kruty Heroes Military Institute of Telecommunications and Information Technology, Kyiv 01011, Ukraine [email protected]
Abstract. The subject of research in the scientific article was the theoretical and practical foundations of applying the cyberontological approach in the professional training of lecturers. To achieve the research purpose, the following tasks were solved: the current state of research and publications in the area was analyzed; theoretical, practical bases for applying the cyberontological approach in the professional training of lecturers were developed. Cyberontological approach is proposed as an innovative basis of psychological and pedagogical science-cyberpedagogy, which is designed to generalize and systematize scientific knowledge in the field of application of modern information and communication, computer, digital, electronic and Internet technologies in the education system. The paper substantiates the cyberontological model of functional dependence of components and the flow of the information processes in the educational activity of a lecturer-student. Based on the accumulated positive experience of using computer technologies in teaching students, the use of a cyberontological approach in the professional training of lecturers is proposed. Several examples of applying a cyberontological approach with elements of gamification in the training of cybersecurity specialists are considered. The scientific significance of the work is that for the first time a functional cybernetic model of the dependence between the components of the pedagogical system has been developed. The practical significance of its application in practice is that on the basis of the cybernetic model, game mechanics can be developed to increase student motivation in modern conditions. It is possible to use a computer game (gamification) as the basis for building a training complex for training military information and cybersecurity specialists. Keywords: Cyberontological approach · professional training · lecturer · ontology · gamification · higher education institution · cybersecurity · mathematical model
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 1068–1079, 2023. https://doi.org/10.1007/978-3-031-36115-9_95
Professional Training of Lecturers
1069
1 Introduction 1.1 Problem Statement Both around the world and in our country, a person learns throughout his life. This is training in a general academic school, vocational school, or higher education institution. The state spends a lot of money on financing the education system, so the problem of improving the efficiency of the learning process is urgent. Its optimization requires not only improving the content and methodology of studying individual subjects, but also developing the theoretical foundations of didactics with the involvement of both the humanities (for example, Psychology) and exact sciences (Mathematics, Cybernetics). Now all over the world, when analyzing the educational process, the consideration of the “lecturer – student” system from the point of view of management theory is based. Almost all educational institutions have shifted their academic activities to digital platforms due to the recent COVID-19 epidemic [1]. Given the above, the cybernetic approach is also relevant for the higher education system. 1.2 Literature Review. Analysis of Recent Research and Publications In the scientific research of V. Pleshakov [2, 3] formed and acquired further formation in the works of N. Voloshyn, O. Mukohorenko, L. Zhohin [4], K. Meteshkin, O. Morozov, L. Fedorchenko, N. Khairov [5] the definition of cyberontology as the existence and / or life activity of a person in an innovative alternative reality of cyberspace (cyberreality), determined by the level of development of self-consciousness and motivational needs of the individual, as well as by the complex of objective and subjective micro-, macro-, meso- and megafactors of society. Based on the analysis of human interaction with computer, electronic, digital technology, a cyberontological approach to education is built, according to which, training and upbringing of a person is determined by the conditions of its existence, life activity and interaction with computer technology, with other people and the world as a whole, based on the integration of two spaces: real and virtual. According to V. Bespalko [6], N. Voloshyn [4], A. Ihibaev, A. Toleukhanov [7], I. Kozubtsov [8], R. Maier [9], V. Pleshakov [10], the cyberontological approach is nothing more than a phenomenon of psychological and pedagogical science. It begins the foundations of cyberpedagogy, which is designed to generalize and systematize scientific knowledge in the field of application of modern information and communication, computer, digital, electronic and Internet technologies in the education system. 1.3 Highlighting Aspects that are Understudied The analysis of recent studies has established that the chosen object of study “cyberontological approach” attracted the attention of foreign scientists, but the description of the functional cyberontological model of dependence and the flow of the information processes in human educational activities is not sufficiently reflected. Based on the above, this current research area was chosen.
1070
O. Silko et al.
1.4 Purpose of the Article To review the theoretical foundations and practical success in applying the cyberontological approach with elements of gamification in the professional training of lecturers of higher educational institutions (HEI). 1.5 Research Objectives (Goals) To achieve this purpose, the following objectives are set: 1. Analyze the current state of research and publications on the keywords of the terms “cyberontological approach”, “cyberpedagogy”. 2. To review the theoretical foundations and practical success in applying the “cyberontological approach” in the professional training of future lecturers of the HEI.
2 Research Methods 2.1 Research Tools The main means of scientific and theoretical research. A set of scientific methods that are comprehensively substantiated and consolidated into a single system. Basic research tools: methods of theoretical analysis and generalization of scientific literature, on the topic of the research; generalization to formulate conclusions and recommendations for effectiveness. 2.2 Reliability and Accuracy of Results Reliability of the results of the study is ensured by the correctness of the use of mathematical apparatus and research methods. 2.3 Methodological Basis of the Study The object of scientific and theoretical research is not just a single phenomenon, a specific situation, but a whole class of similar phenomena and situations, their totality. The methodological basis of the research is the ideas of L. Vyhotskyi, P. Halperin, Y. Babanskyi, S. Rubinshtein, (pedagogical psychology), N. Viner, K. Shennon, F. Rozenblatt, A. Kolmohorov, V. Hlushkov, (cybernetics), V. Maier, D. Novikov, (mathematical modeling of learning), V. Bespalko, Y. Mashbits, (cybernetic approach in pedagogy, programmed learning and automated training systems). Thus, the basic theory for mathematical modeling is cybernetics.
Professional Training of Lecturers
1071
3 Research Results 3.1 Theoretical Foundations of the “Cyberontological Approach”, the “Cyberpedagogy” Component With the advent of “cyberspace”, a virtual dimension, a new environment of life activity and a factor of social change were formed (M. Chitosca [11], S. Petriaiev [12]). As a result, modern person lives and interacts with other people and the world as a whole in parallel in two socializing environments – classical objective (material) reality and alternative innovative reality of cyberspace (cyberreality) – that potentially and objectively affect the formation and transformation of subjective (phenomenon of the psyche) reality. In this regard, it is advisable to talk about the “parallel” existence of a person in cyberspace, about an alternative ontology of modern human civilization – cyberontology. The types of indirect activities have become the main vectors of socialization (integration) of a person with cyberspace. According to some scientists O. Voinov [13], N. Voloshyn [4], V. Pleshakov [10], S. Petriaiev [12]) in particular: 1. 2. 3. 4.
Communication in cyberspace. Leisure in cyberspace. Knowledge in cyberspace. Work in cyberspace.
