e-Learning, e-Education, and Online Training: 8th EAI International Conference, eLEOT 2022 Harbin, China, July 9–10, 2022 Proceedings, Part I 303121160X, 9783031211607

Citation preview

Weina Fu Guanglu Sun (Eds.)

453

e-Learning, e-Education, and Online Training 8th EAI International Conference, eLEOT 2022 Harbin, China, July 9–10, 2022 Proceedings, Part I

Part 1

Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering Editorial Board Members Ozgur Akan Middle East Technical University, Ankara, Türkiye Paolo Bellavista University of Bologna, Bologna, Italy Jiannong Cao Hong Kong Polytechnic University, Hong Kong, China Geoffrey Coulson Lancaster University, Lancaster, UK Falko Dressler University of Erlangen, Erlangen, Germany Domenico Ferrari Università Cattolica Piacenza, Piacenza, Italy Mario Gerla UCLA, Los Angeles, USA Hisashi Kobayashi Princeton University, Princeton, USA Sergio Palazzo University of Catania, Catania, Italy Sartaj Sahni University of Florida, Gainesville, USA Xuemin Shen University of Waterloo, Waterloo, Canada Mircea Stan University of Virginia, Charlottesville, USA Xiaohua Jia City University of Hong Kong, Kowloon, Hong Kong Albert Y. Zomaya University of Sydney, Sydney, Australia

453

The LNICST series publishes ICST’s conferences, symposia and workshops. It reports state-of-the-art results in areas related to the scope of the Institute. LNICST reports state-of-the-art results in areas related to the scope of the Institute. The type of material published includes • Proceedings (published in time for the respective event) • Other edited monographs (such as project reports or invited volumes) LNICST topics span the following areas: • • • • • • • •

General Computer Science E-Economy E-Medicine Knowledge Management Multimedia Operations, Management and Policy Social Informatics Systems

Weina Fu · Guanglu Sun (Eds.)

e-Learning, e-Education, and Online Training 8th EAI International Conference, eLEOT 2022 Harbin, China, July 9–10, 2022 Proceedings, Part I

Editors Weina Fu Hunan Normal University Changsha, China

Guanglu Sun Harbin University of Science and Technology Harbin, China

ISSN 1867-8211 ISSN 1867-822X (electronic) Lecture Notes of the Institute for Computer Sciences, Social Informatics and Telecommunications Engineering ISBN 978-3-031-21160-7 ISBN 978-3-031-21161-4 (eBook) https://doi.org/10.1007/978-3-031-21161-4 © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 This work is subject to copyright. All rights are reserved 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

We are delighted to introduce the proceedings of the 8th European Alliance for Innovation (EAI) International Conference on e-Learning, e-Education and Online Training (eLEOT 2022). This conference brought together researchers, developers, and practitioners around the world who are leveraging and developing e-educational technologies as well as related learning, training, and practice methods. The theme of eLEOT 2022 was “New Trends of Information Technology and Artificial Intelligence in Education”. The technical program of eLEOT 2022 consisted of 111 full papers, which were selected from 226 submissions following at least three single-blind reviews per paper, including two invited papers in oral presentation sessions at the main conference tracks. The conference tracks were as follows: Track 1 – Information Technology Promoted Teaching: Platforms and Systems; Track 2 – Artificial Intelligence-based Educational Modes and Methods; Track 3 – Automatic Educational Resource Processing; and Track 4 – Educational Information Evaluation. The technical program also featured two keynote speeches, which were “Mining the interesting patterns to aid the education system”, to focus on the advantages of using pattern mining models for knowledge discovery and pattern analysis, as well as how pattern mining helps to improve the effectiveness of studying/learning in an education scheme, by Jerry Chun-Wei Lin from Western Norway University of Applied Sciences, Norway, and “Machine Learning based Small Bowel Video Capsule Endoscopy Analysis: Challenges and Opportunities”, presenting a detailed comparative and critical analysis of existing research methodologies for small bowel capsule endoscopy and evaluating these methods based on the aspects considered significant by clinical experts such as population study, bias, data split type, validation method, and prospective nature of the experiments, by Irfan Mehmood from University of Bradford, UK. Coordination with the steering chair, Imrich Chlamtac, was essential for the success of the conference. We sincerely appreciate his constant support and guidance. It was also a great pleasure to work with such an excellent organizing committee team for their hard work in organizing and supporting the conference. In particular, we are grateful to the Technical Program Committee who completed the peer-review process and helped to put together a high-quality technical program. We are also grateful to Conference Manager Kristína Havlíˇcková for her support and to all the authors who submitted their papers to the eLEOT 2022 conference. We strongly believe that the eLEOT conference provides a good forum for all researchers, developers, and practitioners to discuss all science and technology aspects that are relevant to e-learning and e-education. We also expect that the future eLEOT conferences will be as successful and stimulating as this year’s, as indicated by the contributions presented in this volume. March 2023

Weina Fu Guanglu Sun

Organization

Steering Committee Imrich Chlamtac

University of Trento, Italy

Organizing Committee General Chair Guanglu Sun

Harbin University of Science and Technology, China

Technical Program Committee Chair Gautam Srivastava

Brandon University, Canada

Web Chair Xin Liu

Harbin University of Science and Technology, China

Publicity and Social Media Chair Khan Muhammad

Sungkyunkwan University, South Korea

Special Issue Chair Gautam Srivastava

Brandon University, Canada

Workshop Chair Ao Li

Harbin University of Science and Technology, China

Sponsorship and Exhibits Chair Suxia Zhu

Harbin University of Science and Technology, China

viii

Organization

Publications Chair Weina Fu

Hunan Normal University, China

Demos Chair Song Li

Harbin University of Science and Technology, China

Posters and PhD Track Chair Lili Liang

Harbin University of Science and Technology, China

Local Chair Hongwei Wu

Harbin University of Science and Technology, China

Technical Program Committee Adam Zielonka Amin Taheri-Garavand Arun Kumar Sangaiah Ashutosh Dhar Dwivedi Dan Zhang Dang Thanh Dongye Liu Fanyi Meng Feng Chen Fida Hussain Memon Fuguang Guo Guanglu Sun Hari Mohan Pandey Heng Li Jerry Chun-Wei Lin Jianfeng Cui Keming Mao Khan Muhammad Lei Ma Marcin Wo´zniak Mu-Yen Chen

Silesian University of Technology, Poland Lorestan University, Iran Vellore Institute of Technology, India Technical University of Denmark, Denmark Xinyang Vocational and Technical College, China Hue Industrial College, Vietnam Inner Mongolia University, China Harbin Institute of Technology, China Xizang Minzu University, China Jeju National University, South Korea Henan Vocational College of Industry and Information Technology, China Harbin University of Science and Technology, China Edge Hill University, UK Henan Finance University, China Western Norway University of Applied Sciences, Norway Xiamen University of Technology, China Northeastern University, China Sungkyunkwan University, South Korea Beijing Polytechnic, China Silesian University of Technology, Poland National Cheng Kung University, Taiwan

Organization

Norbert Herencsar Peng Gao Ping Yu Shichen Huang Shuai Liu Shui-Hua Wang Tenghui He Thippa Reddy Gadekallu Uttam Ghosh Weina Fu Wuxue Jiang Xiaochun Cheng Xiaogang Zhu Xin Qi Xinchun Zhou Xinyu Liu Xuanyue Tong Yanning Zhang Yating Li Yudong Zhang Yongjun Qin

Brno University of Technology, Czech Republic Hunan Normal University, China Jilin University, China Hunan Normal University, China Hunan Normal University, China Loughborough University, UK Hunan Normal University, China Vellore Institute of Technology, India Vanderbilt University, USA Hunan Normal University, China The Education University of Hong Kong, Hong Kong Middlesex University, UK Nanchang University, China Hunan Normal University, China Baoji University of Arts and Sciences, China Hunan Normal University, China Nanyang Institute of Technology, China Beijing Polytechnic, China Hunan Normal University, China University of Leicester, UK Guilin Normal College, China

ix

Contents – Part I

IT Promoted Teaching Platforms and Systems Design of Online Teaching Assistant Platform for Technical Courses of Physical Education Specialty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Changmin Lv, Shujun Zhang, Yingying Huang, and Mengyang Liu

3

Design of Online Art Appreciation Course Teaching System Based on Interactive Scene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rong Yu

14

An Empirical Study on the English Teachers’ Informatization Ability in Yunnan Ethnic Minority Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiaojie Ning and Haixia Zhou

28

Design of Aerobics Network Teaching System Based on Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rong Sun and Zhaoqi Fu

39

Design of Online Open Wushu Sanda Teaching Platform Based on Hybrid Reality Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zhaoqi Fu, Dunke You, Wen Liu, and Rong Sun

50

Design of Intelligent Financial Education Assistant System Based on Blockchain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Wen Tang and Zhaohu Zhang

62

Enterprise Strategic Financing Risk Management Auxiliary Education System Based on Data Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Xiang Zou and Jing Zhu

75

Leg Posture Correction System for Physical Education Students Based on Multimodal Information Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Lei Yang and Yueguo Jia

91

Online Interactive Teaching System of Sanda Course Based on P2P Network . . . 103 Yueguo Jia and Lei Yang Intelligent Computer Aided Instruction System Based on Cloud Computing . . . . 118 Shiliang Liu and Caifeng Gao

xii

Contents – Part I

Construction of Online Ideological and Political Education Platform Based on Artificial Intelligence Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Huijuan Li and Xiuying Dong Intelligent Interactive Mobile Teaching Platform in Colleges and Universities Based on Artificial Intelligence Network . . . . . . . . . . . . . . . . . . . 145 Chaojun Zhu Construction of Multimedia Online Education Platform Based on Fuzzy Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 Dan Yu Design and Implementation of Mobile Intelligent Education System Based on Cloud Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 Dan Yu Design of Online Preschool Education Decision Support System Based on Data Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185 Nan Li and Min Li Construction of Online Education Model of Marketing Specialty Based on Cloud Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 Lingli Mao and Zhichao Xu Design of Tibetan Vocabulary Online Learning System Based on Multi-terminal Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 Min Li and Nan Li Design of Online Teaching System for Theory of Variable Order Fractional Gradient Descent Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 Zhichao Xu, Chao Song, Li Li, and Lingli Mao Research on Collaborative Learning Behavior Recognition of Online Education Platform Based on Data Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243 Zhi Zhang Design of Online Education System for Ideological and Political Courses of Traditional Chinese Medicine Based on MOOC Model . . . . . . . . . . . . . . . . . . . 255 Na Zhao and Zhongwen Zheng Assistant Teaching System of Human Resource Management Course Based on Data Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 Ying Ye

Contents – Part I

xiii

Design of Virtual Experiment Online Teaching System for Economics and Management in Colleges and Universities Based on Virtualization Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281 Jianhua Zhang, Yingji Luo, and Qiuhui Yang Design of Online Teaching System Based on Clustering Algorithm . . . . . . . . . . . 296 Xiaobo Xue and Lan Zhang Design of Distance Teaching System for Textile Pattern Design Course Based on Online and Offline Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311 Lan Zhang and Xiaobo Xue Online Training System of Distribution Network Equipment Operation and Maintenance Security Based on Cloud Model . . . . . . . . . . . . . . . . . . . . . . . . . . 324 Lizhen Zhang, Lu Liu, Yuexing Hu, Feng Gao, Wei Jin, and Chaojun Zhu Design of Intelligent Education Management Information System in Colleges and Universities from the Perspective of Big Data . . . . . . . . . . . . . . . . 338 Haiyan Zhao, Shiyuan Liu, Lin Xiao, Kun Liu, and Fuxia Liu Mobile Terminal-Based Remote Counseling Education System for Middle School Students’ Mental Health . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351 Fuyu Du and Jing Zhu Design of Online Teaching System for Architecture Major Based on Virtual Reality Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365 Jing Zhu and Fuyu Du Multimodal Information Processing Method of College English Course Online Education System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378 Baoling Feng and Linan Wang Design of Intelligent Accounting Education System Based on Data Processing Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 391 Yuchan Luo Training Simulation System of Aviation Electromechanical Specialty Based on Multimodal Information Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403 Xin Zhang and Dan Zhao Practical Teaching Platform of Aircraft Maintenance Skills Based on Virtual Reality Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 415 Mingfei Qu and Lirong Zhang

xiv

Contents – Part I

Information Teaching Management System of College Employment Guidance Course Based on Hybrid Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 430 Qinghui Ma, Shuang Wang, and Fei Wang Teaching and Training System of Power Grid Monitoring Technology Based on Multimodal Information Processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 445 Lirong Zhang, Ying Xiao, and Mingfei Qu AI based Educational Modes and Methods Image Watermark Removal Method of Classroom Teaching Recording and Broadcasting System Based on Deep Learning . . . . . . . . . . . . . . . . . . . . . . . . . 463 Yan Chao and Chen Chen Research on Online Digital Painting Creation Education Model Based on Cloud Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474 Rong Yu Intelligent Course Scheduling Method of Single Chip Microcomputer Application Technology Based on Reinforcement Learning . . . . . . . . . . . . . . . . . . 487 Jiaofeng Wu and Weiwei Zhang Intelligent Prediction Method of Online Open Course Passing Rate of Automation Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 499 Weiwei Zhang and Jiaofeng Wu Transmission Line Visual Inspection Method Based on Neural Network Online Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 513 Zhaohu Zhang, Zan Li, and Wen Tang Intelligent Integration Method of Innovation and Entrepreneurship Education Management Information Based on ANNS . . . . . . . . . . . . . . . . . . . . . . . 529 Anteng Xiu Extraction Method of Emotional Elements of Online Learning Text Information Based on Natural Language Processing Technology . . . . . . . . . . . . . 542 Haolin Song, Dawei Song, and Yankun Zhen Simulation Analysis of Entrepreneurial Behavior Selection Mechanism of Higher Vocational College Students Based on Teaching Big Data . . . . . . . . . . 553 Jing Zhu and Xiang Zou College Library Assisted Online Teaching Model Based on Cloud Service . . . . . 565 Caifeng Gao

Contents – Part I

xv

Innovative Application of Big Data Technology in Network Teaching Model of University Courses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 578 Changdong Xu and Song Wang Intelligent Statistical Method of Accounting Information Teaching Data Based on SVM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 589 Chen Chen and Yan Chao Research on Online Teaching Method of Children’s Sports Enlightenment from the Perspective of Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601 Changmin Lv, Shujun Zhang, Yingying Huang, and Mengyang Liu Deep Level Intelligent Mining Method of Online Education Decision Information in Economic Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 615 Jiajie Wang Research on Data Mining Technology of Online Intelligent Education Under Collaborative Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627 Zhi Zhang Resource Assisted Online Teaching Model of University Library in the Context of Ecological Civilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 639 Qu Long and Qiong Hao Online Education and Learning Model of Applied Optics Course Based on Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654 Yankun Zhen and Haolin Song Online Teaching Method of Biochemistry Course of Nursing Specialty Based on Association Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 668 Jin Chen, Kai Yang, Ying Pang, Menglai Shen, Ping Liu, and Jie Lu Music Distance Education Resource Sharing Method Based on Big Data Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 683 Jun Zhou and Hui Lin Real-Time Tracking Method of Students’ Targets in Wushu Distance Teaching Based on Deep Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 695 Jie Zhang and Na Ma Research on Anomaly Detection of Distributed Intelligent Teaching System Based on Cloud Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 709 Fayue Zheng, Lei Ma, Hongxue Yang, and Leiguang Liu

xvi

Contents – Part I

Abnormal Traffic Detection Method of Educational Network Based on Cluster Analysis Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 724 Lei Ma, Jianxing Yang, and Fayue Zheng Simulation Operation System of Civil Aviation Professional Electromechanical Equipment Based on Human-Computer Interaction . . . . . . . . 737 Dan Zhao and Xin Zhang Design of Online Teaching Method for Subject Knowledge of Mathematics Teachers in Higher Vocational Colleges Based on Convolutional Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 751 Chunyan Yu and Junyan Wang 3D Reconstruction Method of Virtual Teaching Laboratory Model Based on Akaze Features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765 Mingxiu Wan and Yangbo Wu Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 779

Contents – Part II

Automatic Educational Resource Processing An Auxiliary Recommendation Method for Online Teaching Resources of Ideological and Political Courses in Colleges Based on Content Association . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hongmei Gu and Lin Chen

3

Design of Multiple Interactive Sharing System for Electric Power Subject Course Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hongmei Lei and Ni Wang

16

In Depth Mining Method of Online Higher Education Resources Based on K-Means Clustering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Anteng Xiu

31

Online Education Resource Recommendation System of International Finance Course Based on Preference Data Collection . . . . . . . . . . . . . . . . . . . . . . . Daifu Qiao

44

Intelligent Sharing Platform of Agricultural Online Education Resources Based on Blockchain Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Song Wang and Changdong Xu

56

Economic Management Course Recommendation Algorithm in Smart Education Cloud Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jiajie Wang

69

Research on the In-Depth Recommendation Method of Grammar Topics for English Online Teaching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Yi Yu and Haijun Wang

81

Design of College Art Education Curriculum Resource Allocation System Based on Virtual Grid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Jie Lu and Jin Chen

94

Control Method of Input Information Amount of Intelligent Education Resource Database Based on Data Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 Xin Wang and Pan Zhang

xviii

Contents – Part II

Design of Multimedia Courseware Synchronous Display System for Distance Teaching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 Pan Zhang, Pan Xu, and Xin Wang Design of Music Teaching Resource Sharing System Based on Mobile Terminal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 Hui Lin, Jun Zhou, and Hongping Huang Remote Sharing System of Chinese Educational Resources Based on Information Fusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 Jiang Cai, Mingming Zhang, and Jingya Zheng Intelligent Education Information Resource Integration and Sharing System Based on Cloud Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 Jingya Zheng, Jichao Yan, and Jiang Cai Collaborative Filtering Recommendation Method for Online Teaching Resources of Elderly Care Specialty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 Wei Chen and Zhixiong Jian Personalized Recommendation System of Ideological and Political Online Teaching Resources Based on Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . 186 Xiuying Dong and Huijuan Li Balanced Allocation of Ideological and Political Network Teaching Resources in Universities Based on Big Data Clustering . . . . . . . . . . . . . . . . . . . . . 199 Lin Chen and Hongmei Gu Research on the Balanced Allocation Model of Online Education Resources of Economics and Management in Colleges and Universities Based on Parallel Clustering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 Yingji Luo and Jianhua Zhang Research on Data Classification of Online English Teaching Platform Based on Deep Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 Li Miao and Qian Zhou Data Optimization Query Method of Online Education System Based on Decision Tree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 Yiqian Zhang and Yue Wang Research on Online Mathematics Teaching Resource Integration Model Based on Deep Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 Yue Wang and Yiqian Zhang