To effectively solve the actual problems of human education and training, the need to apply a “cyberontological approach” in education and the beginning of an innovative branch of pedagogical science – cyberpedagogy (V. Bespalko [6], N. Voloshyn, L. Zhohin, O. Mukohorenko [4], K. Meteshkin, O. Morozov, L. Fedorchenko, N. Khairov [5]) is justified. At the same time, a new fifth vector of human socialization into education through cyberspace was identified. So, the fifth vector is the education of a person in cyberspace – this is the process of human education, which is determined by the conditions of his life activity and interaction with oneself, with other people and the world as a whole in the context of the integration of classical objective (material) reality and innovative alternative reality of cyberspace (cyberreality), both of which potentially and objectively affect the formation and transformation of subjective (phenomenon of the psyche) reality. The main role of the cyberontological approach is to regulate the development of personality and human life activity in cyberspace, taking into account modern conditions and trends of education, as well as near and far prospects for the evolution of mankind. Nowadays, the use of cyberspace as an educational environment, a channel of training and educational communications has become commonplace. The human educational process in cyberspace is interdisciplinary, as a result of which “Cyberpedagogy” shows positive and negative trends in education [14]. It was not without reason that the author Y. Mashbits expressed concern at the time: “will the widespread use of the computer lead to the oblivion of the past – our roots, to a less defined future, to the detriment of the cultural and spiritual values of our heritage? The habit of systematically using a computer can make a person neglect his own capabilities, lead him to excessive dependence on the computer, cause atrophy of thinking, isolating a person from the world around them” [15].
1072
O. Silko et al.
A similar opinion was held by (V. Bespalko [6]), noting that a computer cannot effectively teach, relying on “human”, expressed in verbal form, pedagogy that is understandable to a living lecturer who has an innate pedagogical intuition and an understanding of polysemous natural language. The computer needs special pedagogy, expressed in the unambiguous language of mathematics and formal logic, and which describes welldefined rules of action (algorithms) in well-defined pedagogical situations (tasks). The author offers the basics of such pedagogy “Cyberpedagogy” (from cybernetics and pedagogy). Cybernetics (V. Hlushkov) refers to the science of managing complex technical, biological, and social systems that can perceive, store, and process information. From the point of view of cyberpedagogy, the processes of teaching and upbringing can be reduced to managing the development of various personal qualities of students through purposeful and coordinated influences on the part of lecturers and parents (R. Mayer [9]). The purpose of training is to transfer a set of knowledge to students, to form skills and abilities, to develop their ability to observe, reason and effectively interact with the world around them. The main directions of “cyberpedagogy” according to (K. Meteshkin, O. Morozova, L. Fedorchenko, & N. Khairov [5]; D. Novikov [16]): 1. Analysis of the pedagogical system from the point of view of management relations and information flows exchanged between the management and managed subsystems. 2. Optimization of the learning process, finding such forms and methods of organizing the educational process in which the functioning of the education system would be most effective, that is, at the lowest cost, it would bring maximum benefit. 3. Practical use of electronic devices and automated training systems for managing the learning and testing process; programmed learning. Among the modern methods of research of pedagogical systems, a special position is occupied by methods of mathematical and simulation modeling. Their essence lies in the fact that the real pedagogical system is replaced by an abstract model – some idealized object that has the most essential qualities of the system under study. By changing the initial data and parameters of the model, it is possible to investigate the ways of development of the system, determine its state at the end of training [9]. This is the advantage of this approach in comparison with the method of qualitative analysis. In order to clearly understand the “cyberontological approach”, we have developed a functional cyberontological model of dependencies and the flow of the information processes in human educational activities (see Fig. 1). Through “Threat agents”, cybernetic destructive information influences on the trained individual are created in order to lead to the loss of “Assets” in the future. Assets include future knowledge, diploma, job, salary, and so on. The challenge is for the learner to understand the need for “Assets” and to make an effort in the game to form quasi-professional solutions that would neutralize the cybernetic destructive information influences created by “Threat agents”.
Professional Training of Lecturers
1073
3.2 Practice of Applying the Cyberontological Approach in Educational Institutions The practice of applying the cyberontological approach in education developed on an intuitive level, through gamification (N. Rybka [17]). With the advent of computer technology, the motivation of students and cadets to traditional teaching methods began to decline paradoxically rapidly. Game methods of teaching adults have reached the first milestone. The phenomenon of the game is that, being entertainment, recreation, it is able to develop into learning, creativity, a model of the type of human relationships and manifestations in work. It is through the gradual introduction of the cyberontological approach into practice that this approach has contributed to the rethinking of the game from the point of view of the learning method. The pedagogical game has a clearly formed goal of teaching students (cadets) and its corresponding pedagogical results, which can be justified, highlighted in an explicit form and are characterized by an educational and cognitive orientation.
Fig. 1. Functional cyber ontological model of dependencies and the flow of the information processes in human educational activity
This is confirmed by the development of an innovative concept of independent training of Ground Forces cadets on training equipment by playing on a virtual computer [18] and methods of teaching electrical engineering disciplines [19]. It is important to correctly identify the motivation of players and forward it in the direction of learning, and not to cybercrime (V. Leshchina, I. Kozubtsov, & L.Kozubtsova [20]), the consequences of which can be catastrophic (I. Kozubtsova, L. Kozubtsov, T. Tereshchenko, & T. Bondarenko [21]). If the method of pedagogical game is turned into a pedagogical technology, then you can get a wide practical application. Given this circumstance, gamification began to be considered from the angle of game pedagogical technology in education, having a huge
1074
O. Silko et al.
potential for a positive impact on the effectiveness of the educational process, which began to attract more and more attention of researchers (K. Tseas, N. Katsioulas, & T.Kalandaridis [22, p. 25]; V. Bugaeva [23, p. 135]; S. Petrenko [24]; N. Rybka[17]). Publication (D. Kaufmann [25]) confirms that the author has made significant progress in gamifying online higher education. For the purposes of further practice of applying the cyberontological approach in education, software developers are invited to participate in the creation of a game cybersecurity strategy similar to that used in computer games. Its emulation, as a computer game, is rationally used in the training of military information and cyber security specialists to acquire practical skills in working out the settings of routers, firewalls, etc.on a conditional training information system [26]. As a result of virtual 3D modeling of educational experiments in a in game form, it is impossible for experimental equipment to come out of working order due to erroneous actions. 3.3 The Problem in the Use of Higher Educational Institutions with Specific Learning Conditions in the Educational Process Higher educational institutions with specific training conditions are equated with higher military educational institutions (HMEI) – an institution of higher education of state ownership that trains cadets (trainees, students), adjuncts at certain levels of higher education for further service in the positions of officers (Non-Commissioned Officers and Warrant Officers) in order to meet the needs of the Armed Forces of Ukraine, other military formations formed in accordance with the laws of Ukraine, central executive authorities with a special status, the Security Service of Ukraine, the Foreign Intelligence Service of Ukraine, other intelligence agencies of Ukraine, the central executive authority, what implements the state policy in the field of state border protection. The result of using game technologies for training cadets in the HMEI confirmed high efficiency, especially in difficult conditions [18]. However, as practice has shown, the dynamic range of creative activity of a lecturer in higher military educational institutions is too small and is limited to the forms and methods of teaching. Organization and training of officers for the needs of the Armed Forces of Ukraine and the security and defense sector is carried out In accordance with the Order of the Ministry of Defense of Ukraine. 3.4 Discussion of Research Results Gamification and a cyberontological approach can not be implemented for all academic disciplines of the educational component. Gamification is clearly prohibited in special professional academic disciplines during which information with restricted access for educational purposes circulates in the classroom. An interesting combination is the cyberontological approach and gamification when creating a computer game “Become the Head of a field information and communication node” in the course of studying tactical and special disciplines [27]. Assessment and feedback mechanisms are essential components towards effective teaching in higher education and are continuously monitored [28].