Contents – Part II

xix

Resource Matching Method of Online Education Platform Based on Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270 Qinpei Fan Remote Sharing of National Music Teaching Resources Based on Big Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 Jing Zhan Mining Algorithm of Massive Online Financial Education Resources Based on Apriori TIDS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 293 Yuchan Luo Design of Educational Resource Sharing System for Financial Management Specialty Based on Data Mining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305 Jingbo Li and Meifu Li Personalized Recommendation System of College Students’ Employment Education Resources Based on Cloud Platform . . . . . . . . . . . . . . . . . . . . . . . . . . . . 318 Fei Wang, Yanming Huang, and Qinghui Ma Optimization and Recommendation Method of Distance Education Resources Based on Particle Swarm Optimization Algorithm . . . . . . . . . . . . . . . . 334 Jie Gao and Yi Huang Prediction of Online Learning Resource Demand Based on BP Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347 Yi Huang and Jie Gao Educational Information Evaluation Research on the Evaluation Method of Ideological and Political Effect of Online Courses in Internet of Things . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363 Jiajing Cai, Junying Feng, Jinmei Shi, YaJuan Zhang, Shangyu Meng, and Jianfeng Yao Research on Evaluation Method of College Students’ Innovation and Entrepreneurship Training Mode Based on Deep Neural Network . . . . . . . . . 377 Zhongwen Zheng Evaluation Method of Oral English Digital Teaching Quality Based on Decision Tree Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388 Li Miao and Qian Zhou

xx

Contents – Part II

Research on Quality Evaluation of Accounting Online Education Based on Artificial Intelligence Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401 Meifu Li and Jingbo Li An Analysis on the Training Mode of Master Students in Artificial Intelligence Field for Electronic Information Professional Degree—Take Hunan Normal University as an Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 410 Weina Fu and Shuai Liu The Problems and Analysis of Artificial Intelligence Specialty Construction in Universities Under the Present Situation of Artificial Intelligence Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 423 Weina Fu and Shuai Liu Firmware Security Verification Method of Distance Learning Terminal Based on MD5 Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 436 Ni Wang, Hongbo Yu, Zhongwen Guo, and Hongmei Lei Early Warning System of Computerized Accounting Teaching Data Quality Based on Artificial Intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 451 Chenyuyan Li, Jin Li, and Ying Ye Performance Evaluation Model of Substation Battery Pack Based on New Series Parallel Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 465 Jianfeng Yao, Ping Zhou, Li Huang, and Gaoming Liu An Online Vocal Music Teaching Timbre Evaluation Method Based on Feature Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482 Rui Wang, Jianli Qi, and Daifu Qiao Construction of Teaching Quality Evaluation Model of Hotel Management Specialty Based on Association Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 495 Wen Hua, Yi Liu, and Yipin Yan Performance Appraisal System of Teachers in Higher Vocational Colleges Based on Fuzzy Comprehensive Evaluation Model . . . . . . . . . . . . . . . . . . . . . . . . . 509 Yipin Yan, Fangyan Deng, and Wen Hua Real-Time Collection Method of Learning Status Data in Distance Teaching Based on Internet of Things . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 524 Qiong Hao, Zhiwen Chen, and Qu Long Research on English Translator Speech Recognition System Based on Deep Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 537 Zhiyu Zhou

Contents – Part II

xxi

Evaluation Method of English Online and Offline Mixed Teaching Quality Based on Three-Dimensional Teaching . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549 Zhiyu Zhou Sports Online Intelligent Education Effect Evaluation System Based on Deep Learning Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 562 Na Ma and Jie Zhang Evaluation Method of Teaching Effect of Intelligent Education Model for Electromechanical Equipment Technology Specialty . . . . . . . . . . . . . . . . . . . . 573 Zhixiong Jian, Yonghao Zhao, and Wei Chen Research on Effect Evaluation Method of Ideological and Political Classroom Teaching Reform in Colleges and Universities Based on Particle Swarm Optimization Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 588 Lili Shao and Peng Zang Evaluation Method of Classroom Teaching Quality of Marxist Theory Course Based on Deep Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 601 Peng Zang Research on Teaching Effect Evaluation of Innovation and Entrepreneurship Based on Collaborative Filtering Algorithm . . . . . . . . . . . . 614 Haijun Wang and Yi Yu Design of Teaching Quality Evaluation System for Law Major Education Based on Fuzzy AHP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627 Fuxia Liu and Haiyan Zhao Assessment and Evaluation System of Preschool Education Curriculum Based on Big Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 640 Jing Zhan Application of Artificial Intelligence in Pre-school Education Professional Talent Training in the Era of Big Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 654 Lingjun Meng, Qiyuan Xin, and Qinpei Fan Effect Evaluation of Online Basic Japanese Lessons Based on Data Mining Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 671 Lina Wang and Baoling Feng Research on Online and Offline Mixed Teaching Quality Evaluation of Higher Mathematics Based on Hierarchical Analysis Method . . . . . . . . . . . . . . 685 Junyan Wang and Chunyan Yu

xxii

Contents – Part II

An Intelligent Evaluation Method of MOOC Learning Efficiency Based on Koch Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 696 Yangbo Wu, Mingxiu Wan, Ying Lin, and Giap Weng Ng Author Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 711

IT Promoted Teaching Platforms and Systems

Design of Online Teaching Assistant Platform for Technical Courses of Physical Education Specialty Changmin Lv(B) , Shujun Zhang, Yingying Huang, and Mengyang Liu Zibo Normal College, Zibo 255100, China [email protected]

Abstract. In order to increase the number of online users that can be accommodated by the physical education teaching auxiliary platform, so as to improve the actual teaching ability of physical education technology courses, an online teaching auxiliary platform for physical education professional technology courses is designed. The B/S three-tier architecture is used to provide the necessary conditions for the connection of database application structure and teaching process management module, and complete the overall design of physical education online teaching auxiliary platform. On this basis, set the key session interface, calculate the specific value of the weight of the evaluation index through the way of curriculum standardization, complete the online evaluation of the technical courses of physical education, and then combine the relevant hardware equipment structure to realize the design and application of the online teaching auxiliary platform of the technical courses of physical education. The experimental results show that, compared with the traditional online teaching system, with the application of teaching assistance platform, the number of logged in online users has increased significantly, which can play a strong role in promoting the actual teaching ability of physical education technology courses. Keywords: Physical education · Professional courses · Online teaching · Auxiliary platform · B/S architecture · Database

1 Introduction As the name suggests, online teaching is a teaching method based on the network. Through the network, students and teachers can carry out teaching activities even if they are thousands of miles apart; In addition, with the help of network courseware, students can learn anytime and anywhere, which really breaks the restrictions of time and space. For workplace people with busy work and uncertain learning time, network distance education is the most convenient way of learning [1]. With the rapid development of information technology, especially from the Internet to mobile Internet, a cross time and space way of life, work and learning has been created, which has fundamentally changed the way of knowledge acquisition. Teaching and learning can be free from the © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 3–13, 2022. https://doi.org/10.1007/978-3-031-21161-4_1

4

C. Lv et al.

restrictions of time, space and place, and the channels of knowledge acquisition are flexible and diversified. Online education platform, i.e. online training system, is a tool software for implementing online training and online education. It is a distance online education college that can be customized and expanded by using network technology and software technology. It helps the government, industry or enterprises quickly establish their own proprietary knowledge base system through simple and easy-to-use courseware and test question introduction and production functions, and provides training demand investigation, training goal setting, curriculum system design, training plan management, training process monitoring and assessment evaluation to help customers effectively implement employee training and assessment tasks. With the rapid development of computer technology and broadband network technology, great changes have taken place in people’s work style, life style and learning style. At the same time, a profound change is also quietly carried out in the field of education. Computer network learning goes deep into the school and all aspects of teaching, which has promoted the qualitative change of teaching mode and educational concept. In the traditional teaching mode of physical education, the main task of students is to listen to the teacher’s explanation and complete the cognition of knowledge; The main task of teachers is to study the teaching materials, complete the teaching plan and design the teaching process according to the content of the teaching materials. This model highlights the role and guidance of teaching materials, highlights the dominant position of teachers, and ignores students’ subjective initiative [2]. The online teaching platform, the use of network advantages, according to the main content of physical education classroom teaching, combined with the latest computer, multimedia and other technologies, in the teaching process to create the scene and method to guide students to learn, make the teaching resources in a more diversified, flexible way, vividly displayed. Beyond the limitation of time and space, it provides a platform for students to study independently after class, review and test; Provide a platform for teachers to integrate teaching resources and teach students according to their aptitude; Provide a platform for teachers and students to communicate in time. Based on these innovative points, online network teaching auxiliary platform has become the favored professional course teaching mode of colleges and universities.

2 Overall Design of Physical Education Online Teaching Auxiliary Platform The hardware execution environment of online teaching assistant platform is based on B/S three-tier architecture, and can realize the adjustment and planning of course progress under the action of database application structure and teaching process management module. The specific design methods are as follows. 2.1 B/S Three-Tier Architecture The online teaching auxiliary platform for technical courses of physical education specialty adopts the three-tier architecture design of B/S mode, namely user layer, business layer and data layer. The specific architecture mode is shown in Fig. 1.

Design of Online Teaching Assistant Platform

5

Fig. 1. B/S architecture of online teaching assistant platform

According to Fig. 1, the B/S architecture of online teaching assistance platform can be divided into: User Layer Teachers and students can interact with the system through the presentation layer to complete the required operations. Different users have different permissions, and they can access different interfaces, including administrator interface, teacher interface and student interface. The presentation layer is realized through WebForms of ASP.NET and adopts the way of browser, making the interface friendly and easy to use. Business Layer The business layer is the bridge between the presentation layer and the data access layer. It contains the logic related to the tasks of the core physical education curriculum. It is composed of many modules, which are stored on the server side according to different functions. The business logic layer in this auxiliary platform includes administrator module, teacher module and student module, each module is composed of different contents such as user management, online test and course information management. The business logic layer is connected to the data access layer through ADO.NET controls. Data Layer Data layer is the basis for the realization of online teaching assistance platform, which is composed of user information database, curriculum task formulation database, education process management database, etc. [3]. The user information database mainly includes the basic information of administrators, teachers and students. The curriculum task formulation database provides data support for the online test module, including student numbers, educational course options, etc. The message forum database provides data support for the message forum module, including the content composition of the post number and post type. B/S is a network architecture, also known as browser/server architecture. In this mode, WEB browser is the main application structure of online teaching assistance

6

C. Lv et al.

platform. It unified the client and centralized most of the system functions on the server, greatly simplifying the process of platform development, platform maintenance and platform application. 2.2 Database Application Structure The demand analysis of online teaching assistant platform database is to analyze the needs of teachers and students. It is mainly to collect information, analyze and sort out the information, so as to provide guarantee for the follow-up professional and technical courses of physical education. It is the most difficult and time-consuming stage in the whole design process, but it is the basis of the design process; It is the starting point of database design, but it affects the design of other stages of the whole. It requires a comprehensive and detailed investigation of the specific situation of curriculum application, and then collect the needs, basic data and data flow of teachers and students according to the overall design goal of the auxiliary platform. It includes three stages of demand information collection, analysis and review [4]. The construction of database application structure must follow the following design principles: the teaching auxiliary host must be able to express the needs of teachers and student users in rich languages, including describing the relationship between various things, and can quickly establish the course service mapping relationship between data layer server and business layer server. On the one hand, it can realize the proper arrangement of physical education professional and technical courses, On the other hand, it can also increase the number of real-time connections of online users, so as to effectively ensure the educational effect of online courses. Figure 2 reflects the complete E-R diagram structure of online teaching auxiliary platform database.

Fig. 2. E-R diagram of database of online teaching auxiliary platform

Design of Online Teaching Assistant Platform

7

As can be seen from Fig. 2, the database application structure must have strong adaptability because the physical education online teaching curriculum arrangement always presents a constantly changing existence state. 2.3 Teaching Process Management Module In order to facilitate PE Teachers’ users to manage the teaching process, teachers can build an automatic notification system based on Google calendar on the teaching assistance platform. The specific methods are as follows: ➀ Teachers use Google account to create Google Calendar, add daily teaching arrangements and work items, and set three reminder methods of e-mail, pop-up window and SMS for personal schedule management [5]. ➁ Teachers import contact information from Gmail address book and set different sharing permissions for students and colleagues in the calendar. For example, students only have permission to view teaching arrangements, while colleagues have permission to view and modify work matters. ➂ For student cadres who need to contact frequently for notification, guide them to create Google Calendar and open SMS reminder. In this way, once they encounter notification matters such as teaching arrangement adjustment and homework assignment, teachers only need to modify the schedule content in the calendar, and Google Calendar can automatically send reminder SMS to shared users in advance to realize the automatic notification function [6]. ➃ For work matters such as meetings that require colleagues to participate in cooperation, teachers can send an invitation notice to relevant personnel in advance by “sending it to a friend”. Gmail email will automatically update the calendar content after receiving the confirmation information, which is convenient for teachers to arrange their daily work.

3 Online Evaluation of Technical Courses of Physical Education Specialty 3.1 Session Interface Settings Session interface is the most important interface for the online teaching auxiliary platform of technical courses of physical education specialty. However, in the application platform system, the instantiated session is a lightweight class. Creating and destroying it will not occupy a lot of resources. This is really important in the actual project, because in the client program, session objects may be created and destroyed continuously. If the cost of session is too large, it will have an adverse impact on the system. But it is worth noting that the session object is non thread safe, so in your design, it is best for a thread to create only one session object. Under the function of the online teaching assistant platform for technical courses of physical education specialty, session can be regarded as an intermediate interface between data connection and transaction management. Teachers and users can imagine a session as a buffer of persistent objects. Hibernate can detect the changes of these persistent objects and refresh the database in time [7]. In specific teaching cases, the session interface can also be regarded as a persistence layer manager, because it contains some persistence layer related operations, such as storing persistent objects to the database

8

C. Lv et al.

and obtaining them from the database. Hibernate session is different from httpsession in JSP Application. When the term session must be used, the platform host refers to the session in Hibernate, and all subordinate application nodes can be called user sessions. It is worth noting that session is not thread safe. Therefore, when designing software architecture, multiple threads should be avoided sharing the same session instance. Let r and p represent two different interface setting parameters. For the online teaching auxiliary platform of physical education technical courses, the inequality condition of p > r is always true. ur represents the session interface marking coefficient when the parameter value is r, up represents the session interface marking coefficient when the parameter value is p, i represents the coding characteristic value of physical education professional and technical courses, β represents the practice coefficient of educational courses, and y represents the average value of course practice. By combining the above physical quantities, the session interface setting standard can be defined as:  W =

p 

2 up − ur

r=1 +∞ 

(1) |β · y|

2

i=1

When setting the session interface, it is stipulated that the value of the minimum practical quantity of physical education professional technical courses must be greater than the physical natural number “1”. 3.2 Course Standardization Because the meaning of each curriculum evaluation index is different from each other, and the form of expression is also different. Some are absolute indicators. Some are relative indicators, others are average indicators; The function trend of technical course teaching of physical education specialty is also inconsistent, some belong to positive indicators, some belong to inverse indicators, and there is no comparability between various indicators [8]. If the so-called standardized treatment or dimensionless treatment is not carried out, the comprehensive evaluation cannot be carried out, and the real significance and value of the comprehensive evaluation will be lost. The standardized treatment of teaching evaluation indicators of technical courses of physical education specialty is to eliminate the dimensional (unit) influence of indicators through certain mathematical transformation, that is, to transform indicators with different properties and dimensions into a relative number – quantitative value that can be integrated. Theoretically, the standardized treatment of teaching evaluation indicators of technical courses of physical education specialty is very important, but in practice, people often ignore its importance, ignore the nature and significance of various indicators, simplify the treatment methods, and are used to adopting linear treatment methods. Of course, according to the nature of some indicators, the change of the actual value causes a corresponding proportional change of the quantitative value. It is understandable to use the linear processing method, but for most indicators, the impact of the change of the actual value of the indicators on the change of the quantitative value is not equal.

Design of Online Teaching Assistant Platform

9

Let T represent the unit performance duration of physical education professional and technical course evaluation, χ represent the online teaching coefficient index, λ represent the auxiliary teaching characteristic value, and φˆ represent the online teaching practice expression coefficient of physical education professional and technical course, s1 and s2 represent the practice quantity of two different physical education online teaching courses, s1 = s2 , and for the auxiliary application platform, the inequality condition is always true. With the support of the above physical quantities, the result of course standardization can be expressed as follows by simultaneous formula (2): Q=

ˆ λW − inf(φ)  sup2 (s2 − s1 ) χ · |T |

(2)

In order to ensure the effectiveness of technical courses of physical education specialty, the design of online teaching auxiliary platform must refer to the numerical calculation results of course standardization. 3.3 Evaluation Index Weight In order to quantify the implementation process of technical courses of physical education specialty, the weight of evaluation indicators must establish corresponding quantitative tools. Its foundation is extension set theory, which includes extension set theory, correlation function and extension relationship. In extenics, the concept of “correlation function” is established. Through the correlation function, the degree and change of elements with certain properties can be described quantitatively. That is, the elements belonging to the same domain can also be divided into different levels according to the value of the correlation function, and the relationship of “within class, i.e. same class, i.e. different” can be developed into “within class, i.e. different levels”. In order to reflect this property, the correlation function on the real axis is established, and the concept of the highest application level in the real variable function is extended to the concept of distance, which is the basis for expanding the qualitative description to the quantitative description. Let A represent the transfer amount of physical education courses per unit time, α represent the minimum teaching evaluation index, ω represent the maximum value result of α index, and m and n represent two different quantitative evaluation coefficients, For the online teaching auxiliary platform for technical courses of physical education specialty, the inequality condition of m = n is always true, lm represents the physical education curriculum practice weight value when the quantitative evaluation coefficient is m, ln represents the physical education curriculum practice weight value when the quantitative evaluation coefficient is n, μ represents the quantitative evaluation difference, and ξ represents the scalar of online teaching knowledge, h indicates supplementary education indicators. With the support of the above physical quantities, the calculation result of the evaluation index weight of the online teaching auxiliary platform of physical education professional and technical courses can be expressed as:  2 l − l 2 μ−1 m n Q  K= √  A  α=1 (ξ − 1)2 × h ω 

(3)

10

C. Lv et al.

So far, the calculation and processing of various physical coefficient indexes are completed, and the design and application of online teaching auxiliary platform for technical courses of physical education specialty are realized without considering other interference conditions.

4 Case Analysis In order to verify the practical application value of online teaching auxiliary platform for technical courses of physical education specialty, the following comparative experiment is designed. The experimental environment is Intel Core (TM) i5 [email protected] GHz, the memory is 16 GB, the development language is Python, the programming environment is Eclipse4.2. The specific experimental process is as follows: Step 1: Select two Internet computers with exactly the same configuration as the experimental object; Step 2: Input the new online teaching auxiliary platform into the computer of the experimental group and the traditional online teaching system into the computer of the control group; Step 3: Control the system connection time of the experimental group and the control group to be consistent, so that the students majoring in physical education can freely access the above teaching host; Step 4: Analyze the actual changes in the number of online users in the experimental group and the control group. Figure 3 reflects the numerical recording results of the number of online users under ideal conditions.

Number of online users/ People

500 400 300 200 150 100 50 0

10

20

30

40 50 60 70 Experim ent time/Min

80

90

100

Fig. 3. Ideal number of online users

It can be seen from the analysis of Fig. 3 that before the experimental time reaches 70 min, the number of access users of physical education online teaching platform always keeps increasing, and the unit increase range is exactly the same; When the experimental time is between 70–100 min, the number of access users of physical education online teaching platform always maintains an absolutely stable numerical state; In the whole

Design of Online Teaching Assistant Platform

11

experimental process, the maximum number of sports online teaching platform access users was 300, which increased by 200 compared with the minimum number of 100. Table 1 records the actual changes in the number of online users in the experimental group and the control group. Table 1. Experimental values of the number of online users (group I) Experiment time/Min

Number of online users/Person The experimental group

The control group

10

100

100

20

160

117

30

213

144

40

272

196

50

304

225

60

355

270

70

407

292

80

460

322

90

514

341

100

554

367

It can be seen from Table 1 that during the first group of experiments, the number of users connected to the physical education online teaching platform in the experimental group and the control group kept increasing, but the increase in the experimental group was significantly greater than that in the control group. From the perspective of limit value, the maximum value of 554 in the experimental group increased by 254 compared with the ideal maximum value of 300; Compared with the ideal maximum value of 300, the maximum value of 367 in the control group increased by 67, which was much smaller than that in the experimental group. It can be seen from the analysis of Table 2 that during the second group of experiments, the number of online teaching platform access users in the experimental group and the control group also maintained an increasing numerical change trend, but the overall increase was less than that in the first group. From the perspective of limit value, the maximum value of 503 in the experimental group increased by 203 compared with the ideal maximum value of 300; Compared with the ideal maximum of 300, the maximum value of 304 in the control group increased by 4, and the increase range is still less than that in the experimental group. To sum up, with the application of the new online teaching assistance platform for physical education professional and technical courses, the number of connected online users does show a significant increase in numerical change trend, which can play a strong role in promoting the actual teaching ability of physical education technology courses and meet the actual application requirements.