Professional Training of Lecturers
1075
Gamification is a good alternative educational practice to promote programming teaching, it allows better engagement of students in their learning. Students acquire a reasonable level of abstraction and logic and develop reflections on various course concepts [29]. Further study and adequate discussion of the results is expected. A more comprehensive study could reveal additional dimensions of the survey and shed light on how students perceive assessment, evaluation and feedback in higher education in general. Demonstrate some important aspects of this and indicate that improved quality of assessment and feedback can have a positive impact on student satisfaction. A more comprehensive study could reveal additional dimensions of the survey and shed light on how students perceive assessment, evaluation and feedback in higher education in general. As can be clearly seen from Table 1 the following learning methods are the most accepted for the formation of knowledge: didactic games; practical training; teaching others (mutual learning) and independent work [26]. Table 1. Comparative characteristics of different teaching methods Teaching methods
Solved tasks form
Verbal
Visually
(lecture)
knowledge
gaining skills thinking memory language experience
5%
++
–
–
–
++
+
++
+
+
–
+
+
+
++
–
reading
10%
listening
20%
Work with multimedia (audiovisual)
20%
Viewing the drawing
30%
Demonstration 30%
develop
Video viewing 50% Practical
Working with a book (reading)
10%
+
+
+
+
+
Educational discussions
50%
++
–
++
+
++
++
Didactic games
70%
++
–
++
+
++
++ (continued)
1076
O. Silko et al. Table 1. (continued)
Teaching methods
Solved tasks form knowledge
Practical training
75%
Teaching 90% others is the application of what has been learned Independent work Oral and written control
80%
develop gaining skills thinking memory language experience
+
++
++
+
–
++
++
–
++
+
++
++
++
++
++
+
+
++
–
+
+
++
Note: Note: ++ – solve very well; + – solve partially; – – solve poorly
4 Summary and Conclusion Thus, cybersocialization of the individual affects all segments of society, people of different ages, social positions and statuses. The degree of socialization depends on the individual characteristics of people. At the moment, the socialization of people is actively developing in five vectors. Given the accumulated positive experience of using computer technologies in the training of people of different ages, it is considered appropriate to use a cyberontological approach in the professional training of future lecturers. Cyberontological and game approaches in education aim to create such educational conditions that a person, learning in the course of a computer game, does not even suspect that learning is taking place, obtaining new knowledge. In a computer game, there is no source of knowledge that is easily learned by students. The learning process develops in the language of actions as a result of active contacts with each other, unobtrusively. 4.1 Expanding the Boundaries of the Scientific Field The scientific result obtained in the work expands the scientific boundaries of pedagogical sciences in the field of application of game mechanics in the preparation of higher education students in conditions of low motivation. Thus, the expansion takes place in the systemic unity of the philosophy of education and the theory and methodology of vocational education. 4.2 Scientific Novelty. Scientific Justification For the first time, a functional cyberontological model of dependencies and the flow of the information processes in the educational activities of a lecturer of a higher education institution is proposed, that allows us to find out all the interdisciplinary connections of others in cyberspace and the assets of students.
Professional Training of Lecturers
1077
4.3 Practical Use It is possible to use a computer game (gamification) as the basis for building a training complex for training military information and cybersecurity specialists. 4.4 Prospects for Further Research and Study The research is expected to bring more academic and applicable value. To do this, you need to pay extra: clarification of current problems of applying the gamification of training in higher military educational institutions. Acknowledgment. The authors would like to express their respect to the organizers of the 3rd International Conference on Artificial Intelligence and Logistics Engineering (ICAILE2023) (March 11 – March 12, 2023, Wuhan, China) and at a tragic time for Ukrainian researchers, Modern Education and Computer Science Press provided a grant to publish their scientific achievements free of charge. We are sincerely grateful.
References 1. Ahmed, N., Nandi, D., Zaman, A.G.M.: Analyzing student evaluations of teaching in a completely online environment. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 14(6), 13–24. https:// doi.org/10.5815/ijmecs 2. Pleshakov, V.A.: Gerontology and psychology of information sphere security: aspect of human cybersocialization in the social networks of the Internet environment. Bull. Orthodox St. Tikhon’s Univ. Humanit. Ser. IV Pedagogy Psychol. 4(19), 131–141 (2010). (in Russian) 3. Pleshakov, V.A.: Prospects of the cyberontological approach in modern education. Bull. Moscow City Pedagogical Univ. Ser. Pedagogy Psychol. 3(29), 1–18 (2014). (in Russian) 4. Voloshina, N.M., Zhogina, L.M., Mukogorenko, O.S.: On the issue of cybersocialization of Ukrainian society. In: Military Education and Science: Present and Future: XIII International Scientific and Practical Conference, Kiev, VIKNU, pp. 126–127, 24 November 2017. (in Ukrainian) 5. Meteshkin, K.A., Morozov, O.I., Fedorchenko, L.A., Khayrova, N.F.: Cybernetic pedagogy: ontological engineering in teaching and education, p. 207. HNAGH, Kharkiv (2012). (in Russian) 6. Bespalko V.P.: Cyberpedagogics. Pedagogical Foundations of Computer-Controlled Learning (E-Learning), p. 240. T8RUGRAM/Public Education, Moscow (2018). (in Russian) 7. Igibaeva, A.K., Toleukhanova, A.D.: Cyber pedagogy – as an innovative branch of pedagogical science. Bull. KazNPU Named after Abai Ser. Pedagogical Sci. 68(4), 12–17 (2020). (in Russian) 8. Kozubtsov, I.N., Kozubtsova, L.M.: Cyber pedagogy as a new trend in education of the XXI century: problems and risks. In: International Scientific and Practical Conference: “Pedagogical Education in the XXI Century: Priorities and Searches”, Astana, pp. 402–409, 7 October 2022. (in Russian) 9. Mayer, R.V.: Cybernetic Pedagogy: Simulation Modeling of the Learning Process, p. 138. Glazov, GGPI (2013). (in Russian) 10. Pleshakov, V.A.: Cybersocialization as an innovative socio-pedagogical phenomenon. Teacher XXI Century 3(1), 32–39 (2009). (in Russian)
1078
O. Silko et al.
11. Chitosca, M.I.: The internet as a socializing agent of the “M Generation.” J. Soc. Inform. 5, 3–21 (2006) 12. Yu, P.S.: “Cybersocialization” – A = A. Inf. Law 3(12), 25–30 (2014). (in Ukrainian) 13. Voinova, O.I., Pleshakov, V.A.: Cyberontological Approach in Education, p. 244. Norilsk, Research Institute (2012). (in Russian) 14. Kozubtsov, I.M., Beskrovny, O.I., Kozubtsova, L.M., Palaguta, A.M., Mironenko, O.V.: Trends in the educational and scientific space: problems and risks. Bull. Mikhail Ostrogradsky Kremenchug National Univ. 1(132), 40–48 (2022). (in Ukrainian) 15. Mashbits, E.I.: Psychological and Pedagogical Problems of Computerization of Education, p. 192. Pedagogy, Moscow (1988). (in Russian) 16. Novikov, D.A.: Theory of Educational Systems Management, p. 416. Public Education, Moscow (2009). (in Russian) 17. Rybka, N.M.: Graization and experience of using computer games in teaching philosophy in technical institutions of higher education. Inf. Technol. Training Tools 67(5), 213–225 (2018). (in Ukrainian) 18. Kozubtsov, I.M.: The concept of independent training of cadets of the ground forces on training facilities by playing on a virtual computer. In: Prospects for the Development of Weapons and Military Equipment of the Ground Forces. Collection of Abstracts of the Second All-Ukrainian Scientific and Technical Conference, Lviv, p. 77, 28–29 April 2009. (in Ukrainian) 19. Kozubtsov, I.N.: Teaching electrical engineering disciplines by virtual computer game method. In: Electrical Technologies, Electric Drive and Electrical Equipment of Enterprises: Collection of Scientific Papers of the II All-Russian Scientific and Technical Conference. USNTU Publishing House, Ufa, vol. 2, pp. 107–110 (2009). (in Russian) 20. Leshchina, V., Kozubtsov, I., Kozubtsova, L.: Role of motivative characteristics in cyber security ontology. Sci. Pract. Cyber Secur. J. (SPCSJ) 6(1), 15–23 (2022) 21. Kozubtsova, L.M., Kozubtsov, I.M., Tereshchenko, T.P., Bondarenko, T.V.: On the cyber security of military personnel playing geolocation games while staying at departmental facilities of critical information infrastructure. Cybersecurity Educ. Sci. Technol. 1(17), 76–90 (2022). (in Ukrainian) 22. Tseas, K., Katsioulas, N., Kalandaridis, T.: Gamification in higher education. M.S. Thesis, Department, Electrical and Computer Engineering, University of Thessaly, Volos, Greece (2014). https://doi.org/10.4102/hts.v73i3.4527 23. Yu, B.V.: Gamification as a way of forming active professional behavior of future IT industry specialists. Pedagogy Psychol. 56, 129–135 (2018). https://doi.org/10.5281/zenodo.577567. (inUkrainian) 24. Petrenko, S.V.: Gamification as an innovative educational technology. Innov. Educ. 2(7), 177–185 (2018). (in Ukrainian) 25. Kaufmann, D.A.: Reflection: benefits of gamification in online higher education. J. Instr. Res. 7, 125–132 (2018) 26. Kozubtsova, L.M., Kozubtsov, I.M., Lishchina, V.A., Shtanenko, S.S.: Concept of the educational and training complex for training military specialists in information and cybersecurity on the basis of computer games (gamification). Cybersecurity Educ. Sci. Technol. 2(18), 49–60 (2022). (in Ukrainian) 27. Ponomarev, O.A., Pivovarchuk, S.A., Kozubtsov, I.M.: On the use of the computer game “Become the head of a field information and communication node” in the course of studying tactical and special disciplines. Collection of Reports and Abstracts of Materials of the II International Scientific and Technical Conference on Communication, Informatization and Cybersecurity Systems and Technologies: Current Issues and Development Trends, Kiev, VITI, pp. 174–175, 1 December 2022. (in Ukrainian)
Professional Training of Lecturers
1079
28. Rigopoulos, G.: Assessment and feedback as predictors for student satisfaction in UK higher education. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 14(5), 1–9 (2022). https://doi.org/10. 5815/ijmecs 29. Ouahbi, I., Darhmaoui, H., Kaddari, F.: Gamification approach in teaching web programming courses in PHP: use of KAHOOT application. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 13(2), 33–39 (2021)
Exploring the Perceptions of Technical Teachers Towards Introducing Blockchain Technology in Teaching and Learning P. Raghu Vamsi(B) Department of Computer Science and Engineering and IT, Jaypee Institute of Information Technology, A-10, Sector 62, Noida 201307, India [email protected]
Abstract. This paper presents a study that assesses technical teachers’ perceptions of learning blockchain technology and its applications to plan efficient faculty development programs for industry, academia, and research. With universities and technical councils introducing courses to bridge the gap between academic learning and industry demands, the demand for blockchain developers has been on the rise. The study’s findings indicate that technical teachers generally hold positive attitudes towards learning blockchain technology, recognizing its potential applications in various fields such as finance, healthcare, and energy. Quantitative and qualitative study conducted on 79 engineering teachers and professionals who participated in an online faculty development program on Blockchain Technology and its Applications at JIIT University, India by providing survey questionnaire. The study highlights the importance of continuous adaptation and learning of new technologies for teachers to equip students with the skills they need to succeed in the industry. The study also provides insights into technical teachers’ perceptions towards integrating blockchain technology in teaching and learning and offers implications for planning efficient faculty development programs. Query ID="Q1" Text="This is to inform you that corresponding author has been identified as per the information available in the Copyright form." ‘ Keywords: Blockchain technology · Faculty development · Motivation · Performance · Professional development · Technical education
1 Introduction The field of teaching requires educators to engage with students to facilitate their learning, which involves various tasks such as planning, delivering, assessing, and reflecting on content. Effective teaching demands a methodical understanding of learning to establish frameworks that encourage students to take responsibility for their own learning [1–3]. As new technological paradigms emerge, teachers must adapt and develop the necessary attitudes, knowledge, skills, and abilities to become competent instructors [4, 5]. Faculty development programs are essential to disseminate expertise from teachers to students. The term “Perceptions” refers to teachers’ attitudes, behavior, self-beliefs, © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 1080–1089, 2023. https://doi.org/10.1007/978-3-031-36115-9_96
Exploring the Perceptions of Technical Teachers
1081
and views regarding new technology, which can influence their beliefs about the importance of integrating it into teaching and learning. Continuous adaptation and learning of new technologies is necessary for teachers to equip students with the skills they need to succeed in the industry [6–9]. Prior research has focused on exploring the perceptions of teachers and students towards various technological paradigms, such as ICT [10–12], e-learning systems [13], flipped classrooms [14], and online teaching during the COVID19 pandemic [15]. However, the rise of blockchain technology as a secure and distributed database technology, which offers solutions for financial transactions, healthcare record protection, and energy and carbon emission trading, among others, demands attention from the academic community. With universities and technical councils introducing courses to bridge the gap between academic learning and industry demands, the demand for blockchain developers has been on the rise. To facilitate efficient faculty development programs for industry, academia, and research, it is crucial to assess technical teachers’ perceptions of learning blockchain technology and its applications [16–18]. Therefore, this study aims to assess technical teachers’ perceptions of learning blockchain technology and its applications to plan efficient faculty development programs for industry, academia, and research. The remainder of the paper is structured as follows: Sect. 2 presents related work, Sect. 3 presents the methodology of the work, Sect. 4 presents the results with discussion, and Sect. 5 concludes the paper.