12

C. Lv et al. Table 2. Experimental values of the number of online users (Group II)

Experiment time/Min

Number of online users/Person The experimental group

The control group

10

100

100

20

154

108

30

204

130

40

251

141

50

302

170

60

343

193

70

397

221

80

426

270

90

485

286

100

503

304

5 Conclusion Compared with the traditional online teaching system, the new online teaching auxiliary platform for technical courses of physical education specialty, with the support of B/S three-tier architecture system, improves the connection ability of database application structure and teaching process management module, and calculates the specific value of evaluation index weight by setting session interface. From the practical point of view, with the application of this new auxiliary platform, the numerical level of the number of online users shows a trend of appropriately increasing, which can effectively improve the actual teaching ability of sports technology courses, play an important role in promoting, and meet the needs of practical application.

References 1. Huang, J.: Successes and challenges: online teaching and learning of chemistry in higher education in china in the time of COVID-19. J. Chem. Educ. 97(9), 2810–2814 (2020) 2. Jeffery, K.A., Bauer, C.F.: Students’ responses to emergency remote online teaching reveal critical factors for all teaching. J. Chem. Educ. 97(9), 2472–2485 (2020) 3. Andrada, A., Murdocca, R.M., Dondo, J., et al.: Didactic prototype for teaching the MQTT protocol based on free hardware boards and node-RED. IEEE Lat. Am. Trans. 18(2), 376–382 (2020) 4. Baker, R.M., Leonard, M.E., Milosavljevic, B.H.: The sudden switch to online teaching of an upper-level experimental physical chemistry course: challenges and solutions. J. Chem. Educ. 97(9), 3097–3101 (2020) 5. Mussig, J., Clark, A., Hoermann, S., et al.: Imparting materials science knowledge in the field of the crystal structure of metals in times of online teaching: a novel online laboratory teaching concept with an augmented reality application. J. Chem. Educ. 97(9), 2643–2650 (2020)

Design of Online Teaching Assistant Platform

13

6. Rodic, M.V., Rodic, D.D.: Plans vs Reality: reflections on chemical crystallography online teaching during COVID-19. J. Chem. Educ. 97(9), 3038–3041 (2020) 7. Hermanns, J., Schmidt, B., Glowinski, I., et al.: Online teaching in the course “Organic Chemistry” for nonmajor chemistry students: from necessity to opportunity. J. Chem. Educ. 97(9), 3140–3146 (2020) 8. Yan, X.X., An, X.W., Dai, W.B., Sun, N.L.: Image segmentation teaching system based on virtual scene fusion. Comput. Simul. 38(4), 331–337 (2021)

Design of Online Art Appreciation Course Teaching System Based on Interactive Scene Rong Yu(B) College of Art and Design, Nanning University, Nanning 530200, China [email protected]

Abstract. The teaching of art appreciation course is quite different from that of other professional courses, and there is a high demand for interaction between teachers and students. In order to solve the defects of the online art appreciation course teaching system and obtain better teaching effect, an online art appreciation course teaching system based on interactive scene is designed. The system hardware design unit includes teaching data processor selection unit, system operation controller selection unit and peripheral circuit design unit. The software module includes student state detection module, interactive scene setting module and database construction module. Through the design and development of the above hardware units and software modules, the operation and application of the online art appreciation course teaching system are realized. The experimental results show that compared with the comparison system without interactive scene, the system response time obtained by using interactive scene design system is shorter and the accuracy of student state recognition is higher, which fully proves that the design system has certain feasibility and effectiveness. Keywords: Interactive scene · Online art appreciation course · Teaching system · Yolo algorithm · Data processor

1 Introduction With the increasing attention of the state to school education and the improvement of campus network facilities, in order to better, more convenient and faster learn and understand knowledge, we should combine the current advantages and change the previous traditional teaching methods to the current intelligent teaching methods, so that the learning efficiency can be continuously improved. In addition, the design and implementation of an intelligent network teaching system for art majors is one of the feasible paths for intelligent teaching in today’s society. Although the network technology has developed rapidly and the network teaching of art majors has also achieved certain results, there is still a big gap compared with the network teaching system of art majors in developed countries in Europe and America. In the more advanced teaching system, students in Europe and the United States have better learning ability, and more and more teachers use online teaching, and the method is more and more suitable for students, so a new feasible idea is proposed to improve the current teaching system. © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 14–27, 2022. https://doi.org/10.1007/978-3-031-21161-4_2

Design of Online Art Appreciation Course Teaching System

15

This topic aims at the network education of an art appreciation course, which has great particularity in the professionalism of the course, the focus of teaching and the cultivation of professional knowledge. In the past, the guidance method of an art appreciation course basically adopted one-to-one method, which not only limited the utilization rate of teachers’ time, but also students had to spend a lot of time hiring teachers to provide guidance for themselves. In addition, it was difficult for teachers to pay attention to students’ teaching in their spare time. Luo Lili [1] takes students as the main body, integrates the two modes of network teaching and traditional teaching, and combines the design of database to realize the function of online and offline hybrid intelligent auxiliary teaching system, which can meet the requirements of autonomous learning and improve students’ learning enthusiasm. However, in practical application, the system response time is long, and the accuracy of student state recognition is low. Therefore, it is proposed to adopt a more intelligent way to replace the existing teaching methods. Under this condition, teachers can arrange their own teaching tasks according to everyone’s common learning time, and timely count the students’ understanding of a module, so as to increase the corresponding teaching proportion to the module. In addition, according to the comprehension level of most students listening to the class, the courses are divided into key points and difficult points, which can not only maximize the learning efficiency for students, but also allow students to obtain the greatest knowledge reserve effect within a limited learning time. Therefore, according to the actual situation of students, it is imperative to develop an intelligent online teaching system suitable for art appreciation courses. Judging from the existing research results, the existing online art appreciation course teaching system has certain defects and cannot obtain better teaching effects. Therefore, an online art appreciation course teaching system design research based on interactive scenes is proposed. While completing the task of knowledge explanation, the online teaching teacher is allowed to conduct real-time teaching interaction with students in class, and give targeted feedback according to the real-time performance of the whole class and individual students, so that the teaching teacher can directly participate in the students’ question answering and puzzle solving, and the tutor is only responsible for solving unexpected emergencies in class. Through such teaching methods, make full use of teaching teacher resources, let thousands of students receive equal education regardless of location, time and manpower, let students really participate in classroom learning, experience the attention of teachers, and stimulate students’ learning interest and learning autonomy while enhancing their dependence on teachers. After all, interest is the best teacher. At the same time, it is hoped that the development of the system will truly popularize high-quality teaching resources all over the country, so that students can gain in class, so that teachers can better prepare lessons according to students’ feedback in each class, solve the waste of time and resources caused by repeated teaching without gain, strengthen the sense of dependence and trust between teachers and students, and improve the teaching effect of art appreciation course.

16

R. Yu

2 Hardware Design of Online Art Appreciation Course Teaching System Hardware design is the basis and premise of system operation. In order to improve the problems existing in the existing system and improve the teaching effect of art appreciation course, the hardware unit of the system is designed, mainly including the selection unit of teaching data processor, the selection unit of system operation controller and the design unit of peripheral circuit. The specific design process is as follows: 2.1 Selection Unit of Teaching Data Processor Considering the function, real-time and flexibility required by the design system, this study has the following requirements for the selected teaching data processor: (1) Can run embedded Linux system: The operating systems that the selected design system runs are Linux system and real-time kernel. Therefore, when selecting a teaching data processor, it is required that the selected processor can run the embedded Linux system smoothly, so as to provide guarantee for the operation of the design system. (2) With hardware floating-point arithmetic capability: Because the advantage of the ARM processor lies in the aspects of transaction control and task scheduling, the data computing capability is not as good as that of the DSP processor. However, in order to reduce the structural complexity of the design system, this study did not introduce a DSP processor. At this time, it is necessary to select an ARM processor that supports hardware floating-point arithmetic. (3) Development platform, customizability and portability: Some ARM chip programming requires the chip provider to provide professional development software and programming tools, and its programming development relies on the compiler, debugger, etc. provided by the supplier, resulting in the developed software not having customizability and low portability. The portability of the design system software among the hardware platforms is a very important issue. The reasonable selection of the hardware platform with good compatibility is very helpful to improve the portability of the developed software and bring great convenience to the upgrading and maintenance of the hardware platform in the future [2]. Considering the above factors, the design system chooses to use the (ARM) BCM2835 chip produced by Broadcom as the teaching data processor. The selected BCM2835 chip belongs to the ARM11 architecture microprocessor and adopts the BGA package, and the hardware design technical requirements are very high [3]. The BCM2835 chip structure is shown in Fig. 1.

Design of Online Art Appreciation Course Teaching System

17

Fig. 1. BCM2835 chip structure diagram

As shown in Fig. 1, the BCM2835 chip has a relatively simple structure and rich internal resources, which can provide great convenience for the development of the design system. 2.2 System Operation Controller Selection Unit According to the design system requirements, Samsung’s S3C4510B is selected as the system operation controller, which has the advantages of low power consumption and high performance, and is most suitable for system applications that are sensitive to price and power consumption [4]. The S3C4510B controller features are described as follows: One is the architecture. With integrated system, full 16/32 RISC architecture, support for big and little endian mode, debugging scheme based on JTAG interface, boundary scan interface, etc.; The second is the system manager. External bus controller with bus request/response pins and support for EDO/conventional or SDRAM memory, while also being able to program access cycles; The third is the integrated instruction/data Cache. Cache can be configured as internal SRAM and supports LRC replacement algorithm. In addition, the S3C4510B controller has abundant pins to support it to complete a variety of functions. Pin related definitions are shown in Table 1. Table 1. S3C4510B controller pin definition table Pin

Signal

Describe

3

nUADTR1

UART1 data terminal ready

4

UATXD1

UART1 data transmission

5

NUADSR1

UART1 data device is ready (continued)

18

R. Yu Table 1. (continued)

Pin

Signal

Describe

17

nDTRB

HDLC Ch-B terminal ready

19

nRTSB

HDLC Ch-B transfer request

23

nCTSB

HDLC Ch-B transfer clear

30

RXD_10M

10M receive data

36

RX_ERR

Receive error

45

TX_ERR

Send error

55

FILTER

Ceramic capacitor

58

TCK4

Test clock

59

TMS

Test mode selection

60

TDI

Test data input

61

TDO

Test data output

62

nTRST

Reset signal

Different from other controllers, the connection method of the address bus of the S3C4510B application system is relatively simple. Since S3C4510B uses a 32-bit address bus, all addresses can be regarded as byte addresses. The address bus provides a 4GB linear addressing space. When a word access signal is issued, the storage system ignores the lower 2 bits A[1:0]. When signaling a halfword access, the memory system ignores the lower A[0] bits. Therefore, the S3C4510B generates components through an onchip address bus to hide the process, and only needs to connect the address bus of the

Fig. 2. Schematic diagram of S3C4510B address bus conversion

Design of Online Art Appreciation Course Teaching System

19

S3C4510B with the address bus of the memory one by one. The S3C4510B address bus conversion is shown in Fig. 2. Through the above process, the selection and configuration of the system operation controller is completed, which provides effective help for the development and application of the design system. 2.3 Peripheral Circuit Design Unit Peripheral circuit is one of the key components of designing system hardware and undertakes the important task of connecting hardware units. Due to space limitations, only the switching signal input and output circuits are shown [5]. The switch signal input and output circuit is shown in Fig. 3.

Fig. 3. Schematic diagram of switching signal input and output circuit

The above process completes the design and selection of the hardware unit of the design system, but still cannot realize the teaching of the online art appreciation course, so the system software module is designed and developed.

3 Software Design of Online Art Appreciation Course Teaching System The design system software module includes student state detection module, interactive scene setting module and database construction module. The specific design process is as follows: 3.1 Student Status Detection Module In class state detection such as students’ concentration and hands raising is one of the preconditions for completing interactive scenes and improving the teaching effect of

20

R. Yu

art appreciation course [6]. Aiming at the student state detection and classification, the design system uses the widely used Yolo algorithm to complete the student state detection. In class, the system camera recording will be turned on to record students’ behavior in class. Firstly, the video in class will be processed in frames, and the output student picture set will be input into Yolo model for target state detection and positioning. The detection targets are divided into positive targets, negative targets and hand raising targets, in which positive targets represent students’ focused state and negative targets represent students’ non focused state. Finally, the location of the detection target is written into the database for system call. Yolo algorithm steps are as follows: Step 1: Input image. After image preprocessing, output 300 × An image of 300 pixels is used as the input of the algorithm; Step 2: Feature extraction. The layers and sizes of feature maps selected by Yolo algorithm are shown in Table 2. Table 2. Level and size of feature map Layers of feature maps

Sizes of feature maps

Conv4_3

38 × 38

Conv7

19 × 19

Conv8_2

10 × 10

Conv9_2

5×5

Conv10_2

3×3

Convll_2

1×1

According to the different layers of the feature map in Table 2, a priori box with the same size is set on the same feature map, and the size of the priori box in each layer of the feature map increases layer by layer to match the real target. There are n×n center points in the feature map of each layer, and each center point will generate a default box of different size. For example, each center point of the Conv4_3 layer uses 4 default boxes, then the calculation method of the number of a priori boxes generated by this layer is 38 × 38 × 4, and the calculation result is 5776 a priori boxes. By analogy, a total of 8732 a priori boxes are generated in the six layers. Each a priori box corresponds to a set of bounding boxes, which are convolved through two different 3 × 3 convolution kernels to output a set of independent detection values, including the confidence of the category to which the detection target belongs and the position information of the bounding box; Step 3: Loss calculation. It is calculated by weighting the category loss function and the position loss function; Step 4: Model training [7]. In the training process, the ratio of the area of the intersection of the two boxes to the sum of the areas of the two boxes is denoted as IoU . For each real box in the picture, find the a priori box with the largest IoU to match, and ensure that at least one first the test box matches the ground-truth box. The bounding boxes

Design of Online Art Appreciation Course Teaching System

21

corresponding to such a priori boxes are used as positive samples for training, and the remaining unmatched a priori boxes are used to calculate the IoU value between them and each real box. Generally, the threshold of IoU is set to 0.5. If the IoU value is greater than the threshold, the a priori frame is matched with the corresponding real frame with the largest IoU value, and for the a priori frame whose IoU value and all real frames are less than the threshold, such a priori frame is used as a negative sample; Step 5: Prediction process. Preprocess the picture to 300 × 300 pixels input Yolo algorithm to predict the category confidence and location information of each a priori box, filter out the a priori boxes whose prediction result is background or whose confidence is less than the threshold, and the remaining a priori boxes are arranged according to the confidence. Decode the first 400 a priori frames with the highest confidence to the position in the original image, and screen out the intersecting a priori frames with IoU greater than 0.5 through non maximum suppression processing to obtain the final classification target, including the category and position information of the detection target. In the Yolo algorithm shown above, the core step is loss calculation. The calculation formula is: 1 (1) L(x) = (Lconf (x) + αLloc(x)) N In formula (1), L(x) represents the loss value; N represents the number of a priori boxes; Lconf (x) represents the classification loss function; α represents the default value of 1; Lloc(x) represents the position loss function. In formula (1), the calculation formula of the classification loss function Lconf (x) is: Lconf (x) = −

N 

 p   0  p log cij xij log ci −

(2)

i=1 p

In formula (2), xij indicates that the i a priori box matches the real box of the j category p, and a value of 0 indicates that no a priori box matches the real box; p indicates the p type of target; ci represents the probability that the i prediction box is of type p. In formula (1), the calculation formula of the position loss function Lloc(x) is: Lloc(x) =

N  

  xijk smoothL1 lim − gjm

(3)

i=1

In formula (3), k represents the type of object; smoothL1(·) represents the position determination function; lim and gjm represent the width and height of the prior frame, respectively. Through the above process, the detection of student status can be completed, which provides effective support for the realization of online art appreciation course teaching. 3.2 Interactive Scene Setting Module The interactive scene setting module mainly sets the interactive process of online course teaching accordingly. During the teaching process, the lecturer interacts with students

22

R. Yu

through interactive components. The types of interaction include order maintenance of the lecturer, question and answer interaction between teachers and students, question and answer interaction between students, volume feedback interaction, preferred screen projection interaction and group PK interaction [8]. Among them, the business process of order maintenance of the lecturer is as follows: The lecturer triggers the classroom order maintenance component according to the students’ concentration before and during the class, and the system detects the students’ concentration every 3 s through the camera in the classroom. The lecturer adjusts the strategy of class order maintenance according to the proportion of students whose concentration is not up to standard. If more than 30% of students are identified as not paying attention in class, the lecturer will maintain the overall order of the class, otherwise remind the students to take class seriously through the buzzer of the answering machine. The business process of question and answer interaction is: For the question and answer interaction strategy, when the teacher calls the roll, in order to give all students a chance to answer questions, give priority to the students who have not answered the questions, and then choose the students who answer the questions less and raise their hands less. The system first detects whether someone raises his hand. If a student raises his hand, the teacher selects the student who raises his hand less often to answer. If no one raises his hand, the teacher randomly selects a student to answer. If the first student answers wrong, in order to test the overall effect of students’ listening, the teacher asks another student to answer the question in the same way. If the second student still answers wrong, the teacher explains the knowledge point in detail, and then praises the students who answer correctly. For the volume interaction strategy, the teacher asks questions to individuals or groups, and makes corresponding verbal feedback according to the volume of students. If the teacher asks questions to the whole class, the lecturer judges the volume of students’ answers according to the system radio equipment in the classroom, divides the volume of students into low volume, medium volume and high volume, and starts different strategy branches according to the volume feedback, including praising students, encouraging students or explaining problem strategies. For the loud feedback, the system starts the praise script, and the speaker teacher praises the students, such as: “the students answer loudly, so keep it up”. In response to the feedback of medium volume, the system activates the students’ encouraging words, and the lecturer encourages the students, such as: “The students are not very loud, please read it again with the teacher”, and the students follow the second reading. In response to the low volume feedback, the system judged that the students did not understand the content of the question, and the lecturer re-explained the relevant knowledge and asked the question again in the same way. If the lecturer asks a question to the individual, the system will record the sound through the clicker, and the teacher will give feedback according to the received volume. The general question and answer mode is generally used in the follow-up questioning of English teachers, and the individual question mode is generally used in personal reading in English class [9]. Due to space limitations, this study only shows the interactive process of optimal screen projection, as shown in Fig. 4.