2 Related Work Nyme et al. [19] analyzed the student evaluations of teaching in a completely online environment to uncover the key factors that influence the quality of teaching in a virtual classroom environment. The authors utilize Educational Data Mining (EDM) techniques to analyze data gathered from the student evaluations of computer science students in three online semesters at X University. They use Weka, sentimental analysis, and word cloud generator to carry out the research. The decision tree classifies the factors affecting the performance of the teachers, and the authors find that student-faculty relation is the most prominent factor for improving the teaching quality. Tushar et al. [20] showed the status of ICT education and finds the gaps between rural and urban institutions for providing ICT education in secondary and higher secondary institutions in Bangladesh. The authors use primary data collected using a survey questionnaire that is answered by ICT teachers engaged in those institutions. This research found that that there exist low facilities in rural institutions compared to the urban institutions because studentscomputer ratio (SCR) is 46 in rural areas whereas SCR is 22 in the urban area. Finally, authors presented recommendations to meet the identified gaps. Chen et al. [21] explored the potential applications of blockchain technology in education, highlighting its features, benefits, and current applications. The authors propose innovative applications of blockchain technology and discuss the challenges of implementing it in education. Araujo et al. [22] presented study on the integration of Flipped Classroom and Adaptive Learning techniques for Blockchain courses in higher education. The study evaluates student acceptance and the learning impact of out-of-class material with and without Adaptive Learning techniques, with positive results observed. Chang et al. [23] investigated the determinants affecting the acceptance of blockchain technology in the tourism
1082
P. R. Vamsi
industry on Jeju Island, Korea. The authors analyze the effect of blockchain trust transparency on performance and effort expectancy, finding that facilitating conditions are the most influential factor in blockchain technology acceptance. The role of blockchain technology in the sustainable development of students’ learning process is discussed by Popa et al.[24]. Two specialized studies conducted in Romania are presented, examining teachers’ and students’ perceptions of technological integration in the learning process. The authors find that implementing a reward system through blockchain technology significantly improves students’ motivation and creativity.
3 Methodology of Study The methodology of this study involved the development and distribution of a questionnaire to participants of an online Faculty Development Program (FDP) at Jaypee Institute of Information Technology in July 2022. The questionnaire consisted of 20 questions categorized into four motivational factors: job expectations, social desire, personal desire, and technical capabilities. Each question was surveyed using a five-point scales, with a range of strongly agree to strongly disagree. The questionnaire was distributed for on-spot completion and collection, and the response rate was limited by the FDP participants’ attendance on the survey date, which was held on the last day of the FDP and just before the conduction of the final quiz. The collected questionnaires were compiled into a Google Spreadsheet and analyzed using Python programming with the Pandas data science library [25–27]. The analysis involved calculating the mean value for each question, with a mean value of 5 indicating complete agreement with the question merit, and a mean value of 1 indicating complete disagreement with the question merit. Detailed statistical results of the survey were compiled and are presented in Table 1.
4 Results and Discussion Total 79 members participated in the survey, with 51.9% female and 48.1% male participants. Of the total, 48.1% have a Master’s degree in Technology, 45.5% have a Doctorate degree in computer science and information technology, and 6.2% have a non-technical degree. Out of the 79 participants, 56.96% are Assistant professors, 18.98% are Associate professors, 18.98% are doctorate research scholars, and 2.6% are software professionals. 31.6% of the participants are in the age range 25–34 years, 59.5% are in the age range of 35–44 years, and 8.9% are in the age range 45–54 years. 36.7% of the participants have more than 15 years of work experience, 26.5% have 10–14 years of work experience, 17.7% have 5–9 years of work experience, and 15.2% have 1–4 years of work experience. When asked about their intention behind learning Blockchain technology, 48.1% answered to improve their knowledge on the topic, 20.5% answered that this technology helps in their ongoing research work, 20.2% answered to become sound in the latest technology, and 6.3% answered to shift their work from academics to industry. Despite women making up the majority of the participants, middle-aged professors are more cognizant of the value of Blockchain technology and understand that it is already crucial for industry 4.0. Many participants believe that learning Blockchain technology would help them progress in their jobs in technology and research because it is a novel subject. 5% of the participants intend to leave academia and work in industry.
Exploring the Perceptions of Technical Teachers
1083
From Table 1, the study found that personal desire was the strongest motivator for faculty members to take up blockchain technology courses, while social desire was the lowest. A significant number of participants were Assistant Professors with a Master’s degree in Technology and over 15 years of work experience. Job expectations were the second strongest motivator, with respondents showing awareness of the potential employment opportunities in the blockchain sector. The social desire category was significant, with over 80% of respondents believing that blockchain development is a prestigious profession that will maintain their social status. The technical capabilities category was the third motivator, with respondents considering their technical skills as essential for taking a blockchain course. The mean score for the job expectations category was 4.1, while the social desire category had a mean score of 3.5 out of 5. On average, 35.7% of faculty members fell behind in the four categories, while 35.4% strongly agreed with all four. Overall, 71.1% agreed or strongly agreed with the four categories, while 28.9% were neutral, disagreed, or strongly disagreed. Figure 1 shows the grade distribution of participants who attended the Blockchain FDP, which aims to improve their programming skills in Blockchian and smart contracts. Participants must attend 80% of sessions to qualify for the final quiz, and their grades are used to assess motivation. The course average is 3.5, which is a B + and slightly higher than the normal average of C +. This suggests that participants’ technical abilities are above average, which is consistent with their confidence in the technical capabilities category of the survey. This positive observation could be attributed to their prior knowledge in computer networks, cryptography, and programming. Selected participants’ evaluations of a Blockchian FDP were analyzed, following All India Council for Technical Education (AICTE) guidelines. 5 items from 16 were chosen, and the average response for each question item was 4.5 out of 5. The participants have a high level of confidence in the knowledge and skills learned in the course. The participants suggested correlating blockchian theoretical aspects with application development and research areas, as well as holding hands-on training workshops on full stack blockchian application development and research directions. The study found that 48.1% of the participants intended to learn Blockchain technology to improve their knowledge, while 20.5% answered that this technology helps in their ongoing research work. 20.2% answered that they intended to become sound in the latest technology, and 6.3% answered that they wanted to shift their work from academics to industry. Interestingly, despite women making up the majority of the participants, middle-aged professors were more aware of the value of Blockchain technology and understood its importance for industry 4.0. With these results, this study found that faculty members are motivated to learn Blockchain technology for various reasons, including improving their knowledge, ongoing research work, and staying up-to-date with the latest technology. Personal desire was found to be the strongest motivator, while social desire was the weakest. The study also found that academic staff members were well-informed about employment options in the sector, and the potential for new research initiatives and technology to pay off. The
1084
P. R. Vamsi
Table 1. Technical Teacher’s feedback on their motivation towards Blockchain Technology Course at FDP conducted at JIIT Noida. (Scale: 5 = Strongly agree, 4 = Agree, 3 = Neutral, 2 = Disagree, and 1 = Strongly disagree) Category
Survey Questions
%of Answers
Mean (out of 5)
Job expectation
5
4
3
2
1
C1.1
I want to learn blockchain application development, because this is a well-paid domain compared to most of the other software domains
40.5
40.5
13.9
3.8
1.3
4.2
C1.2
I want to learn blockchain application development, because it will be easy for me to find a good job
26.6
40.5
27.8
2.5
2.5
3.9
C1.3
I want to learn blockchain 46.8 application development, because it will be easy for complete my research work / publish research papers
34.2
17.7
1.3
0
4.3
C1.4
I want to learn blockchain 36.7 application development, because my current job role demanded me learn it
35.4
25.3
2.5
0
4.1
Category 1 average
4.1
Social desire C2.1
I want to learn blockchain application development, because this is going to be one of the prestigious professions that will maintain my class level in the society
39.2
32.9
22.8
2.5
2.5
4
C2.2
I want to learn blockchain application development, because my family (parents) told me to do so since cryptocurrencies like Bitcoin is gaining importance
31.6
22.8
25.3
5.1
15.2
3.5
(continued)
Exploring the Perceptions of Technical Teachers
1085
Table 1. (continued) Category
Survey Questions
%of Answers
C2.3
I want to learn blockchain application development, because my friends/peers told me to do so
32.9
21.5
Mean (out of 5) 25.3
7.6
12.7
Category 2 average
3.5
3.6
Personal desire C3.1
I want to learn blockchain 40.5 application development, because this will contribute to society by solving community/world problems
41.8
17.7
0
0
4.2
C3.2
Blockchain application development is interesting and fun and I feel good when I deal with trust and privacy issues
38
45.6
15.2
1.3
0
4.2
C3.3
I chose blockchain application development because I like to build and fix trust and privacy issues in business chain
40.5
39.2
20.3
0
0
4.2
C3.4
I chose blockchain application development because I like to develop and design new systems
34.2
48.1
17.7
0
0
4.2
C3.5
I chose blockchain application development because I like to work on projects in teams to solve problems
44.3
36.7
16.5
2.5
0
4.2
C3.6
I chose blockchain application development because I feel I have the ability to present new systems/products to the public/students than software professionals
40.5
36.7
20.3
2.5
0
4.2
Category 3 average
4.2 (continued)
1086
P. R. Vamsi Table 1. (continued)
Category
Survey Questions
%of Answers
Mean (out of 5)
Technical capabilities C4.1
I chose blockchain application development because I like networking and security subjects
C4.2
38
35.4
22.8
3.8
0
4.1
I chose blockchain 26.6 application development because I have been always doing great in distributed networking and have great grip on subject
41.8
22.8
8.9
0
3.9
C4.3
I chose blockchain 30.4 application development because I have been always doing great in network security and have great grip on subject
35.4
24.1
10.1
0
3.9
C4.4
I chose blockchain 31.6 application development because I have been always doing great in full stack web development and have great grip on subject
35.4
22.8
10.1
0
3.9
C4.5
I chose blockchain 25.3 application development because I have been always doing great in networking, security and full stack development and have great grip on subjects
41.8
25.3
7.6
0
3.8
C4.6
I chose blockchain application development because I like to develop security applications, analyze and address the trust and privacy issues
34.2
40.5
19
6.3
0
4
C4.7
I chose blockchain application development because I like to develop cryptocurrencies
25.3
38
25.3
11.4
0
3.8
(continued)
Exploring the Perceptions of Technical Teachers
1087
Table 1. (continued) Category
Survey Questions
%of Answers
Mean (out of 5)
Number of Parcipants
Category 4 average
50 40 30 20 10 0
Count
3.9
A-A+
B-B+
C-C+
D
F
10
40
18
8
3
Fig. 1. Final grades of the participants in FDP
findings of this study could be used to design effective training programs for faculty members in Blockchain technology. Important observations of this study are. • This study examined the perceptions and motivations of faculty members towards learning Blockchain technology. Results showed that most participants were assistant professors with a master’s degree in technology and over 15 years of work experience. • Personal desire was found to be the primary motivator, followed by job expectations and technical capabilities. Social desire was the lowest motivator. Middle-aged faculty members had a higher level of awareness about the importance of Blockchain technology in industry 4.0. • Many participants considered leaving academia to work in the industry due to the high demand for blockchain developers and better pay. Policymakers and educators can use these findings to design curriculum and training programs that focus on personal motivation and job expectations. • Finally, this study suggests promoting the social value of learning Blockchain technology to maintain social class status. The study identified an opportunity for future research to explore additional motivators for taking Blockchain technology courses.
5 Conclusion The purpose of this study was to examine the motivational behaviors and performance of participants in online faculty development programs focused on blockchain technology. Based on the survey results, it is clear that individuals pursuing careers in teaching and research recognize the importance of academic professionals in solving technical challenges facing society. Furthermore, respondents understand that programming, networking, and security concepts are essential for successful blockchain technology research. However, to maintain motivation and confidence in pursuing blockchain technology as
1088
P. R. Vamsi
a primary focus, participants require adequate support in terms of teaching blockchain application development and research tools. The study recommends that participants be informed about the correlation between basic computer science courses and full-stack application development with blockchain technology at the beginner and intermediate levels of academic teaching. This knowledge will enable them to effectively address research problems related to or with blockchain technology and maintain their motivation and confidence. Ultimately, this will contribute to the development and adoption of blockchain technology in various industries, benefiting society as a whole.