Design of Online Art Appreciation Course Teaching System

23

Fig. 4. Schematic diagram of the preferred screen projection interaction process

The above process completes the setting of the interactive scene and helps the design system to complete the teaching interactive task, so as to improve the students’ interest in learning and enhance the teaching effect of the online art appreciation course. 3.3 Database Building Blocks The database is mainly responsible for the recording and storage of teaching data and system operation data, and is the basis for the stable operation of the design system. The database is mainly constructed in the form of logical tables. Due to space limitations, only some of the logical tables are displayed, as shown in Table 3. Through the design and development of the above hardware units and software modules, the operation and application of the online art appreciation course teaching system are realized, which provides help for the development of the online art appreciation course teaching.

4 Experiment and Result Analysis In order to verify the application performance of the above design online art appreciation course teaching system, a simulation comparison experiment is designed. The specific experimental process is as follows:

24

R. Yu Table 3. Part of the database display table

Field name

Field code

Type

U_Id

Number

Int

U_name

User name

char

U_ema

Mailbox

char

U_pw

Password

char

R_id

Role

Int

P_id

Pass or not

Int

S_sex

Gender number

char

S_add

Address

char

S_date

Time

Int

S_ph

Telephone

Int

4.1 Experimental Preparation Stage The experimental preparation stage mainly undertakes the task of formulating the experimental data acquisition process. On this basis, it obtains accurate experimental data and prepares for the analysis of subsequent experimental results. The experimental data is mainly the feedback results of course teaching. The feedback results of course teaching include the process after the teacher explains the teaching content. The core process includes the implementation of student course evaluation, student after-school test, in class report generated by the system according to students’ performance, parents’ message, parents’ praise and other processes. After submitting the course evaluation, students need to take a classroom test to test the students’ listening effect. Students shoot the problem-solving process through the question answering camera and upload it to the background of the system for problem judgment. The system will generate students’ in class report according to the students’ behavior records in class. After the tutor confirms that it is correct, upload the students’ report. After parents log in to the system, they can download it on the parents’ service page or view the in class report online. You can also leave messages, comments and likes in the parent comment area, and give feedback to teaching institutions while parents communicate with each other, so that institutions can understand the real needs of parents and improve teaching development strategies in time. The feedback logic of course teaching is shown in Fig. 5. The above process completes the preparation of experimental data and provides support for the subsequent online art appreciation course teaching experiment. 4.2 Analysis of Experimental Results In order to quantify the performance of the online art appreciation system and show the results as follows: The system response time data obtained through experiments are shown in Table 4.

Design of Online Art Appreciation Course Teaching System

25

Fig. 5. Schematic diagram of curriculum teaching feedback logic

Table 4. Data sheet of system response time Number of experiments

Application interactive scene design system

Interactive scene comparison system is not applied

1

3.56s

6.98s

2

4.02s

7.15s

3

2.19s

7.58s

4

3.57s

8.50s

5

3.20s

9.45s

6

3.66s

6.20s

7

4.59s

7.24s

As shown in Table 4, with the increase of test times, the system response time of the design system and the comparison system fluctuates. The response time of the design system is between 2.19s and 4.59s, the response time of the comparison system is between 6.20s and 9.45s, and the longest system response time difference is 4.86s. The response time of the application interactive scene design system is shorter than that of the non application interactive scene comparison system, indicating that the design system has stronger real-time performance. The accuracy of student state recognition is obtained through the experiment, as shown in Fig. 6.

26

R. Yu

Fig. 6. Data chart of student state recognition accuracy

As shown in Fig. 6, the accuracy of the application of interactive scene design system is generally higher than that of the non application of interactive scene comparison system. When the number of tests reaches 10, the accuracy of the non application of interactive scene comparison system is 10%, and the accuracy of the application of interactive scene design system is 40%, with a difference of about 30%, compared with the non application of interactive scene comparison system, the student state recognition accuracy of the application of interactive scene design system is higher. The above experimental data show that compared with the non application of interactive scene comparison system, the application of interactive scene design system obtains shorter system response time and higher accuracy of student state recognition, which fully proves that the application performance of the design system is better.

5 Conclusion In order to solve the defects of the online art appreciation course teaching system and obtain better teaching results, an online art appreciation course teaching system based on interactive scenes is designed. Using Linux system, ARM processor, bcm2835 chip and Samsung S3C4510B as system operation controller can reduce power consumption and improve detection accuracy. Yolo algorithm is used to detect students’ status and improve the effectiveness of online art appreciation course teaching. Set up the interactive scene module, increase the interaction between teachers and students, mobilize students’ interest in learning, and improve learning efficiency. Build the database module, record and store the teaching data and system operation data, so that the design system is feasible. After experimental verification, the design system can greatly shorten the system response time, improve the accuracy of students’ state recognition, provide a more effective system support for the development and implementation of online art appreciation teaching, and also provide a certain reference for the related research of the teaching system. However, due to the limited research time, the number of tests in this paper is less, which may produce a little error in practical application. It is necessary

Design of Online Art Appreciation Course Teaching System

27

to increase the number of tests in the follow-up work practice, improve the system, and better serve users. Fund Project. Nanning College’s 2021 Professor Cultivation Project “Research on Cultural and Creative Design and Artistic Creation of Weizhou Island’s Characteristic Landscape Resources”, Project Number: 2021JSGC22.

References 1. Lili, L.: Design of Online and Offline Hybrid Intelligent Assistant Teaching System. Microcomput. Appl. 35(08), 105–108 (2019) 2. Gang, L., Fang, W.: Research on art innovation teaching platform based on data mining algorithm. Clust. Comput. 22(2), 13867–13872 (2019) 3. Danaher, M., Schoepp, K., Kranov, A.A.: Teaching and measuring the professional skills of information technology students using a learning oriented assessment task. Int. J. Eng. Educ. 35(3), 795–805 (2019) 4. Long, Y.: Research on art innovation teaching platform based on data mining algorithm. Clust. Comput. 22(6), 14943–14949 (2019) 5. Eberle, J., Hobrecht, J.: The lonely struggle with autonomy: a case study of first-year university students’ experiences during emergency online teaching. Comput. Hum. Behav. 121(3), 106804 (2021) 6. Dama, C., Langford, M., Dan, U.: Teachers’ agency and online education in times of crisis. Comput. Hum. Behav. 121(3), 106793–106793 (2021) 7. Boccia, M., Guariglia, P., Piccardi, L., et al.: The detail is more pleasant than the whole: global and local prime affect esthetic appreciation of artworks showing whole-part ambiguity. Atten. Percept. Psychophys. 82(2), 3266–3272 (2020) 8. Chen, C., Pan, Y., Li, D., et al.: A virtual-physical collision detection interface for AR-based interactive teaching of robot. Robotics Comput. Integrated Manufacturing 64(2), 101948 (2020) 9. Liu, P., Su, J., Zhang, L., et al.: Simulation of balanced allocation of database MAR based on ALAD algorithm. Comput. Simul. 38(09), 420–423+443 (2021)

An Empirical Study on the English Teachers’ Informatization Ability in Yunnan Ethnic Minority Area Xiaojie Ning(B) and Haixia Zhou School of Foreign Languages and Literature, Yunnan Normal University, Kunming, China [email protected]

Abstract. This paper studies the teachers’ informatization ability by sampling 1508 English teachers from secondary schools in the eight ethnic minority prefectures of Yunnan Province. The questionnaire data collected were analyzed with SPSS 22.0. The questions addressed were 1) The current situation of English teachers’ information-based teaching ability; 2) The relationship between English teachers’ information-based teaching ability and teaching level as well as English proficiency. Thus, this study came to the conclusion, namely, 1) The informationbased teaching of English teachers in the minority areas of Yunnan Province is in a moderate level. 2) There is a significant positive but weak correlation between informatization teaching ability and teaching level and English proficiency. Keywords: English teacher · Information teaching ability · Yunnan ethnic minority areas

1 Introduction This paper aims to explore the information-based teaching ability of English teachers in minority areas of Yunnan Province. Through the investigation, this paper explores the current situation of English teachers’ information-based teaching ability in the eight minority prefectures of Yunnan Province. The 21st century is an information age. With the rapid development of knowledge economy in China, high-tech and information technology have penetrated into all aspects of human’ lives, bringing a lot of convenience to communication. Nowadays, information technology and education are increasingly integrated. In 2010, the Chinese government officially issued the Outline of the National Medium and Long Term Reform and Development Plan (2010–2020). Chapter 19 of the document clearly proposes to “speed up the process of educational informatization”. The Implementation Plan of Education Poverty Alleviation in Yunnan Province (2017) clearly states that it is necessary to strengthen the construction of teachers in rural areas, accelerate the promotion of compulsory education informatization, and basically achieve the full coverage of multimedia teaching in compulsory education schools by the end of 2020. Therefore, information technology © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 28–38, 2022. https://doi.org/10.1007/978-3-031-21161-4_3

An Empirical Study on the English Teachers’ Informatization Ability

29

has a revolutionary impact on the development of education, which must be paid high attention to. Information teaching means that teachers can use information teaching means reasonably to complete teaching tasks with high quality. At this stage, education informatization has come to a new stage and micro class, flipped classroom, network teaching and research all reflect the integration of modern information and education, update education ideas, innovate teaching methods, and improve teaching efficiency (Zhang, 2014). Since the first year of webcast in 2016, information-based teaching has made new progress. The popularity of the novel coronavirus in 2020 has pushed the information technology to the peak. Under the background of educational informatization, the integration of information technology and disciplines has become the trend of education development (Liu, 2007). Because of geographical location, historical development and other factors, there are great differences among different regions in China. These differences are mainly reflected in economy, culture and education. It mainly manifested in the differences between coastal and inland areas, eastern and western regions, urban and rural areas, non-minority areas and ethnic minority areas. From the perspective of education, there is a big gap in the development of basic education in China, which mainly displays that there is the better development of education informatization, complete teaching resources and accessories, and moderate development in the central region. The development of educational informatization in west China is poor, and the teaching resources are insufficient, which is urgent to be improved. The research on the teacher’s information-based ability in China only started 20 years ago. The relevant research mainly focuses on the concept, current situation and development of teachers’ information-based teaching ability. However, there is a lack of empirical quantitative questionnaire survey. In the past, scholars focused on the research of teachers’ information-based teaching ability in the developed areas, but few in remote areas, especially in ethnic minority areas. Besides, there is no research on the relationship between teachers’ information ability and demographic variables as level of teaching ability, English proficiency etc. Most of the previous studies related used only the descriptive statistics of SPSS and few measured the data with the programs of SPSS such as correlation and regression.

2 Methodology 2.1 Research Questions This research attempts to ask the following questions: 1) What is the general situation of information teaching ability in the ethnic minority area in Yunnan? 2) What is the relationship between information teaching ability and English teaching level and English proficiency? 3) Are there any important predictors of information teaching ability which affect the teachers’ English teaching level and English proficiency?

30

X. Ning and H. Zhou

2.2 Sampling and Samples In this study, convenience sampling was used. The sample teachers came from the eight ethnic minority prefectures of Yunnan Province with a total of 1518 teachers of English as shown in Table 1. Table 1. Personal information. N Frequency Gender Age

Male

Percent

201

13

Female

1317

87

20–30

466

31

31–40

664

44

41–50

351

23

37

2

757

64

51or above School location

Junior high school Senior high school

419

36

Years of teaching

1–5 year

360

24

6–10 year

349

23

11–20 year

520

34

21–30 year

257

17

32

2

31 years or above Education

Postgraduate Graduate Senior college

Academic ranking title

3 93

70

5

Secondary vocational education

2

0

Professor

3

0

287

19

Lecturer

806

53

Assistant

333

22

Senior

None Major

41 1405

English Non-english

89

6

1381

91

137

9 (continued)

An Empirical Study on the English Teachers’ Informatization Ability

31

Table 1. (continued) N Frequency School location Post

Percent

City

750

49.41

Country

768

50.59

Full time

1344

88.54

Part-time

131

8.63

34

2.24

9

0.59

Special post Substitute

Note: Special post teachers is a special policy implemented by the central government for rural compulsory education in central and western regions. Through the open recruitment of college graduates to teach in rural schools at or below the county level in the central and western regions, we should guide and encourage college graduates to engage in rural compulsory education to innovate the supplementary mechanism for teachers in rural schools

2.3 Instruments Questionnaire Questionnaire is currently the most common method for evaluating TPACK. This questionnaire adapts five-point, seven-point or nine-point Likert scale. According to The Standard of Educational Technology Competence for Primary and Secondary School Teachers issued by the Ministry of Education (2014), and the questionnaire of Wang (2009) Doctoral Dissertation “Research on the development of teachers’ informationbased teaching ability” and Dang’s (2013) thesis “Analysis and Countermeasures of Middle School English Teachers’ Information-based Teaching ability”, the author modified the questionnaire according to the research needs. The questionnaire was divided into two parts. The first part is personal information including teaching age, education background, professional title, age, gender, major. Besides, the English proficiency and teaching level are included. The second part is about the current situation of teaching information. In this part, Five-point Likert Scale was adopted, which ranges from strongly disagree to strongly agree (1 = strongly disagree, 2 = disagree, 3 = uncertain, 4 = agree, 5 = strongly agree). The questionnaire consists of six factors, including teaching monitoring ability, teaching implementation ability, teaching cognition ability, teaching design ability, teaching research ability, and teaching innovation ability. There are 30 question items in the questionnaire, of which number 6, number 26 and number 28 are reverse questions. More details as follow. The Cronbach’s Alpha Reliability Statistics of this questionnaire is .945, which is regarded as very high (Table 2).

32

X. Ning and H. Zhou Table 2. Information of questionnaire

Factors

Number

Teaching cognitive ability

1,2,5,7,12

Teaching design ability

3,4,8,11,20

Teaching implementation ability

6,9,10,13,21

Teaching monitoring ability

14,16,18,19,24

Teaching research ability

15,17,22,26,28

Teaching innovation ability

23,25,27,29,30

2.3.1 Self-Rated English Proficiency and Self-Rated Teaching Ability of the Subjects Table 3. Self-rated English proficiency and teaching ability of the subjects English proficiency Frequency Poor

Teaching ability Percent

Frequency

Percent

43

2.83

19

1.25

Average

540

35.57

447

29.45

Good

812

53.49

885

58.30

Excellent Total

123 1518

8.10 100

167 1518

11.00 100

Table 3 presents the participants’ self-evaluation of English proficiency and teaching ability. More than half of participants choose good in English proficiency and teaching ability, about 30% participants consider their English and teaching are average level. On the whole, most English teachers rate their English proficiency and teaching ability are good, but some English teachers evaluate their English proficiency and teaching ability are average. 2.3.2 Data-Analysis The quantitative data collected were treated with SPSS22.0 1. Frequency was used to measure the personal information. 2. Descriptive Statistics was utilized to find out the current situation of English teachers’ information ability. 3. Spearman Correlation Coefficient was employed to test the relationship between information teaching ability and teaching ability and English proficiency of the subjects.

An Empirical Study on the English Teachers’ Informatization Ability

33

4. Linear Stepwise Regression was employed to find out the important predictors of information teaching ability which affect teaching ability and English proficiency of the subjects.

3 Results and Discussions 3.1 The General Situation of the Subjects’ Teaching As is shown in Table 4, this study carried out descriptive statistics analysis on the overall factors of the teachers’ information-based teaching ability of the English teachers in the eight prefectures of Yunnan Province. Table 4. Descriptive analysis of five factors of information ability Descriptive Statistics N

Mean

Std. deviation

Monitor

1518

3.83

0.85

Research

1518

3.42

1.05

Cognition

1518

4.03

0.67

Implementation

1518

3.85

0.79

Design

1518

4.01

0.79

Table 4 shows the mean and Std. D. of the five factors. The scores of each factor are ranked from the highest to the lowest as follows: cognition (Mean = 4.03, Std = 0.67), design (Mean = 4.01, Std = 0.79), implementation (Mean = 3.85, Std = 0.79), monitor (Mean = 3.84, Std = 0.85), research (Mean = 3.42, Std = 1.05). It means that the overall level of information-based teaching ability of English teachers in the eight prefectures surveyed is relatively high. The subjects’ ability of information monitoring, cognition, application and design is relatively strong, but the ability of information research is intermediate. At present, the middle and high school English teachers had a strong sense of information-based teaching, and were willing to apply information technology to their English teaching. This result confirms the theory that teachers are expected to have an ability to conduct academic research in terms of analysis, design, development, implementation, and evaluation of technology integrated curriculum (Ge and Han, 2017). Schmidt (2009) conducted a research of 124 pre-service teachers in the United States and the questionnaire had seven dimensions which was the same as those put forward by Mishra and Kohler in 2005. The seven dimensions are pedagogical knowledge, technical knowledge, content knowledge, pedagogical content knowledge, technology content knowledge, technology pedagogical knowledge, and technology pedagogical and content knowledge. This questionnaire completed and detailed the items by organizing the items in a logical way. It is

34

X. Ning and H. Zhou

one of the most widely used Technological Pedagogical Content Knowledge (TPACK) questionnaire and it has laid the foundation for many researches. Although teachers hold a positive attitude to the application of information technology in teaching, their own information teaching concept and skills are still lagging behind, and the application of information technology only stays on the use of basic functions (Li, 2009). This result confirms the previous theory and has a certain guiding significance for improving information teaching ability and English teaching in ethnic minority areas of Yunnan Province, namely; the research ability of the subjects is still relatively weak, which needs to be further improved. Research ability of English teachers in Yunnan minority areas is relatively weak, in the future studies, it is necessary to strengthen and enrich the research in this field, so as to provide more support for the informatization research ability. For educators and education managers, it is urgent to improve the ability of information technology research, and take some necessary measures, such as providing targeted information research training and effective online learning platform. In the future studies, ethnic minority teachers can also be taken as an independent sample to make clear the ability of English teachers in trilingual context and the relationship between trilingual and informatization. 3.2 The Relationship Between Information Teaching Ability and English Teaching Ability and English Proficiency 3.2.1 The Relationship Between English Proficiency and Information Teaching Ability

Table 5. Spearman coefficient correlation between information teaching ability and English proficiency Correlation Monitor English proficiency

Spearman correlation Sig.(2-tailed)

.245* 0.00

Research .10* 0.00

Cognitive .264** 0.00

Implementation .248** 0.00

Design .252** 0.00

Spearman Coefficient Correlation analysis of the relationship between the information teaching ability and English proficiency is displayed in Table 5. The correlation suggests that there was a significant, positive but relatively weak relationship between English proficiency and the five factors, i.e., monitor ability (r = .245, p = .000 < 0.01), research ability (r = .10, p = .000 < 0.01), cognitive ability (r = .264, p = .000 < 0.01), implementation ability (r = .248, p = .000 < 0.01) and design ability (r = .252, p = .000 < 0.001). It means the subjects’ information teaching ability and English proficiency positively affected each other to some degree.