References 1. Allocation of teachers who participated in professional development in ICT skills, Organisation for Economic Co-Operation and Development (OECD) (2022) 2. Mardiana, H.: Lecturers’ adaptability to technological change and its impact on the teaching process. JPI (Jurnal Pendidikan Indonesia) 9, 275–289 (2020) 3. Darby, A., Newman, G.: Exploring faculty members’ motivation and persistence in academic service-learning pedagogy. J. High. Educ. Outreach Engagem. 18, 91–119 (2014) 4. Gomez, D.R., Swann, W., Willms Wohlwend, M., Spong, S.: Adapting under pressure: a case study in scaling faculty development for emergency remote teaching. J. Comput. High. Educ. 35, 1–20 (2022) 5. Phuong, T.T., Foster, M.J., Reio, T.G., Jr.: Faculty development: a systematic review of review studies. New Horiz. Adult Educ. Hum. Resour. Dev. 32, 17–36 (2020) 6. Daniels, E.: Logistical factors in teachers’ motivation. Clearing House J. Educ. Strat. Issues Ideas 89, 61–66 (2016) 7. Welch, M., Plaxton-Moore, S.: Faculty development for advancing community engagement in higher education: current trends and future directions. J. High. Educ. Outreach Engagem. 21, 131–166 (2017) 8. Sadiku, S.: Factors affecting teacher motivation. Int. Sci. J. Monte (ISJM) 4, 98–113 (2021) 9. Penprase, B.E.: The fourth industrial revolution and higher education. High. Educ. Era Fourth Ind. Revolution 10, 978–981 (2018) 10. Saito, H., Umeda, K.: The development of teaching skills using ICT in teacher training: practices in first-year introduction for ICT. In: Proceedings of the 2019 3rd International Conference on Education and Multimedia Technology, New York, NY, USA (2019) 11. Liesa-Orús, M., Latorre-Cosculluela, C., Vázquez-Toledo, S., Sierra-Sánchez, V.: The technological challenge facing higher education professors: perceptions of ICT tools for developing 21st century skills. Sustainability 12, 5339 (2020) 12. Razak, S., Khan, S., Hussein, N., Alshikhabobakr, H., Gedawy, H., Yousaf, A.W.: Integrating computer science and ICT concepts in a cohesive curriculum for middle school - an experience report. In: Proceedings of the 52nd ACM Technical Symposium on Computer Science Education, New York, NY, USA (2021) 13. Hashim, N.A., Mukhtar, M., Safie, N.: Factors affecting teachers’ motivation to adopt cloudbased E-learning system in Iraqi deaf institutions: a pilot study. In: Proceedings of the 2019 International Conference on Electrical Engineering and Informatics (ICEEI) (2019) 14. Mishra, C.: Faculty perceptions of digital information literacy (DIL) at an Indian University: an exploratory study. New Rev. Acad. Librarianship 25, 76–94 (2019) 15. Redmond, P., Heffernan, A., Abawi, L., Brown, A., Henderson, R.: An online engagement framework for higher education. Online Learn. 22, 183–204 (2018) 16. Ravindran, U., Bhardwaj, P., Vamsi, P.R.: Blockchain design for securing supply chain management in coffee retailer network. Int. J. Sci. Res. Sci. Technol. 7, 492–502 (2021)
Exploring the Perceptions of Technical Teachers
1089
17. IEEE Standard for Application Technical Specification of Blockchain-based E-Commerce Transaction Evidence Collecting. IEEE 18. Ravindran, U., Vamsi, P.R.: A secure blockchain based finance application. In: Proceedings of the 2021 Thirteenth International Conference on Contemporary Computing (IC3–2021) (2021) 19. Ahmed, N., Nandi, D., Zaman, A.G.M.: Analyzing student evaluations of teaching in a completely online environment. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 14(6), 13–24 (2022). https://doi.org/10.5815/ijmecs.2022.06.02 20. Saha, T.K., Shahrin, R., Prodhan, U.K.: A survey on ICT education at the secondary and higher secondary levels in Bangladesh. Int. J. Mod. Educ. Comput. Sci. (IJMECS) 14(1), 17–29 (2022). https://doi.org/10.5815/ijmecs.2022.01.02 21. Chen, G., Xu, B., Lu, M., Chen, N.-S.: Exploring blockchain technology and its potential applications for education. Smart Learn. Environ. 5(1), 1 (2018). https://doi.org/10.1186/s40 561-017-0050-x 22. Araujo, P., Viana, W., Veras, N., Farias, E.J., de Castro Filho, J.A.: Exploring students perceptions and performance in flipped classroom designed with adaptive learning techniques: a study in distributed systems courses. In: Brazilian Symposium on Computers in Education (Simpósio Brasileiro de Informática na Educação-SBIE), vol. 30, no. 1, p. 219 (2019) 23. Chang, M., Arachchilage, C.S.M.W., Kim, M., Lim, H.: Acceptance of tourism blockchain based on UTAUT and connectivism theory. Technol. Soc. 71, 102027 (2022) 24. Popa, I.-C., Orzan, M.-C., Marinescu, C., Florescu, M.S., Orzan, A.-O.: The role of blockchain technologies in the sustainable development of students’ learning process. Sustainability 14(3), 1406 (2022) 25. Punch, K.F., Oancea, A.: Introduction to Research Methods in Education. Sage (2014) 26. Cohen, L., Manion, L., Morrison, K.: Research Methods in Education, 7th ed., Routledge (2011) 27. Ouda, O.K.M., AL-Asad, J.F., Asiz, A.: An assessment of student’s motivations to join College of Engineering: Case study at Prince Mohammad Bin Fahd University - Saudi Arabia (2014)
Author Index
A Acheme, Ijegwa David 152, 165 Aderounmu, Ganiyu A. 239 Adigwe, Wilfred 165 Ahmed, Aiman Magde Abdalla 944 Akhatov, Akmal 437 Akhter, Mshura 129 Akhund, Tajim Md. Niamat Ullah 178 Akinyemi, Bodunde O. 239 Akter, Lubana 129 Alifov, Alishir A. 101 Anabtawi, Wasim 271 Anton, Yakovlev 416 B Bai, Chuntao 514, 876, 911 Bazilo, Constantine 120 Beskrovnyi, Oleksii 1068 Bilanovych, Alisa 310 Borhan, Rownak 299 C Cai, Qi 573, 734 Cao, Jingjing 681 Chang, Ming 1006 Chen, Chen 983 Chen, Hongbao 963 Chen, Jinming 963 Chen, Liping 963 Chen, Ning 640, 754 Chen, Qingfeng 719 Chen, Xi 628 Chen, Yan 65, 963 Chen, Yi 598 Chen, Yuan 25 Chimir, Igor 35 Cui, Bin 524 Cui, Zhiyu 1028 D Dai, Jinshan 65, 899 Das, Prantho kumar 1039
Degila, Jules 239 Delyavskyy, Mykhaylo 382 Deresh, Olha 217 Dey, Samrat Kumar 129, 1039 Ding, Yao 833, 954 Dong, Pan 854 Dong, Sijie 899 Doronina, Iryna 404 Du, Lijing 3, 573, 734 Du, Weihui 651, 696 Du, Yong 514 Durnyak, Bohdan 187, 197 E Efiong, John E. 239 F Fan, Simeng 983 Fang, Can 651 Fang, Liang 777 Feng, Bin 886 Feng, Gong 681 Feng, Wei 1006 Filimonov, Sergey 120 Filimonova, Nadiia 120 Frontoni, Emanuele 288, 310 G Gao, Ying 598 Ge, Jingjing 822 Gumen, Olena 448 H Havryliuk, Myroslav 372 He, Anqi 524 Hong, Huawei 109 Honsor, Oksana 393 Huang, Bihui 14 Huang, Saixiao 25 Huang, Yixuan 944 Huang, Yujie 844 Huang, Zhicheng 65
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 Z. Hu et al. (Eds.): ICAILE 2023, LNDECT 180, pp. 1091–1094, 2023. https://doi.org/10.1007/978-3-031-36115-9
1092
I Ibrahim, Isa A.
Author Index
239
J Jarin, Mahnaz 334 Jiang, Chun 974 Jiang, Hong 764, 899 Joseph, Isabona 488 Ju, Chengwei 79 Jumanov, Isroil I. 478 K Kaminskyi, Andrii 426 Kaminskyy, Roman 372 Kandiy, Sergey 288, 310 Katerynchuk, I. 206 Khan, Saikat Islam 299 Kobylianska, Olena 288 Kopych, A. 206 Korniyenko, Bogdan 55 Kozubtsov, Ihor 1054, 1068 Kozubtsova, Lesia 1054, 1068 Kubytskyi, Volodymyr 321 Kuno, I. 206 Kuznetsov, Oleksandr 288, 310, 360 Kuznetsova, Kateryna 360 L Lavrenchuk, Svitlana 382 Li, Bin 524 Li, Liwei 931 Li, Shujun 524 Li, Wenhui 719 Li, Xin 764, 899 Li, Yan 764 Li, Yani 899 Li, Zhengxie 558, 664 Lian, Jinxiang 777 Liang, Chengquan 865 Liang, Tingting 90 Liang, Zhaomin 90 Liaskovska, Solomiya 448 Lin, Qian 719 Lishchyna, Nataliia 251, 382 Lishchyna, Valerii 251, 382 Lisovych, Taras 372 Liu, Bing 14 Liu, Ke 801
Liu, Li 833, 954 Liu, Mingfei 944 Liu, Minwei 811 Liu, Qian 681 Liu, Shan 708 Liu, Yanhui 777 Lu, Huiting 787 Lu, Liuqing 1016 Lukianchuk, Iurii 251 Luo, Wei 25 Luo, Ximei 719 Lutskiv, Mikola 187 Lv, Peng 109 M Ma, Jie 854 Mai, Liqiang 833, 954 Makinde, Ayodeji S. 152 Maqboul, Ahmad 271 Martyn, Yevgen 448 Melnyk, Roman 458 Mori, Margherita 3 Mutovkina, Nataliya 921, 995 Myronenko, Olha 1054 N Nazarov, Fayzullo 437 Nehrey, Maryna 404, 426 Neroda, Tetyana 197 Nesteruk, Andrii 55 Nezhyva, Olha 1054 Noman, S. M. 178 Nwankwo, Chukwuemeka Pascal Nwankwo, Wilson 152, 165 O Oberyshyn, Roksolana 393 Ogbonda, Clement 488 Oghorodi, Duke 165 Ojei, Emmanuel 165 Olajubu, Emmanuel A. 239 Olayinka, Akinola S. 152 Oleh, Lisovychenko 416 Othman Othman, M. M. 271 P Pan, Zhikang 754 Panchenko, Taras 321
152, 165
Author Index
Pang, Rongrong 764 Pang, Weinan 886 Pasieka, Mykola 187 Pasieka, Nadiia 187, 197 Perelygin, Sergey 229 Prokopov, Serhii 310 Putrenko, Viktor 404 Q Qi, Xiaoyang 514 Qi, Xin 583 Qian, Xiaorui 109 Qian, Yuntong 618 Qiao, Lichen 608 Qin, Fengcai 974 Qin, Shanyong 811 Qiu, Suzhen 90 R Rahaman, A. S. M. Mostafizur 334 Rahman, Elias Ur 299, 1039 Rahman, Md. Hafizur 178 Rahman, Md. Mahbubur 468 Rahman, Md. Mokhlesur 468 Rahman, Sabbir 468 Rashid, Md. Mamun Ar 1039 Rashidov, Akbar 437 Risi, Ikechi 488 Romanyshyn, Yulia 197 Rong, Donglin 696 S Safarov, Rustam A. 478 Saha, Dola 129 Salehin, Imrus 178 Semotiuk, O. 206 Sharmin, Fateha 299 Sharmin, Nusrat 468 Sheketa, Vasyl 187 Shepita, Petro 187, 197 Shmyhelskyy, Ya. 206 Silko, Oleksii 1068 Smirnov, Oleksii 288 Smirnova, Olga 995 Smirnova, Tetiana 310 Stetsenko, Inna V. 344 Su, Lifang 801 Su, Rengshang 79
1093
Sulim, Viktoriia 382 Sun, Jiawei 696, 719 Sveleba, N. 206 Sveleba, S. 206 T Tang, Jiwei 25 Tavrov, Danylo 141 Temnikov, Andrii 141 Temnikov, Volodymyr 141 Temnikova, Olena 141 The, Quan Trong 229 Tian, Jia 628 Tian, Jintian 745 Tikhonov, Andrii 360 Tolokonnikov, G. K. 261 Tong, Gang 109 Tong, Xinshun 536 Tulashvili, Yurii 251 Tupychak, Lyubov 197 Tushnytskyy, Ruslan 458 U Uddin, Khandaker Mohammad Mohi 299, 1039 Ulianovska, Yuliia 288 Umezuruike, Chinecherem 165 V Vamsi, P. Raghu 1080 Vdovyn, Mariana 217 Velgosh, S. 206 W Wang, Enshi 14 Wang, Haiyan 618 Wang, Jing 573, 734 Wang, Lulu 503 Wang, Meng 764, 899 Wang, Mengqiu 899 Wang, Xiujuan 514, 876, 911 Wang, Ying 573, 734 Wang, Zaitao 876, 911 Wei, Puzhe 583 Wen, Jin 833, 954 Wu, Hongyu 628 Wu, Xia 45 Wu, Yunyue 547
129,
1094
X Xiao, Kai 109 Xie, Mei E. 608 Xie, Sida 899 Xie, Xiaoling 65 Xing, Saipeng 651, 696 Xiong, Kevin 573, 734 Xiong, Shuiping 45 Xiong, Yanbing 573, 734 Xu, Yajie 536 Xv, Li 628 Y Yan, Yang 719 Yang, Haifeng 854, 963 Yang, Lingxue 719 Yang, Xiaozhe 787 Yang, Xue 640 Yao, Zhenhua 1006 Yashchenko, Sergei 120 Yassine, Radouani 681 Ye, Fang 524 Ye, Hui 608 Ye, Yanxin 844 Yemets, Kyrylo 372 Yezhov, Anton 360 Yin, Guanchao 833, 954
Author Index
You, Jiawei 944 Yu, Xunran 628 Yuan, Jun 503 Z Zarichkovyi, Alexander 344 Zdolbitska, Nina 382 Zhan, Xiangpeng 109 Zhang, Cheng 583 Zhang, Chenyu 65 Zhang, Geng E. 79, 1016 Zhang, Lei 745 Zhang, Long 558, 664 Zhang, Peilin 503 Zhang, Xiaoyu 651 Zhang, Yao 503, 608 Zhao, Ting 876, 911 Zhelykh, Vasyl 448 Zheng, Zhong 708, 931 Zhou, Qilai 833, 954, 1006 Zhou, Xiaofen 764, 899 Zhou, Xiaoguang 777 Zhou, Xiaoying 708 Zhu, Lingling 109 Zhu, Shixiong 931 Zomchak, Larysa 217 Zuo, Jing 844