An Empirical Study on the English Teachers’ Informatization Ability

35

Teachers’ information-based teaching ability is related to their English proficiency, which has not been mentioned in previous studies. Therefore, the results of data research enrich the existing theories is concerned. The relationship between informatization ability and proficiency is mutual promotion. For educators, continuous improvement of their information-based teaching ability can promote proficiency, and the improvement of proficiency in turn has a positive effect on information-based teaching ability. In the future research, scholars can increase the research on the relationship between informatization ability and proficiency, and put forward more effective strategies to improve teachers’ information-based teaching ability and English level, so as to improve the teaching quality. 3.2.2 The Relationship Between English Teaching Ability and Information Teaching Ability

Table 6. Spearman coefficient correlation between information teaching ability and English teaching ability Correlation Teaching ability

Spearman correlation Sig. (2-tailed)

Monitor

Research

Cognitive

.197**

.095**

.241**

0.00

0.00

0.00

Implementation .213** 0.00

Design .201** 0.00

Spearman Coefficient Correlation analysis of the relationship between the information teaching ability and English teaching ability is displayed in Table 6. The correlation suggests that there was a significant, positive but very weak relationship between teaching ability and two factors, i.e., monitor ability (r = .197, p = .000 < 0.01), research ability (r = .095, p = .000 < 0.01); there was a significant, positive but relatively weak relationship between teaching ability and three factors, namely, cognitive (r = .241, p = .000 < 0.01), implementation (r = .213, p = .000 < 0.01), design (r = .201, p = .000 < 0.001). It signifies that the subjects’ monitor ability and research ability and English teaching ability influenced each other to a small degree, and cognitive ability, implementation ability and design ability and English teaching ability influenced each other to some degree. Teachers’ information-based teaching ability is related to their English teaching ability, which has not been mentioned in previous studies. Therefore, the results of data research enrich the existing theories are related. The relationship between informatization ability and English teaching ability is mutual promotion. For educators, continuous improvement of their information-based teaching ability can promote English teaching ability, and the improvement of English teaching ability in turn has a positive effect on information-based teaching ability. In the future research, scholars can increase the

36

X. Ning and H. Zhou

research on the relationship between informatization ability and English teaching ability, and put forward more effective strategies to improve teachers’ information-based teaching ability and English teaching ability, so as to improve the teaching quality. 3.2.3 The Important Predictors of Information Teaching Ability Which Affect English Proficiency

Table 7. The important predictors of information teaching ability which affect English proficiency Model summary Model

R

Adjusted R square

1

.264a

.069

2

.303b

.091

3

.308c

.093

a. Predictors: (Constant), cognitive

b. Predictors: (Constant), cognitive

b. Predictors: (Constant), cognitive

b. Predictors: (Constant), cognitive, design

c. Predictors: (Constant), cognitive, design

c. Predictors: (Constant), cognitive, design

c. Predictors: (Constant), cognitive, design, implementation

d. Predictors: (Constant), cognitive, design, implementation

d. Predictors: (Constant), cognitive, design, implementation

Table 7 shows the result of the linear stepwise regression of the six independent variables on the subjects’ English proficiency. The first factor entered was cognitive ability, which explained 6.9% of the total variability of the dependent variable. The second factor entered was design ability, which decided 2.2% and the third factor was implementation ability, which affected 0.2%. Altogether, these three independent factors contributed 9.3% to the variance in English proficiency. Compared with the previous research theories, this study analyzes the predictive factors, and confirms that these three factors (cognitive, design, implementation) can make important predictions for teachers’ English proficiency. This discovery expands the dimensions of existing theories. For educators and education administrators, teachers’ English proficiency can be enhanced from three factors. In the future empirical research, scholars can add predictive factor analysis to find out the important predictors of teachers’ English proficiency in the information-based teaching ability. 3.2.4 The Important Predictors of Information Teaching Ability Which Affect English Teaching Ability Table 8 shows the result of the regression of the six independent variables on the selfassessment of English teaching ability. The first variable entered was cognitive ability, which explained 6% of the total variability of the dependent variable. The second variable

An Empirical Study on the English Teachers’ Informatization Ability

37

Table 8. The important predictors of information teaching ability which affect English teaching ability Model summary Model

R

Adjusted R square

1

.241a

0.06

2

.264b

0.07

a. Predictors: (Constant), cognitive b. Predictors: (Constant), cognitive, implementation c. Dependent variable: 19. English teaching ability

entered were implementation ability, explaining 1%. These two independent variables contributed 7% to the variance in English teaching ability. Compared with the previous research theories, this study analyzes the predictive factors, and confirms that these two factors (cognitive, implementation) can make important predictions for teachers’ English teaching ability. This discovery expands the dimensions of existing theories for educators and education administrators, teachers’ English teaching ability can be enhanced from the two factors. In the future empirical research, scholars can add predictive factor analysis to find out the important predictors of teachers’ English teaching ability in the information-based teaching ability.

4 Conclusion 1. The English teachers’ information-based teaching ability in minority areas of Yunnan Province is in a low level. In the application of information teaching, most teachers only know how to operate some simple basic information technology. Teachers’ information monitoring, design and innovation ability is very weak. 2. There is significant and positive but weak correlation between the subjects’ English proficiency and informatization teaching ability and teaching level. 3. Cognitive ability, design ability and implementation ability are the important predictors of information teaching ability which affect the subjects’ English proficiency. 4. Cognitive ability and implementation ability are the important predictors of information teaching ability which affect the subjects’ English teaching ability.

References Capan, S.A.: Teacher attitudes towards computer use in EFL classrooms. Front. Lang. Teach. 3, 248–254 (2012) Graham, C.R., Burgoyne, N., Cantrell, P., Smith, L., Clair St., L., Harris, R.: TPACK development in science teaching: Measuring the TPACK confidence of in-service science teachers. TechTrends 53(5), 70–79 (2009)

38

X. Ning and H. Zhou

Suárez-Rodríguez, J., Almerich, G., Orellana, N., Díaz-García, I.: A basic model of integration of ICT by teachers: competence and use. Educ. Tech Res. 66, 1165–1187 (2018). https://doi.org/ 10.1007/s11423-018-9591-0 Keengwe, J., Onchwari, G.: Computer technology integration and student learning: barriers and promise. J. Sci. Educ. Technol. 17, 560–565 (2008) Kirschner, P., Wopereis, I.G.J.H.: Mindtools for teacher communities: a European perspective. Technol. Pedagog. Educ. 12(1), 105–124 (2003) Weng, C.H., Tang, Y.: The relationship between technology leadership strategies and effectiveness of school administration: an empirical study. Comput. Educ. 76, 91–107 (2014) 崔丽. 教师信息化教学能力提升 . 中国学刊 05, 76–79 (2014) 樊文芳.中学英语教师信息化教学能力现状分析与对策研究——以甘肃省定西市为例. 兰 州, 西北师范大学 (2013) 胡培培.中小学教师教育技术能力发展的研究. 南京, 南京师范大 (2008) 教育部办公厅. 教育部办公厅关于印发 《中小学教师信息技术应用能力标准 (试行) 的通知 》 [EB\OL].(2014-10-07) [2004-05-28] 马若明. 乡村教师信息化教学能力发展的研究. 兰州, 西北师范大学 (2005) 任瑞红. 中小学教育信息化现状、问题及对策. 济南, 山东师范大学 (2009) 王卫军. 教师信息化教学能力发展研究. 兰州, 西北师范大学 (2009)

Design of Aerobics Network Teaching System Based on Artificial Intelligence Rong Sun(B) and Zhaoqi Fu School of Physical Education, Jiangxi Science and Technology Normal University, Nanchang 330008, China [email protected]

Abstract. The traditional network teaching system uses network video for aerobics teaching, which can not effectively correct students’ movements. At the same time, its recommended resources are difficult to match students’ learning ability, and can not provide targeted intensive training resources. In order to realize remote intelligent aerobics teaching and improve the teaching effect of aerobics course, an aerobics network teaching system based on artificial intelligence is designed. In the process of hardware design, the control module and peripheral function module are optimized. The software part uses the meanshif algorithm to track the students’ bone information collected by Kinect, so as to correct the aerobics movement. At the same time, the neural network is used to improve the recommendation algorithm. According to the students’ learning and action correction, aerobics teaching resources are recommended to meet the needs of students at different stages. The system performance test results show that after the application of the designed system, the action standard rate of students is increased to more than 90%, and the accuracy rate of Resource Recommendation is higher than 90%. Keywords: Artificial intelligence · Aerobics · Network teaching · Teaching system · Meanshif algorithm

1 Introduction With the increasing maturity of Internet technology, computer technology and information technology, distance education and education system based on network have made great development and progress. Moreover, at this stage, multimedia technology has fully integrated network communication technology, and teaching work is carried out more in network conditions, which has completely changed the teaching mode [1]. Integrate information technology into the process of teaching and management mode innovation, promote development with innovation, and promote the reform of education service supply mode, teaching and management mode. In the process of development and progress of human society, it has been constantly studying the use of various machines to replace part of human labor, so as to reduce labor costs, improve work efficiency and improve productivity, which makes artificial intelligence come into being. The proposal © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 39–49, 2022. https://doi.org/10.1007/978-3-031-21161-4_4

40

R. Sun and Z. Fu

of educational modernization strategy not only creates conditions for intelligent teaching, but also promotes the development of intelligent teaching. Network teaching system is an important teaching auxiliary tool to promote the scientization of school teaching methods, teaching means and teaching modes, effectively apply various high-quality network education resources and advanced education technologies, improve the school operation mechanism and management form, and integrate the existing education and teaching resources. The network teaching system mainly uses the Internet to provide rich audio and video resources such as sound, animation and film for teaching. The traditional network teaching system has the problems of large data transmission delay and poor concurrency. The physical education teaching system exists more relying on the school teaching system, and the database capacity and system framework design that can be occupied are limited, which can not meet the needs of independently implementing its own design architecture according to the basic requirements and characteristics of physical education curriculum teaching. Therefore, the design of physical education teaching system is still in a more traditional and primary stage. Although the use of Android mobile teaching system improves the use of the client, it can not meet the needs of teaching when applied to Aerobics Teaching [2]. Bodybuilding is a fashionable sport, which originates from everyone’s pursuit of health and beauty. With the improvement of China’s education quality and the deepening of the education system, aerobics is introduced into the classroom of colleges and universities to shoulder the arduous task of strengthening the national physique and cultivating high-quality sports talents. Therefore, how to improve the quality of aerobics teaching has become the key to the optimization of aerobics teaching. At present, artificial intelligence technology has been applied to many fields of social production and life. However, in the field of culture and education, although schools at all levels have generally introduced many multimedia devices for network teaching, the application of artificial intelligence technology is still in its infancy [3]. With the mature development of artificial intelligence technology, it has great application potential in the field of culture and education, which is not only reflected in the teaching of knowledge with intelligent robots, but also can greatly help the design and application of school teaching system, making the school’s educational administration and teaching system intelligent from the traditional network version, which is helpful to improve the school’s educational administration management level, It is of positive significance to improve teaching efficiency. Therefore, according to the above analysis, this paper will use artificial intelligence technology to research and design aerobics network teaching system to provide more convenient, efficient and intelligent teaching auxiliary services.

2 Hardware Design of Aerobics Network Teaching System Based on Artificial Intelligence The design of the hardware part of the network teaching system mainly serves for the following artificial intelligence algorithm and network teaching research. Therefore, the hardware part of the system should meet the basic requirements of artificial intelligence technology. Figure 1 is the overall block diagram of the hardware part of Aerobics network teaching system.

Design of Aerobics Network Teaching System JTAG

41

USB

Kinect

Power supply module STM32F103C8T6

Camera

Communication module

DDR

Flash

Upper machine

Fig. 1. Hardware block diagram of Aerobics network teaching system

Real time acquisition of system communication signal and real-time data communication with PC. Combined with the needs of network teaching, the selected main control chip should not only have stable and reliable performance, but also have strong data processing ability and rich communication interfaces, but also have high-quality ADC module. Considering the system performance and cost, STM32F103C8T6 microprocessor is selected as the main control chip of the system. Use camera and Kinect equipment to realize human-computer interaction function and improve the effect of Aerobics network teaching. 2.1 Control Module Design STM32F103C8T6 is a 32-bit ARM microprocessor, which can choose to use internal or external clock. Through frequency doubling, the maximum frequency can reach 72 MHz; The chip contains 64 KB flash and 20 KB SRAM space; On chip 12 bit 10 channel ADC, the shortest conversion time is 1US, and the external bus controller with bus request/response pin can be used for real-time signal acquisition; Up to 9 communication interfaces, including 2 I2C interfaces and 3 USART interfaces, which can be used for data communication with identification module and PC; The power supply voltage is 2 V ~3.6 V, and there are three low-power modes: sleep, stop and standby, which ensure the requirements of low-power applications; Support JTAG interface to facilitate development and debugging [4]. The microprocessor supports large and small terminal modes. The internal architecture is big end mode, and the external memory can be big and small end modes. The access cycle of chip control programming can be set to 0 ~7 waiting cycles. The DMA transmit/receive buffer of the chip transmits and receives with 256 bytes. When the system needs to be reset, the main control chip can be reset manually by pressing K1 key and automatically by power on. The following Fig. 2 is the schematic diagram of storage mapping when the main control chip is reset [5]. The quartz crystal with T1 of 32.768 kHz of microprocessor is used as a low-speed external clock, which is connected with pc14 and PC15 pins of STM32. It is generally used for RTC clock. When the power VDD is cut off, RTC can still work normally as

42

R. Sun and Z. Fu 0X3FFFFFF 0X3FF0000

Special function register

Undefined memory 0X2000000 ROM/SRAM FLASH BANK 0 (Inaccessible area)

0X0000000

ROM/SRAM FLASH BANK 0 (Accessible area)

Fig. 2. Schematic diagram of reset storage mapping

long as VBAT supplies power; T2 is an 8 MHz quartz resonator as a high-speed external clock, with the OSC of STM32_ IN.OSC_ The out pin is connected, and the frequency doubling output can be carried out through PLL phase-locked loop to provide 72 MHz working frequency for microprocessor core, or provide appropriate working frequency for on-chip low-speed equipment through frequency division. 2.2 Peripheral Function Module Design In addition to the control module, the peripheral function module of Aerobics network teaching system hardware mainly includes communication module, power supply module, data synchronization module and surrounding circuits. The communication interface of the camera uses video decoding chip ADV7183A. The chip is based on ITC – RBT 656 automatically detects pal composite video and decodes analog signals into digital video signals. The chip adopts a 4:2:2 sampling scheme, which can convert the signals of palsecam and NTSC into standard video signals compatible with itu 601 standard. It describes video digital signals using brightness and two color difference signals [6]. The video interface unifies the input data format and sends it to the image processing unit. When the data is on the 8-bit bus and the LLC1 clock signal arrives, it will remind the synchronizer that the data is valid. The output of this component is synchronous data and 6-bit signal bus. The signal bus contains a 4-bit status bus and a 2-bit content bus. The status bus is used to provide the position information of relevant pixels in the current frame. The 2-bit content bus indicates the Y, Cr or CB values of the current data. The input interface will process all pixel data from the video decoder, which will need to divide the effective video information into luminance component and chrominance component. To do this, the interface will need to start the valid video information (SAV) and end the valid video information (EAV) code to save the input track. These codes are 4-byte pixel sequences that separate valid and invalid pixels from each other. If the pixel data is not required, it can be discarded. The luminance component may be discarded because the existing algorithms only use two chroma [7]. After thresholding CB and Cr,

Design of Aerobics Network Teaching System

43

the input interface will respond to the request put forward by the control system and data storage in RAM. The video interface can be divided into two parts: input synchronization and state detection. CP2102 chip is selected to bridge USB/RS-232. After installing the driver of CP2102 chip and initializing according to the data format defined by USB interface, on the one hand, the USB data transmitted by the upper computer can be converted into the format of RS232 information flow and sent to the main control module, on the other hand, the serial port data can be converted into USB data and transmitted to the upper computer. Input synchronization is performed through a pair of triggers. This pair of flip flops reduces the possibility of metastable state entering the system by allocating a whole clock cycle between the two flip flops. A register loaded with asynchronous clock can synchronously output enough time signals for new valid data. The system clock here is set to 60 MHz. A fourth-order shift register stores the last four input data for the identification of timing reference marks. The LLC1 synchronization clock signal triggers the loading of the shift register. The contents of the shift register and the LLC1 synchronization clock signal provide input to the state generation component (ADC). Table 1 below shows the format of valid content input signal when the counter tracks data. Table 1. Format setting of valid contents of counter Data input

Most significant bit

Least significant bit

0

0

Y

1

1

Cb

0

1

Cr

1

0

—

When data exchange communication is carried out, it is output to the next stage except for status and content signals. The signal bus is only suitable for one clock cycle and is synchronized with the 8-bit output data bus. The signal bus format is always retained in the whole pipeline. Under the above hardware conditions, the software part of Aerobics network teaching system is designed by using artificial intelligence technology to realize the remote network teaching of aerobics course.

3 Design of Aerobics Network Teaching System Software Based on Artificial Intelligence 3.1 Aerobics Movement Correction Design In Aerobics network teaching, the system designed in this paper requires students to use Kinect sensor to realize human-computer interaction, so as to complete the correction and guidance of aerobics. Therefore, the algorithm is used to track Kinect bones, so

44

R. Sun and Z. Fu

as to establish learners’ human motion trajectory, and then carry out aerobics action correction. In the depth image collected by Kinect, the upper 13 bits of the color information of a single pixel are the depth information, and the lower 3 bits are the user index (ID). 0 ~2 bits store the user index value, and 3 ~15 bits store the pixel depth value. Move the pixel data right by 3 bits to obtain the pixel depth value (realdepth = (bufferrun [j]&0xfff8) > > 3). Table 3.1 shows the depth bits and index bits of the depth of field data, There are 13 depth bits (D3~D15) and 3 index bits (D0~D2). The specific depth value and index value can be calculated from the data bits. The maximum depth value of Kinect is about 4000 mm, so that before image preprocessing, a threshold value is set to be 1200 mm ~3500 mm, so as to limit the depth data to 1200 mm~3810 mm. Meanshif algorithm is used to track the learners’ Aerobics actions collected by Kinect sensor. When the meanshif algorithm tracks, it calculates the offset mean value of the current point, then moves the point to the position of the offset mean value of the point, then calculates the offset mean value of the offset point position, and then moves. The iteration will not stop until the end conditions of the iterative algorithm are met. The kernel function is used for nonparametric estimation of meanshif algorithm. There are usually two kinds of kernel functions, unit uniform kernel function and unit Gaussian kernel function. Gaussian kernel function is used in this design. The density of a point can be calculated according to the kernel function. Many data processed in the field of computer vision are high-dimensional data, and the kernel function used in high-dimensional space is multivariable kernel function. In the d -dimensional Euclidean space, given n points xi ∈ Rd and a positive definite d × d bandwidth matrix H , the multivariable kernel density estimation expression is as follows:   n   −1  x − xi (1) fH ,k (x) = n |H | K √ |H | i=1 There are two kinds of bandwidth matrices, namely diagonal matrix and identity matrix. The formula of identity matrix is as follows:   n    −1  x − xi fH ,k xˆ = n |h| K h

(2)

i=1

where K is the kernel function; h is the of the kernel function. When K is  bandwidth centrosymmetric and has K(x) = Qk x2 , where x is a non negative number, K can be called a contour function, and Qk is a standardized constant coefficient. The formula in the above formula can be changed to the following formula:

 

n

x − xi

  d −1 



fH ,k xˆ = nh Qk K

h

(3)

i=1

The eigenvalue probability density of the above formula can be calculated by the above formula. When estimating the standard density gradient, r(x) = −k  (x) can be

Design of Aerobics Network Teaching System

45

set and brought into the above formula to simplify the following mean shift vector: n 

pk (x) =

i=1



n 



  i 2

x−xi 2 − x xi r x−x r h h i=1

n 

2   r x−xi

i=1

(4)

h

The mean shift vector is a direction in which the gradient value of probability density increases, and the mean shift algorithm is an iterative algorithm. The iteration window size of meanshift iterative algorithm is fixed. The specific mean shift iterative algorithm is as follows: (1) Determine the size, position and shape of the window to be iterated in advance. (2) Calculate the center of gravity in the window according to the above iterative algorithm, and determine whether it needs to be weighted according to needs. (3) Move the center of the original window to the center of gravity of iterative convergence. (4) Continue to calculate the center of gravity at the new center position. That is, repeat the process of (2) until the end condition of convergence iteration is reached, stop the iteration, and this position is the end position of convergence to obtain the final tracking path. Using the bone motion model demonstrated by the teacher as the template, the Kinect bone tracking path of the students is matched, so as to determine whether the students’ actions are correct and correct the aerobics actions. 3.2 Recommendation of Aerobics Resources by Neural Network The network teaching system needs to recommend teaching resources with moderate difficulty based on students’ learning level and students’ feedback in aerobics movement learning and correction. After receiving students’ feedback, the teaching system transmits the data to the resource recommendation module, and the recommendation module retrieves the optimal resources and recommends them to students’ users based on the feedback results. This paper combines the widely used CF algorithm with deep neural network, and proposes a collaborative filtering recommendation algorithm based on deep neural network fusion. The algorithm is used to recommend aerobics personalized teaching resources in the teaching system. The algorithm consists of two main parts. The first part uses the auxiliary information of users and projects, such as resource name, resource description and other specific information, to obtain the initial feature matrix of user projects through deep neural network, in which text attribute features are vectorized by LSTM neural network, and other attribute features are vectorized and input into neural network, Finally, the fusion is carried out through the full connection mode. The second part carries out prediction processing through DBN deep confidence neural network. The initial input is the initial characteristic matrix obtained in the first step, and the output

46

R. Sun and Z. Fu

layer will obtain the user’s specific prediction value of the project. Through comparison with the real score, cycle learning until the optimal prediction value is obtained. In this paper, the deep confidence network DBN is composed of three layers of RBM in series and the last layer of BP neural network. The RBM model is based on state energy, and the current state energy function is shown in the calculation formula. I J I  I  (ui − gi )2  tij uij lij − i=1 s i lj (5) − E(z, v; ξ ) = − 2 i=1 j=1

i=1

In the above formula, the visual layer data is generally represented by u; Hidden layer data is represented by l; T represents the weight between the visible layer u and the hidden layer l. gi and si represent the offset of visual node i and hidden node j respectively, tij represents the connection weight between the i-th visual node and the j-th hidden node, and ξ is specified as the parameter set. According to the log likelihood probability gradient, the weight update criterion of RBM can be obtained, and the Gibbs sampling method is used to obtain the parameters of DBN depth confidence network. After each iteration completes the pre training process of DBN, the BP feedback neural network obtains the training results, feeds them back from top to bottom, adjusts and optimizes the overall training, uses the minimum mean square error MSE function to detect the error between the predicted score and the real score, and predicts the score according to the highest probability value in the probability matrix, So as to recommend the aerobics resources with the highest score to the corresponding learners, so as to help learners get more targeted learning resources or targeted intensive training. From the above design content, the software function of Aerobics network teaching system is realized by using artificial intelligence technology, and the software part is transplanted to the system hardware environment, that is, the design of Aerobics network teaching system based on artificial intelligence is completed.

4 Test Experiment The aerobics network teaching system based on artificial intelligence is designed above. This section will test the practical feasibility of the system. 4.1 Experimental Contents Testing the software can effectively ensure the stability of system performance and the perfection of function, mainly including function testing, performance testing and so on. For this system, we first developed the student login and registration function, then further developed the login and registration function of other roles in the system, and then tested the login and registration function. The function of each system component is improved through testing, and the whole system is gradually realized. After testing the function of the system in this paper, it is completed in the form of comparison in the system performance test. Using the traditional network teaching system as a comparison, the system performance is evaluated by comparing the response rate of the system and the standard rate of students’ aerobics movement after the application of the system.

Design of Aerobics Network Teaching System

47

4.2 Experimental Results The main task of function testing is to detect whether each function in the system has been completed, including using the method of unit test to test each functional module in the system, ensure that each functional module can be used normally, and feed back the predicted output. The following table is some examples of testing the system to test whether it meets the functions required by the requirements specification. For example, the test is shown in Table 2. Table 2. User management test cases Test module

Test content

Expected results

Test result

Application loading

Start Chrome browser; Enter the URL address of the system

The browser displays the system login page

Meet expectations

User login

User name and password are empty

Unable to log on to the Prompt user input system cannot be blank, please log in again

Incorrect user name or password

Unable to log on to the Prompt the user for system input error, please log in again

After successful login, check the legitimacy of the user entering the page

Users can enter different pages according to different roles

Meet expectations

It can be seen from Table 2 above that the system functions normally and can operate normally. Response time is the time that the system reacts to the user’s input or request. The faster the response time, the faster the response speed of the corresponding system and the better the performance. The movement standard rate refers to the matching degree between students’ Aerobics movements and standard movements. The higher the matching degree, the better the teaching performance of the corresponding system. Table 3 shows the response time comparison between the designed system and the mobile teaching system in reference [2] under different conditions and the standard rate of Aerobics movements of students. It can be seen from Table 3 that the response time of the system in this paper is shorter than that of the Mobile teaching system, and the response speed meets the actual needs of Aerobics Teaching at present. From the students’ movement standard rate, the students’ movement standard rate using this system is higher than 90%, which is higher than the highest value of 88.6% of the Mobile teaching system, indicating that the effect of aerobics teaching using this system is better. The main reason is that the control module and peripheral function module are optimized in the design system, which improves the control performance and response speed of the system. At the same time, meanshif

48

R. Sun and Z. Fu Table 3. Function comparison test results

Number of system test users

Paper system Response time/ms

Mobile teaching system Action standard rate/%

Response time/ms

Action standard rate/%

100

83.9

93.6

244.1

79.1

150

98.8

94.5

325.3

86.3

200

112.9

94.4

389.2

79.4

300

121.7

95.3

4278

80.7

400

128.5

94.2

515.4

88.6

500

130.6

91.1

686.6

84.5

algorithm is used to track the students’ bone information collected by Kinect, so as to achieve the purpose of correcting Aerobics movements and improve the standard rate of students’ movements. Figure 3 below shows the accuracy comparison between students’ needs and system recommended resources when students use the system. In the comparison process, students of a university are divided into 5 groups with 20 people in each group to make targeted recommendations.

Fig. 3. Comparison of accuracy of system recommended resources

It can be seen from Fig. 3 that when students in each group use the two teaching systems, the accuracy of the recommended resources recommended by the system in this paper is relatively higher, higher than 90%. In this paper, the improved neural network recommendation algorithm is used to recommend the teaching content, which can accurately eliminate the influence of other factors, and the recommended resources can better meet the needs of students to learn aerobics.

Design of Aerobics Network Teaching System

49

Above all, the aerobics network teaching system based on artificial intelligence designed in this paper has good operation and practical application effect.

5 Conclusion Aerobics is a popular sport. As one of the important contents of physical education curriculum, it is also welcomed by students. Influenced by the obvious fitness effect and easy to learn characteristics of aerobics, many scholars at home and abroad have designed and developed the aerobics network platform, but most of the people whose platform design is aimed at are fitness enthusiasts of non professionals in society, which is not suitable for campus aerobics course teaching. By deepening the application, we can give play to the role of information technology in education and teaching reform and development, Relying on educational informatization, accelerate the construction of learner centered teaching and learning methods. Therefore, this paper designs an aerobics network teaching system based on artificial intelligence from two aspects: hardware control module and peripheral function module optimization, software part calibration of aerobics movement and recommendation of aerobics teaching resources. The performance of the system is tested to verify the feasibility of the system being put into teaching. In the actual aerobics teaching, the use of the system effectively enhances students’ interest in learning, improves the effect of Aerobics network teaching, and improves the teaching disadvantages of traditional aerobics teaching. In the next step, the specific aerobics teaching courses will be classified and taught to further improve the teaching quality.

References 1. Zhang, Y.: Mobile live broadcasting platform for aerobics video teaching based on Android. Inf. Technol. 44(05), 41–44+53 (2020) 2. Li, X., Wu, Q.: Research on optimization design of modern apprenticeship teaching system applied to multimedia network video technology. J. Chongqing Inst. Technol. 33(03), 174–179 (2019) 3. Xu, Y., Jin, G.: Artificial intelligence matching simulation of targets in multi-feature cascaded image database. Comput. Simul. 38(03), 437–441 (2021) 4. Zhu, M., Shen, J., Wang, J.: The design and implementation of a scalable Internet of things teaching development system. J. East China Normal Univ. (Nat. Sci.) 2021(03), 78–95 (2021) 5. Zhang, Z., Chen, J.: Construction practice of cloud experiment teaching system for information security specialty based on cloud platform. Exp. Technol. Manage. 38(09), 251–255 (2021) 6. Zeng, H., Wang, Z., Qiu, C.: A study on building a model of artificial intelligence teaching application based on meta-module method. China Educ. Technol. (11):71–76+87 (2021) 7. Zhang, J., Li, C., Zhu, Y.: Massive video teaching resource management system based on cloud platform. Mod. Electron. Tech. 43(21), 151–155 (2020)

Design of Online Open Wushu Sanda Teaching Platform Based on Hybrid Reality Technology Zhaoqi Fu(B) , Dunke You, Wen Liu, and Rong Sun School of Physical Education, Jiangxi Science and Technology Normal University, Nanchang 330038, China [email protected]

Abstract. IN view of the lack of interaction in Wushu Sanda Teaching, an online open Wushu Sanda Teaching Platform Based on hybrid reality technology is designed. In the hardware part, the serializer packs the image data into FPD link signal and outputs it to the deserializer circuit board through coaxial cable. The deserializer is responsible for unpacking the received FPD link signal, restoring it to mipi-csi signal and transmitting it to the image receiving and processing chip. In the software part, the logic layer includes two components: knowledge agent and interaction agent, which constitutes the architecture of learning and management system for online open platform. The collaborative interaction model is designed based on hybrid reality technology to realize the interaction behavior between users and the environment. According to the overall requirements of the platform, each functional module is designed to ensure teaching interaction. The test results show that when the number of concurrent users is 5000, the response time of the online open Wushu Sanda Teaching Platform Based on hybrid reality technology is 0.9 s, which is 0.5 s and 0.49 s shorter than that based on virtual technology and big data analysis technology, respectively. Therefore, the platform realizes each functional module and has good performance. Keywords: Hybrid reality technology · Online opening · Martial arts · Teaching platform · Platform design · Online education

1 Introduction At present, the world is in an overall peaceful environment, so Wushu has gradually shown its characteristics of fitness, entertainment and leisure. Among China’s traditional national sports, Wushu Sanda occupies a very important part. It is a highly antagonistic form of sports that comprehensively reflects skills, intelligence and courage. Sanda is also a very important branch of Wushu, so its development process will naturally be related to fitness and entertainment. Wushu Sanda as a competitive sports fitness, its popular development space needs to be further developed. In the teaching and teaching of physical education curriculum in Colleges and universities, it is clearly pointed out that physical education curriculum in Colleges and universities should carry forward national traditional sports in China, which reflects the connotation of national traditional sports. © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 50–61, 2022. https://doi.org/10.1007/978-3-031-21161-4_5

Design of Online Open Wushu Sanda Teaching Platform

51

With the rapid growth of Wushu Sanda Teaching demand, the actual teaching classroom has been difficult to meet the Wushu Sanda Teaching demand. It is necessary to use the teaching platform to combine computer technology and Wushu Sanda knowledge, and carry out teaching interaction centered on the online open teaching platform, so as to make up for the shortcomings of traditional teaching methods in Colleges and universities. Online open teaching platform is a new online course mode. It enhances the dissemination of knowledge, shares course resources and strengthens the spirit of teamwork. The research and development of online open teaching platform is fast, and some research results have been achieved. She Chunhua and Deng yuxu designed an online teaching simulation platform based on virtual technology to find the causes of different defects [1]. Shi Wanli and Zhang Yuhui use HBase database combined with SQL computing execution engine to analyze education data and transfer the analyzed student, teacher and resource information to the intelligent education information cloud service layer [2]. The education information cloud service layer enables platform users to enjoy platform storage files, course management, course publishing and other services through user authentication, service binding and service provision. The platform can provide personalized teaching and management according to students’ personalized learning behavior. The above teaching platforms can realize open education, support a large number of students’ online learning, and provide effective management for multiple courses. However, there are still problems of insufficient interactivity that need to be further solved. Integrating the cutting-edge technology of the Internet and the concept of education to create a socialized large-scale online open Internet learning model has a far-reaching impact on promoting online education. Wushu Sanda, as a traditional Wushu project, its teaching method is also more traditional and single. Carrying out physical education network teaching is also an inevitable choice in the information age. Hybrid reality technology is not only the result of the integration of the physical world and the digital world, but also the result of the interaction between human, computer and environment. It releases the possibilities limited by our imagination. It appears with the progress of graphics processing ability, computer vision, input system and display technology. Hybrid reality combines the physical world and the digital world, which define the two extremes of the scope called “virtual continuum”. On one side is the physical reality, that is, the human environment; On one side is the corresponding digital reality, while the mixed reality is in the middle of both sides. Hybrid reality allows users to interact with the virtual information in front of them. It is the interaction of hybrid reality between computers, humans and the environment. Therefore, this paper designs an online open Wushu Sanda Teaching Platform Based on hybrid reality technology, so as to promote the wide application of multimedia teaching means in Wushu teaching, optimize Wushu classes and improve the quality of Wushu teaching.

2 Hardware Design of Online Open Wushu Sanda Teaching Platform The function of the hardware part of the online open Wushu Sanda Teaching Platform is to collect, transmit and enhance high dynamic images in real time. The circuit mainly includes image acquisition circuit, image serial circuit and image deserialization circuit. The overall design architecture is shown in Fig. 1.

52

Z. Fu et al.

Sensor circuit board MIPI-CSI Serializer circuit board FPD-Link Deserializer circuit board MIPI-CSI FPGA

Fig. 1. Hardware design architecture

The image sensor is designed on a circuit board and is responsible for collecting high dynamic images. After collecting the image data, the image sensor outputs the image data to the serializer circuit board through mipi-csi interface. The serializer packs the image data into an FPD link signal and outputs it to the deserializer circuit board through a coaxial cable. The deserializer is responsible for unpacking the received FPD link signal, restoring it to mipi-csi signal and transmitting it to the image receiving and processing chip. The deserializer circuit board is connected with the image receiving and processing circuit board through FPC socket and GPIO socket on the image processing chip. The power supply scheme of the system circuit is described in detail below. Since the deserializer chip circuit board is directly connected with the image processing chip through the GPIO socket, it can directly input the vcc5v power supply. Since the power supply required by the deserializer is 1.8V, it is necessary to use the linear regulated power supply chip to generate the 1.8V power supply required by the deserializer chip. In this paper, mp1495sgj-p chip is selected to generate the 1.8V power supply required by the deserializer. Since the image sensor circuit board is connected with the serializer circuit board through FPC socket, it is necessary to generate 1.2V and 2.8V power supply required by the image sensor on the serializer circuit board. Rt9011-cmgj6 is a high-speed linear voltage regulator, which can support 2.5V to 5.5V input and generate 1.2V and 2.8V supply voltage. It is output at Vout1 and vout2 pins respectively. The maximum supply current can reach 700 mA, meeting the requirements of supply current. The principle of power supply circuit is shown in Fig. 2.

Design of Online Open Wushu Sanda Teaching Platform

53

V

V

V

EN1

VOUT2

RT9011-CMGJ6 VIN

VOUT1 C

GND

EN2

GND GND

C

GND GND

Fig. 2. Schematic diagram of power supply circuit

The drive control module selects the EK-TM4C1294XL development board of TI company to generate the controlled square wave signal. The relevant characteristics of the development board are as follows: the 32-bit processor based on ARM Cortex-M4 core can work at a frequency of 120 MHz and produce up to 8 PWM waves. It contains 8 timer modules, each module has 2 16 bit registers, 8 universal asynchronous serial communication modules (UART) and 20 12 bit ADC converters. Tm4c1294 development board receives the duty cycle information of six serializers from the upper computer through the serial port, and generates the corresponding square wave signal through the received information, so as to realize the control of the serializer. The internal vs terminal of the chip is connected to the positive pole of the DC power supply, and the GND terminal is connected to the negative pole of the DC power supply.

3 Software Design of Online Open Wushu Sanda Teaching Platform 3.1 Structure Design of Online Open Platform The design goal of Wushu Sanda Teaching Platform is to closely combine Internet technology with Wushu Sanda Teaching, and make up for the defects of traditional Wushu Sanda Teaching mode, such as lack of high-quality teachers and resources, lack of manpower, and lack of guarantee of teaching quality and process. Taking Wushu Sanda Teaching as the center, we should create an online education platform advanced by the joint efforts to build and share, effective interaction between teachers and students, mass teaching resources of high quality Wushu Sanda, and have advanced, sustainable, iterative and upgrading development to meet the needs of Wushu Sanda Teaching in our school, thus improving the management and teaching level of Wushu Sanda, and improving students’ initiative learning ability and innovative learning ability. Meet the society’s increasing demand for Wushu Sanda Teaching. The online open platform structure of this design is shown in Fig. 3.

54

Z. Fu et al.

Interface layer

Send request

Return data

Freemarket implementation

Configuration file

Logical layer ADO object SQL commit

Dataset return

Data layer

Database

Fig. 3. Online open platform structure

Any application software system needs a friendly operation interface. The interface design adheres to the user experience as the center, with intuitive and concise interface and convenient and fast operation [3]. Viewed from the current mature software UI, the general standard of UI is that the software has a main interface, and all business-related tasks should be completed in the main interface, while other main functions of data input and data output are completed through the navigation menu. For our work in UI, we have adopted easyUI for the teacher working platform. Using easyUI can quickly create an interface effect with unified style. The easyUI main interface layout is divided into four parts. The top is the page log, the left is the function menu navigation bar, and the right shows the main content. The top part lists the task list of the current page, which is simple and friendly for the user experience; Powerful, the page can support the display of various browsers, and easyUI provides many controls. The logic layer includes two components: knowledge agent and interactive agent, which constitutes the architecture of learning and management system for online open platform. The main function of interactive agent element is to create a unified platform for learners to participate in the theme discussion and curriculum activities of Wushu Sanda Course, and promote learners to join the theoretical learning of the course with the help of interactive agent element [4]. In this framework, learners transfer behavior to evaluation elements and submit learning parameters to teaching agents. The evaluation component is mainly to effectively evaluate the learning parameters provided by learners according to the historical information of learners stored in the learner record data. According to the historical information and evaluation information, the teaching agent finds the resources needed by learners in the learning resource database. With the help of component delivery, it uses the multimedia platform to provide learning resources for Learners [5]. MySQL has attracted much attention because of its speed, reliability and adaptability. Through the demand analysis, because the MySQL database has the characteristics of low cost, small volume, open source and free, superior performance and powerful function, it can meet the needs of the system, so the MySQL database is selected in the system. The

Design of Online Open Wushu Sanda Teaching Platform

55

primary keys of all tables are uniformly named ID, and the specified type is long, except for the tables formed by many to many relationships. 3.2 Design Collaborative Interaction Model Based on Hybrid Reality Technology According to the interactive characteristics and requirements of online open Wushu Sanda Teaching Platform, a collaborative interaction model is designed based on hybrid reality technology. The basic elements of interaction model include user, mode, task, operation and representation. The specific attributes of user elements can include user ID, user role, group size, user status, etc. Modal elements conform to the multi sensory interaction mode of human in normal cooperation. Specific collaboration task requirements will affect the collaboration mode between users. The proposed eight collaboration styles include active discussion, visual participation, shared perspective, independent perspective sharing, synchronous completion of specific tasks, asynchronous resolution of general tasks, resolution of different tasks and separation from collaboration [6]. The operation of hybrid reality technology represents the operation that users need to complete in order to achieve their goals, and can support the interaction between a single user and the environment and the interaction between users [7]. Teaching interaction is a process of adjusting and controlling teaching activities to meet the needs of students in special network teaching. This process is based on the analysis of the adjustment and control of teaching objectives, contents, modes, objects, environment and other elements, and optimizes the teaching process through teaching methods, means and modes. In Wushu Sanda Teaching, one operation corresponds to one kind of element action. For an element action, in order to judge the position, time span and direction of its symbol in Wushu Sanda, it is necessary to determine the body part, start and end time and direction corresponding to the action. The body part corresponding to the action can be determined by the joint point performing the action; The start and end time can be determined by the number of frames and frame rate of motion capture data; The judgment of movement direction needs reference. In Wushu Sanda Teaching, the front of the body is usually used as a reference to quantify other directions. Taking the element action of human arm as an example, when judging the direction, first find the final posture of the action, and then subtract the hand node coordinates and shoulder node coordinates to calculate the arm vector. The direction points from the shoulder to the hand, quantifying the horizontal and vertical directions respectively. When quantifying the vertical direction of arm movement, it is necessary to calculate the angle between the arm vector and the positive direction of y axis. The included angle is calculated as follows: α = arccos

p1 |p2 |

(1)

In formula (1), α represents the included angle; p1 is the vertical component; p2 represents the arm vector. Assuming that the included angle between the front of the human body and the x-axis of the camera world coordinate system is β, and the position of a joint point in the world coordinate system is (x0 , y0 , z0 ), the coordinates of the point

56

Z. Fu et al.

after shooting angle normalization are calculated as follows: ⎡ ⎤ cos β 0 sin β (x, y, z)T = ⎣ 0 1 0 ⎦(x0 , y0 , z0 )T − sin β 0 cos β

(2)

In formula (2), (x, y, z) represents the coordinates after normalization; T represents transpose matrix. The actions of each upper limb element are regarded as the category label of Wushu Sanda data, and the judgment of the upper limb action direction is regarded as a classification problem, which is processed by the integrated model based on sub sample aggregation strategy and limit learning machine. During training, input the timing data matrix of the action and the corresponding category label. The activation function is:  (3) f = 1 + (cμ + η) In formula (3), f represents the activation function; c represents training sample; μ represents the weight vector between the input layer and the hidden layer node; η indicates offset. After training, for an unknown action data, the limit learning machine can identify the category label of the action. The discrimination formula is as follows: g(c) = argi max ϑi (c)

(4)

In formula (4), g represents the category label; i indicates classification; ϑ represents the probability that action c belongs to class i. In the collaborative interaction model based on hybrid reality technology, the element of representation can be regarded as the output module in collaboration. Full and effective real-time information display is very important for each user to help users improve their perception of the collaborative system and get effective interactive feedback. 3.3 Design the Functional Module of Wushu Sanda Teaching Platform Platform login mainly includes the login of registered users to the martial arts Sanda Teaching Platform. User login requires user name and password. The system will assign different users access to the system according to their roles and permissions. At the same time, it supports third-party authorized login. On the platform home page, we adopt the bootstrap page framework and add some dynamic effects. For example, the left to right button below can dynamically switch the recommended courses, showing a good interface effect. At the system management end, you can edit the system main page, specify the cover course, specify the classification, etc. Teaching resources are an important part of teaching. Online teaching resources are one of the components of network teaching resources. Rich and colorful teaching resources are an important guarantee for teaching quality. Teaching resources are the teaching materials provided by teachers for the development of teaching, and online teaching resources are one of the teaching materials for teachers to carry out network teaching. Online teaching resources of network teaching mainly include teaching documents, auxiliary resources and network resources. The content of online teaching resources is rich. There are not

Design of Online Open Wushu Sanda Teaching Platform

57

only electronic teaching materials showing teaching contents, but also teachers’ teaching plans, courseware, teaching demonstration, student work display, etc. the teaching forms are also diverse, including documents, pictures, audio and video, animation, etc. This paper will use Professor Wang Sen’s Sanda defense and counterattack classic, this paper will refer to Professor Wang Sen’s Sanda defense and counterattack classic, Zeng Yujiu’s new theory of Wushu Sanda training, Zhou Zhengwei’s Sanda Teaching and training, and the special teaching materials for Sanda Course of national traditional sports major in Physical Education College of Shandong Normal University. The course category list on the left is pre specified in the database. The first level course classification and some second level course classifications are pre specified. The first level category is specified according to the national first level subject directory and cannot be changed. The second level classification teachers can add dynamically. When the system first enters the page, all courses will be loaded, and more unclassified courses can be loaded through the “more” button. If you click the link of secondary category on the right, the courses of the specified category will be loaded. If the student logs in, the student can enter learning through the start learning now button on the right. When adding Wushu Sanda Course Resources, you need to select the corresponding type under the unit according to the type of resources. If you add video, you need to select the video type, if you add exam, you need to select the exam type, if you add homework, you need to select the assignment type, if you add discussion, and if you add HTML, you need to select the HTML type. Before adding an assignment or exam, you need to enter the test questions on the platform to form an assignment or test paper to form a corresponding assignment list and test paper list. When you need to add an assignment or exam, you can select from the corresponding list. The user management function mainly includes permission setting, role definition and account management. It establishes the background data management organization system for the martial arts Sanda learning platform to ensure the normal functions of the system. The main function of system management is to set the authority of each user for data query and analysis, so as to ensure the security of the system. The system management module subsystem mainly includes log information, system information, user information, report generation and information release.

4 Teaching Platform Test 4.1 Platform Function Test Teaching platform testing is mainly to ensure the high-quality operation of the platform and reduce defects and errors. It exists in each life stage of platform development. Each stage of testing is the process of finding errors through testing, finding and solving problems as soon as possible, so as to ensure the smooth development and implementation of software. The hardware environment for platform performance test is as follows: CPU: Intel (R) core (TM) i5-3230M CPU @ 2.60 GHz; Memory capacity: 8g; Graphics card model: AMD Radeon R5 m200; Video memory capacity: 2 G. The software environment is as follows: operating system: Microsoft Windows 10 professional edition; Web server: CPU: 2 pcs, above 2.4 GHz; Application server: CPU: 2 pcs, above 2.4 GHz; Database server: memory: above 8 GB; Hard disk: above 500 g. Functional test cases are mainly divided into basic operations such as course editing, management and search, user login,

58

Z. Fu et al.

addition and deletion, announcement release and deletion of course resources, etc. After testing, the functions of the online open Wushu Sanda Teaching Platform can be realized normally, and the test results are consistent with expectations. 4.2 Platform Performance Test Performance test is mainly to test the performance indicators of the platform, and judge whether the indicators of the platform are consistent with the expected performance indicators, whether they are safe and stable enough, etc. For the martial arts Sanda Teaching Platform, due to the large number of users, it mainly tests the pressure of the platform. Set the number of concurrent users to 2000, 5000 and 10000, continuously increase the number of concurrent users on the platform, record the response time of the platform, conduct 10 experiments in each group, and take the average value in 10 groups. The stress test results of the online open Wushu Sanda Teaching Platform Based on hybrid reality technology are compared with the teaching platform based on virtual technology and big data analysis technology. The platform stress test results of different concurrent users are shown in Tables 1, 2 and 3. Table 1. Response time of 2000 concurrent users Number of experiments

Platform response time (s) Online open Wushu Sanda teaching platform based on hybrid reality technology

Online open Wushu Sanda teaching platform based on virtual technology

Online open Wushu Sanda teaching platform based on big data analysis technology

1

0.44

0.68

0.71

2

0.42

0.64

0.74

3

0.55

0.68

0.77

4

0.48

0.69

0.68

5

0.56

0.66

0.65

6

0.55

0.68

0.66

7

0.44

0.65

0.72

8

0.41

0.67

0.75

9

0.43

0.62

0.71

10

0.42

0.63

0.74

According to the above results, the response time of each platform increases with the increase of the number of concurrent users. However, for the user, this data is still within the acceptable range, and the user will not give up the operation. Taking 5000 concurrent users as an example, the response time of the online open Wushu Sanda Teaching Platform Based on hybrid reality technology is 0.9 s, which is 0.5 s and 0.49 s shorter than the

Design of Online Open Wushu Sanda Teaching Platform

59

Table 2. Response time of 5000 concurrent users Number of experiments

Platform response time (s) online open Wushu Sanda teaching platform based on hybrid reality technology

Online open Wushu Sanda teaching platform based on virtual technology

Online open Wushu Sanda teaching platform based on big data analysis technology

1

0.85

1.33

1.36

2

0.84

1.34

1.34

3

0.98

1.48

1.37

4

0.85

1.36

1.38

5

0.96

1.45

1.35

6

0.99

1.39

1.36

7

0.85

1.36

1.43

8

0.82

1.38

1.42

9

0.91

1.45

1.45

10

0.95

1.49

1.46

Table 3. Response time of 10000 concurrent users Number of experiments

Platform response time (s) Online open Wushu Sanda teaching platform based on hybrid reality technology

Online open Wushu Sanda teaching platform based on virtual technology

Online open Wushu Sanda teaching platform based on big data analysis technology

1

2.06

3.12

2.98

2

2.14

3.24

3.12

3

2.08

3.18

3.08

4

2.26

3.256

2.98

5

2.15

3.3

2.98

6

2.22

3.29

2.99

7

2.03

3.15

3.01

8

2.08

3.28

3.13

9

2.14

3.25

3.15

10

2.17

3.24

3.14

60

Z. Fu et al.

Resource processing speed /ms

teaching platform based on virtual technology and big data analysis technology respectively. From this result, the performance of the function during operation is consistent with the goal. Based on the above analysis, the design platform has fast response time and good response performance. The main reason is that the design platform constitutes an online open platform oriented learning and management system architecture, and uses hybrid reality technology to build a collaborative interaction model to realize the interaction between users and the environment, and improve the response performance. In order to further prove the superiority of this platform, 1000 gb Wushu Sanda teaching video resources are set to be integrated. There are 30 practical tasks in Wushu Sanda Teaching. The processing efficiency of Wushu Sanda teaching video resources is shown in Fig. 4. 100 Before integration

90 80

After integration

70 60 50 40 30 20 10 05

10

15

20

25

30

Wushu Sanda Teaching Tasks / piece

Fig. 4. Processing efficiency of Wushu Sanda teaching video resources

By analyzing Fig. 4, it can be seen that there is a significant difference in the processing efficiency of Wushu Sanda Teaching Video Resources before and after the integration, and the processing speed after the integration is lower than 70 ms. This shows that this platform can access Wushu Sanda teaching video resources, and the retrieval speed is fast, and can quickly provide students with the required Wushu Sanda teaching video resources. The main reason is that the image data is packaged into FPD link signals by the serializer and output to the deserializer circuit board through the coaxial cable. The deserializer is responsible for unpacking the received FPD link signals, restoring them to mipi-csi signals and transmitting them to the image receiving and processing chip, thus improving the call performance.

5 Conclusion This paper designs an online open Wushu Sanda Teaching Platform Based on hybrid reality technology, which greatly improves the application value of teaching. Through the platform of Wushu Sanda Teaching, we can more timely update the information of Modern Wushu Sanda learning, and make Wushu Sanda Teaching reach a new level. However, there are still some improvements and deficiencies in the platform. The overall design of the platform needs further aesthetic treatment, the compatibility is not enough, and it should further meet the requirements of supporting mobile devices. The function

Design of Online Open Wushu Sanda Teaching Platform

61

and content of the platform need to be further optimized and improved. With the increase of teachers and students of Wushu Sanda Course, there is a greater demand for platform data transmission. There may be some other problems, and the platform needs to be gradually improved in the future.

References 1. She, C., Deng, Y.: Construction of experimental online teaching simulation platform based on virtual technology. Mod. Electron. Tech. 43(21), 156–159, 164 (2020) 2. Shi, W., Zhang, Y.: Intelligent education platform design based on big data analysis technology. Mod. Electron. Tech. 43(9), 150–153 (2020) 3. Zhang, J., Wang, H., Ban, J.: An optimal design of vocal music teaching platform based on virtual reality system. Comput. Simul. 38(6), 160–164 (2021) 4. Shen, C., Bai, H.: Current situation and optimization strategies for construction of learning platform for artificial intelligence courses in primary and secondary schools. E-educ. Res. 42(10), 77–83 (2021) 5. Wang, Z., Chen, R., Ren, C.: Cloud platform anomaly detection based on ensemble learning. Comput. Eng. Des. 41(5), 1288–1294 (2020) 6. Zhang, X., Zhang, Y., Wang, Y., et al.: Auxiliary maintenance method for electromechanical equipment integrating digital twin and mixed reality technology. Comput. Integr. Manuf. Syst. 27(8), 2187–2195 (2021) 7. Wei, Y., Zhang, Y., Zhang, H., et al.: Research in third-party perspective technology for headmounted augmented/mixed reality devices. Comput. Eng. 47(6), 284–291 (2021)

Design of Intelligent Financial Education Assistant System Based on Blockchain Wen Tang1(B) and Zhaohu Zhang2 1 Liaoning Institute of Science and Technology, Benxi 117004, China

[email protected] 2 PowerChina Hubei Electric Engineering Co., Ltd., Wuhan 430000, China

Abstract. The traditional financial education assistant system has the problems of long response speed and poor anti-interference effect of financial education information transmission. Therefore, an intelligent financial education assistant system based on blockchain is designed. Collect intelligent financial education data according to the data warehouse technology, analyze the collected intelligent financial education data through statistical analysis methods, build an auxiliary decision-making model based on the comprehensive financial data, and use the blockchain to realize the design of intelligent financial education auxiliary system, so as to realize the data conversion of intelligent financial education. The experimental results show that the average response speed of the intelligent financial education assistant system and the other two intelligent financial education assistant systems are 60.530 s, 67.220 s and 68.133 s respectively; When the noise intensity is 50 dB, 80 dB and 120 dB, the design system always has a high education information transmission accuracy, which shows that the design system can improve the Financial Education Assistance Effect of the intelligent financial education assistance system, and that the intelligent financial education assistance system integrated with the blockchain technology has better performance. Keywords: Blockchain · Intellectualization · Finance · Educational support system · Decision model · Data characteristics

1 Introduction Financial education assistance is not only an important embodiment of ability indicators, but also one of the industries with very strong market demand. The development level of financial education assistance system largely determines the overall management level of a user. With the gradual opening of the national financial market, all users, including telecom users, are also facing great competitive pressure. Financial education assistant system is an indispensable information management system for each unit to manage financial revenue and expenditure. It plays a great role in users’ financial planning, financial control, financial supervision, financial application and so on [1, 2]. How to strengthen the management concept, reduce user costs and improve the competitiveness of users will be one of the key factors for the foothold of intelligent financial system. © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 62–74, 2022. https://doi.org/10.1007/978-3-031-21161-4_6

Design of Intelligent Financial Education Assistant System

63

After years of development, the financial information management of the user’s financial department has a certain scale and efficiency. However, there are also problems inconsistent with the assistance of modern financial education, such as the cooperation between different systems, the interconnection between different systems, and the phenomenon of information island still exists. Relevant scholars have made some progress in this field. In order to solve the problems existing in the reimbursement management of the current hospital, shixiaochuan and others have adopted the R & D mode of “mature software + customized development” under the promotion of blockchain and other technologies to build an intelligent financial system, but the problem of incomplete performance of the auxiliary decisionmaking model has been ignored in the design process [3]. Zhangxinqi designed a hospital financial system based on multi-scale deep learning. In the hardware design of the system, several sensors are added, and the coordinator is used to scan all the information in the financial database. In the software design, through the establishment of database forms, set the weights that can connect the same attribute financial information. According to the multi-layer perceptual network topology, a full interconnection mode based on multi-scale deep learning is designed to realize the system’s deep extraction of data. However, no detailed solution is proposed for the problem that the performance of the auxiliary decision-making model is not comprehensive [4]. In view of the above problems, this paper designs an intelligent financial education auxiliary system based on blockchain, collects intelligent financial education data according to data warehouse technology, analyzes intelligent financial education data through statistical analysis methods, and uses blockchain to build an auxiliary decisionmaking model to realize the design of intelligent financial education auxiliary system. The design system can effectively improve the accuracy of education information transmission and response speed.

2 Hardware Design of Intelligent Financial Education Assistance System The main objective of the demand analysis of the financial education auxiliary basic module is to develop a financial education auxiliary system that can meet the needs of users according to the needs of users, conduct a comprehensive investigation and Analysis on the needs of users, accurately define various design requirements of the development system, and describe in detail what the developed system can do, Whether the developed software can meet the needs of system users and whether it is developed according to the needs of users, so it is necessary to analyze the needs of users, comprehensively investigate and analyze users, and comprehensively clarify the needs of users. Considering the universality of industrial application, the data acquisition system designs analog, digital, switching, frequency and other data acquisition interface circuits and RS485 serial universal interface circuits according to the actual needs. Financial education auxiliary system is mainly used for external signal acquisition and equipment control. The collected signals are divided into three digital signals and three analog signals. The overall hardware architecture of intelligent financial education auxiliary system is shown in Fig. 1:

64

W. Tang and Z. Zhang

Fig. 1. The overall hardware architecture of the intelligent financial education assistance system

It can be seen from Fig. 1 that the system receives the data sent by the data acquisition system through the RS48_5 interface for processing, and displays the parsed information on the 8-in. LCD screen in real time. The operator can interact with the system through the 8-in. capacitive touch screen, which is convenient to operate. At the same time, the relay module control command is also controlled through the RS485 interface. Therefore, if Pro-SiVIC DDS data communication technology is used to send data from sensor objects in Pro-SiVIC, the output sampling frequency should not exceed the value specified in the above table. Ideally, external applications should set the same object data read rate. Otherwise, the data bandwidth flow between the sender and the receiver will be too large, which may lead to communication failure when exchanging a large amount of sensor data. Therefore, the system selects two microcontrollers, which are the MSP430 microcontroller from TI and the C8051 microcontroller from Silicon Laboratories. In the system, C801 and MSP430 communicate data through 11 IO ports, of which 8 IO ports are used for data transmission and reception, and three IO ports are used for data communication control. C805 1F has its own fully differential 24-bit Sigma-Delta analog-to-digital converter (ADC), with high acquisition accuracy and stable performance. The ADC features internal calibration, a sampling rate of up to 1 kHz, an on-chip temperature sensor and an asynchronous, full-duplex serial port. C8051 is responsible for the acquisition of three channels of analog data and the reception of one channel of RS485 sensor communication data, and then sends the data received by AD conversion and RS485 to the MSP430 microcontroller according to the protocol. All the collected data are encapsulated according to the protocol and sent to the data processing system. The specific protocol is shown in Table 1:

Design of Intelligent Financial Education Assistant System

65

Table 1. Data processing protocol Frame header

FEFB

Financial education auxiliary simulation channel

CRC check

Frame type

1 byte

Financial education aid RS485 communication 1

Financial education aid simulation channel 2

Financial education data length

2 byte

Financial education aid RS485 communication 2

Financial education aid simulation channel 1

Wind speed

1 byte

Financial education aid simulation channel 6

Financial education aid RS485 communication 1

Rotary encoding

2 byte

Financial education aid simulation channel 1

Financial education aid RS485 Communication 2

Inclination

6 byte

FEFA

Analog channel 1

As can be seen from Table 1, the data and instructions issued by the data processing system are received, and the reduced instruction set (RISC) architecture is adopted, which has strong processing capability and fast operation speed, and has a wide range of applications in the field of industrial control. DDS defines a service for efficiently distributing application data among the participants of a distributed application, which is not specific to a common object request broker architecture. This specification provides a Platform Independent Model (PIM) and a Platform Specific Model (PSM) that maps this PIM model to a CORBA IDL implementation. MSP430 is a 16-bit single-chip microcomputer. MSP430 is mainly responsible for the acquisition of frequency and digital quantities, the relay breaking control circuit, and the DAC waveform generation circuit with the DAC7614 chip as the core. MSP430 has four DART serial interfaces, three of which use RS48_5 serial communication mode, which can be extended to send and receive data from external AD acquisition modules. In the terminal design, the external AD acquisition module takes C8051 microcontroller as the core, and the external analog signal is converted by AD and then transmitted to the system through the RS485 interface.

3 Software Design of Intelligent Financial Education Assistance System 3.1 Collect Intelligent Financial Education Data Intelligent financial education assistance is mainly for managers to use scientific statistics, calculation and analysis methods to analyze various financial data, financial statements and operation data of users, so as to judge the financial status and operation results of users in a certain period, so as to evaluate the performance of users’ business activities systematically and objectively, Correctly predict the future of user development, so as to make corresponding decisions effectively. The acquisition target of system data has obvious intelligent characteristics. It is to prepare user investment plan according

66

W. Tang and Z. Zhang

to historical data and user development objectives. It has the function of automatic production and management of investment scheme. Individual evaluation of investment projects, such as economic evaluation, risk evaluation and social evaluation. The established intelligent model is used to realize the comprehensive evaluation of investment projects. Intelligent financial education assistance is not only the deep extension of users’ financial data and financial management activities, but also the core and goal of users’ financial management activities. Through the correlation analysis of various financial statements and financial data, we can further master users’ financial status, provide scientific, reasonable, correct and effective reference for users’ managers and decision makers to formulate business strategies. Through the analysis of financial statements, it can effectively evaluate the user’s three abilities, profitability, debt repayment and asset operation ability, predict the user’s future development trend, and provide a reliable basis for the user’s management to correctly carry out business management and financial decision-making. The data acquisition and extraction structure is shown in Fig. 2:

Fig. 2. Data acquisition and extraction structure

As can be seen from Fig. 2, the data collection step provides the whole system with a data support environment oriented to the subject of analysis, and provides functions such as statistical query, OLAP, and data mining. The comprehensive layer adopts massive data warehouse technology, collects and extracts the operational data of each department, analyzes and organizes it, and saves it into a special data warehouse. The design of an intelligent financial decision model must meet the overall goals of the system. It supports not only structured decision-making problems in user financial management decisions, but also semi-structured and partially unstructured decision-making problems. Support users’ financial management issues at different management levels, such as supporting high-level management’s macro-management decisions on financing, investment, profit distribution, and budgeting. 3. Support communication between users in order to support interdependent decision making. 4. Support users to modify and expand the database, model library, method library, knowledge base. 5. Provide convenient man-machine dialogue and good data transmission function. Intelligent financial education assistance

Design of Intelligent Financial Education Assistant System

67

not only provides reliable data and basis for financial activities, but also provides data for the decision-making and management of all aspects of user operation and management activities. Supervise and control, and provide specific basis for assessment and evaluation of user’s operation and management performance. 3.2 Blockchain Builds an Auxiliary Decision-Making Model Blockchain is the underlying technology of bitcoin and the core of building its data and transactions [5, 6]. Blockchain technology is a data chain that packages data into blocks and arranges them in chronological order. Through the technical advantages of computer encryption algorithm, it ensures that the data can not be tampered with and counterfeited [7]. Therefore, the blockchain based Ledger has the characteristics of decentralization, distrust and tampering. The availability of blockchain includes two aspects. On the one hand, from the perspective of developers, the API provided by blockchain applications is difficult to be used in other services and applications. Financial aided decision-making model is a decision-making model that automatically analyzes the financial situation of enterprises and provides accurate and timely information support for managers at all levels of enterprises to make decisions by using the knowledge of relevant computer technology, management and other disciplines and based on the financial comprehensive analysis model. Based on the above description, the calculation formula of capital preservation and appreciation rate can be obtained: H=

δ−q × 100% L

(1)

In formula (1), δ represents the total equity at the end of the period, q represents the amount of newly added (+ decreased) equity during the period, and L represents the total equity at the beginning of the period. According to the calculation results, the highest score of the index can be adjusted to 1.2 times the standard score value, and the lowest score can be adjusted to 0.6 times the standard score value. The average ratio difference per point is the difference between the 1-point ratios of the scores of various financial indicators. The calculation formula is: φ−m (2) D= η−h In formula (2), φ represents the highest ratio of the industry or company, m represents the standard ratio of the industry or company, η represents the highest score of the revised index of the industry or company, and h represents the revised standard score. When designing the model, the object-oriented design idea is adopted, and the major analysis themes and small sub-themes, as well as the dimension objects and dimension levels that are concerned when statistics and analysis of business information are determined according to the requirements. Financial information is a decision-making model. It is a complex computer-aided decision-making model centered on financial analysis. It involves related financial knowledge, operations research knowledge, statistical knowledge, artificial intelligence and computer knowledge. All related disciplines are interconnected and intersected. Integrated Auxiliary Decision Model. Financial information is in the core position in various application systems of the enterprise

68

W. Tang and Z. Zhang

and comprehensively reflects the overall operation of the enterprise. Its main function is to provide managers at all levels of the enterprise with all kinds of financial information related to the enterprise now and in the future, and provide managers with correct and timely decision-making services. According to the application type of data warehouse, a database specially used to store data and an expert system to store relevant knowledge information and models can be established respectively. The special database uses the massive storage, extraction and analysis function of the data warehouse to process and store the collected information. The stored information includes the internal information accumulated by the enterprise for a long time and the information related to the external environment. After determining the subject and dimension, the behavior record structure of users’ selection and use of investment business, i.e. fact table, is determined based on the division principle of the contact point between users and business support system. An analysis topic corresponds to one or more fact tables, and a sub topic corresponds to the fact table one by one in the system. The financial account book report is an important part of the financial management system. Through the enterprise’s account book report, the enterprise can effectively realize relevant financial analysis and statistics, flexibly define and configure various financial statements, and effectively maintain and manage various reports. The enterprise’s financial account book mainly includes account summary table, enterprise balance sheet The profit statement and cash flow statement of the enterprise. Internal information data usually includes enterprise financial data information, enterprise financial data information, enterprise sales data information, enterprise production data information and other related internal data. Support middle-level management control decisions, such as production decision, cost decision, sales decision, plan management control in inventory decision, etc. Support grass-roots operation control decisions, such as production decisions, sales decisions, inventory decisions, etc. The special data warehouse should be able to automatically identify the integrity and effectiveness of the data, check the data source from time to time, and automatically update the data warehouse in real time once the data of the data source is found to change, so as to maintain the integrity and effectiveness of the data. 3.3 Design System Software Data Conversion Function The blockchain distributed ledger sets up multiple nodes through the same algorithm. These nodes work together to maintain the update and operation of the network. Multiple nodes can ignore the distance space for transaction and data sharing at any time. Nodes access the ledger through the public key and private key. Any transaction initiated between nodes will be broadcast in the blockchain network. The bookkeeping node obtains the bookkeeping right through the algorithm, records the transaction information completely and adds it to the blockchain information. The bookkeeping node will receive bookkeeping reward. Other nodes will update their blockchain ledger in time after receiving the broadcast of successful bookkeeping. Once recorded, the information in the ledger cannot be changed at will. Each node has the same real and tamper proof ledger, which is the unique bookkeeping method of blockchain. Data warehouse solves the problem of inconsistent data. While collecting a large amount of transaction level data from the underlying database, the data warehouse integrates, transforms and synthesizes the data. The calculation formula of single indicator data in the conversion

Design of Intelligent Financial Education Assistant System

69

system is: G=r−

t z

(3)

In formula (3), r represents the actual ratio of the calculated indicator, t represents the standard ratio of the calculated indicator, and z represents the ratio difference per minute of the indicator. The calculation formulas for the data of multiple indicators in the conversion system are:  W = V +K (4) In formula (4), V represents the standard score of the calculation index, and K represents the adjusted score. The establishment of the data warehouse and the extraction and conversion of source data in this system use Microsoft’s data warehouse generation tools and data conversion service tools to convert data from different data sources into a unified SQLSEVER database format. The application management of blockchain can be classified into three categories: one is basic information storage, that is, users can establish a database on the blockchain network as a proof of retention. The second is a more complex authentication application, which uses the blockchain network to process complex logical data, such as binding personal identity information with various applications and transaction information, and directly using the combination of the main chain and the branch chain. Link to achieve faster and faster data support. It ensures that no matter what kind of model construction, mathematical method and programming language is used, as long as the effective model can be run independently, each sub-model can become the model resource of the system, which is uniformly managed and scheduled by the system, and the model is easy to be repeated. Use. The subroutine model operation method is adopted, and the model is connected with the system through the plug-in loading method, which ensures the relative fixedness of the system structure and the relative independence of the model, makes full use of the existing resources, and makes the system have better scalability, Maintainability and generality [8–10]. The third is transaction-based applications, such as the free exchange of different currencies through the blockchain network, which requires the overall operation of the entire blockchain network to complete multiple links such as initiation, review, transaction, and accounting. From a financial point of view, the successful implementation of the three functions is completed by relying on the distributed ledger under the blockchain technology.

4 Simulation Analysis 4.1 Test Preparation After configuring JDK and other related basic equipment, log in using the Internet, enter the URL in the address bar of the home page, and enter the login interface after entering the URL. When debugging the MiniUSB interface, after connecting the PC and the MiniUSB interface with a data cable, it is found that the connection cannot be made normally, and the PC cannot identify the peripherals of the data processing system where the MiniUSB interface is located. There is no design error in the analysis of the circuit

70

W. Tang and Z. Zhang

schematic diagram, and there is no leakage or virtual soldering of the components. Supply power to the 5 V system power supply outside the MiniUSB to eliminate the reason that the USB interface of the PC has insufficient driving capability, but the MiniUSB still cannot be connected normally. Use the core version of the programmed Linux system for testing. When the system is powered on and started, you can see the kernel printout information through the serial port. 4.2 Analysis of Results Response Time The intelligent financial education assistant system based on Internet of things and the intelligent financial education assistant system based on SOA are selected for experimental comparison with the intelligent financial education assistant system in this paper to test the response time of the three systems under different concurrent users. The experimental results are shown in Figs. 3, 4 and 5:

Fig. 3. The number of concurrent users is 50. The response speed (s)

As can be seen from Fig. 3, the average response speed of the intelligent financial education auxiliary system in this paper and the other two intelligent financial education auxiliary systems are 10.698 s, 15.303 s and 16.442 s respectively; As can be seen from Fig. 4, the average response speed of the intelligent financial education auxiliary system in this paper and the other two intelligent financial education auxiliary systems are 63.576 s, 72.151 s and 71.988 s respectively; As can be seen from Fig. 5, the average response speed of the intelligent financial education auxiliary system in this paper and

Design of Intelligent Financial Education Assistant System

Fig. 4. The number of concurrent users is 300. The response speed (s)

Fig. 5. The number of concurrent users is 500. The response speed (s)

71

72

W. Tang and Z. Zhang

Fig. 6. Transmission accuracy of educational information under different noise interference

Design of Intelligent Financial Education Assistant System

73

the other two intelligent financial education auxiliary systems are 107.315 s, 114.205 s and 115.968 s respectively. Anti Interference Effect of Financial Education Information Transmission In order to verify the anti-interference effect of the financial education information transmission of this method, the intelligent financial education auxiliary system based on the Internet of things and the intelligent financial education auxiliary system based on SOA are used to compare with the intelligent financial education auxiliary system in this paper. The anti-interference effects of the financial education information transmission under the noise intensity of 50 dB, 80 dB and 120 dB are obtained respectively. According to Fig. 6, when the noise intensity is 50 dB and the number of experiments is 10, the education information transmission accuracy of the Internet of things system is 90.5%, the education information transmission accuracy of the SOA system is 92.6%, and the education information transmission accuracy of the designed system is 98%; When the noise intensity is 80 dB and the number of experiments is 10, the accuracy of educational information transmission of the Internet of things system is 83%, that of the SOA system is 85%, and that of the designed system is 97%; When the noise intensity is 120 dB and the number of experiments is 10, the accuracy of educational information transmission of the Internet of things system is 63.8%, that of the SOA system is 62%, and that of the designed system is 96.8%; The above data shows that the design system always has a high accuracy of education information transmission when the noise intensity is 50 dB, 80 dB and 120 dB, indicating that the design system can improve the Financial Education Assistance Effect of the intelligent financial education assistance system.

5 Conclusion This paper designs an intelligent financial education assistant system based on blockchain. Collect intelligent financial education data according to the data warehouse technology, build an auxiliary decision-making model using the blockchain, and realize the design of intelligent financial education auxiliary system. The experimental results show that: (1) The average response speed of the intelligent financial education auxiliary system in this paper and the other two intelligent financial education auxiliary systems are 60.530 s, 67.220 s and 68.133 s respectively, indicating that the intelligent financial education auxiliary system integrated with blockchain technology has better performance. (2) When the noise intensity is 50 dB, 80 dB and 120 dB, the accuracy of education information transmission of the designed system is 98%, 97% and 96.8% respectively, indicating that the designed system can improve the Financial Education Assistance Effect of the intelligent financial education assistance system.

Fund Project. 2022 Liaoning University of Science and Technology College Students’ Innovation and Entrepreneurship Training Project “Step into the Cloud - Smart Cloud Financial Business Practice” (202211430016).

74

W. Tang and Z. Zhang

References 1. Ti, Y.: Research on financial system design based on oracle database and JAVA language. Tech. Autom. Appl. 39(7), 170–174 (2020) 2. Lei, Z., Shi, X., Lei, Y.: The practice and exploration of hospital fund management based on the intelligent financial system. Chin. Health Econ. 40(8), 81–83 (2021) 3. Shi, X., Lei, Z.: Practice and exploration on the construction of hospital intelligent financial system based on the whole process. Chin. Health Econ. 39(3), 86–88 (2020) 4. Zhang, X.: Design of hospital financial system based on multi-scale deep learning. Microcomput. Appl. 36(11), 143–146 (2020) 5. Niu, S., Xie, Y., Yang, P., et al.: Cloud-assisted attribute-based searchable encryption scheme on blockchain. J. Comput. Res. Dev. 58(4), 811–821 (2021) 6. Xu, Z., Wang, Y., Wang, X.: Research on cloud storage optimization of unstructured big data combined with blockchain. Comput. Simul. 38(7), 304–307, 354 (2021) 7. Zhang, D.: Research on counter-terrorism intelligence fusion and sharing model based on blockchain technology. J. Intell. 40(2), 62, 69–76 (2021) 8. Zeng, H., Hu, Y.: Research on the design and implementation of the teaching assistant system for information technology in primary schools. Digit. Technol. Appl. 39(02), 133–135 (2021) 9. Gao, S.: Design and implementation of java intelligent teaching assistant system. Inf. Rec. Mater. 21(05), 171–172 (2020) 10. Yang, J.Z., Zhang, H.: Design and implementation of an Android based field geology teaching aid system. China Geol. Educ. 28(03), 92-96 (2019)

Enterprise Strategic Financing Risk Management Auxiliary Education System Based on Data Mining Xiang Zou1(B) and Jing Zhu2 1 School of Accounting and Finance, Wuxi Vocational Institute of Commerce, Wuxi 214153,

China [email protected] 2 Department of Business and Trade, Guangzhou Vocational College of Technology and Business, PanYu 511442, China

Abstract. In order to ensure the use effect of enterprise strategic financing risk management auxiliary education system, an enterprise strategic financing risk management auxiliary education system based on data mining is designed. By designing the system hardware structure, combined with data mining algorithm, optimize the functional structure of the system software, simplify the auxiliary education process of enterprise strategic financing risk management, and realize the auxiliary education of enterprise strategic financing risk management. Through the analysis of the system test results, it can be seen that the enterprise strategic financing risk management auxiliary education system based on data mining has high practicability and fully meets the user experience in the process of practical application. Keywords: Data mining · Enterprise strategic financing · Risk management · Supplementary education system

1 Introduction At present, the competition of various enterprises is becoming increasingly fierce, and their requirements for the management of enterprise strategic financing are becoming higher and higher, which also puts forward higher requirements for the professional practitioners related to enterprise strategic financing. The conventional teaching mode can not meet the current development needs, so it is necessary to adjust and improve the teaching methods. Using data mining virtual business social environment teaching system is a better way, which can effectively simulate the units and posts in the real business social environment. Students can learn and train in this virtual environment, and learn the contents and work characteristics of each position, so as to improve their learning and execution ability, improve their overall awareness and vocational skills, and then improve students’ learning interest and professional skills, so as to lay a foundation © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 75–90, 2022. https://doi.org/10.1007/978-3-031-21161-4_7

76

X. Zou and J. Zhu

for future practical work. As for the teaching of enterprise strategic financing management in colleges and universities, it mainly includes theoretical teaching and auxiliary teaching, which involves the collection, analysis and processing of a large number of enterprise strategic financing data. The traditional data information processing method is not only inefficient, but also has poor processing quality due to human errors. Therefore, the effective application of data mining processing method in college teaching is very important [2]. Li minglun [1] proposed to put the safety education of college students in front, built a set of safety knowledge question base based on recommendation algorithm, and developed a practical online learning and testing system of safety knowledge. This method can effectively improve the safety awareness of college students. However, with the increase of the number of students, the response time of the system is long and can not meet the needs of more users. The enterprise strategic financing risk management auxiliary education system based on data mining combines advanced teaching ideas and enterprise management methods, optimizes the functional modules and business of the system, and designs a system that can provide students with a real internship environment. And achieve high consistency in the enterprise strategic financing environment. In this way, interns can really understand the whole process of enterprise organization and management. The use of enterprise strategic financing risk system has significantly improved the training effect, education and teaching level and information level. It can be directly applied to the teaching of enterprise strategic financing management. At the same time, the training of other majors also has strong reference significance.

2 Enterprise Strategic Financing Risk Management Auxiliary Education System Based on Data Mining 2.1 Hardware Structure of Auxiliary Education System for Financing Risk Management The construction of resource bank in the auxiliary education system of financing risk management is different from the construction of online courses, mainly for small mobile learning terminals [2]. Therefore, when developing courseware applied in mobile learning, we should pay attention to this factor. First of all, the courseware needs to consider the size, followed by the choice of granularity. Thirdly, the theme is clear and can stimulate interest [3]. However, it does not mean the repeated construction of the online course resources on the original Internet. In fact, the educational resources developed based on learning objects can undoubtedly play a great role here. The mode structure of the current mobile learning system is shown in Fig. 1. With the advent of 5g network era, the further optimization of wireless internet speed and wireless network security, the further price reduction of various access wireless network terminal equipment and the reduction of wireless internet charges will inevitably promote a new round of mobile learning boom. Therefore, the extended mobile examination and mobile incentive will also highlight its greater application advantages with the in-depth research of mobile learning [4]. All walks of life are constantly applying computer software to improve the efficiency of production and work, and the investment

Enterprise Strategic Financing Risk Management Auxiliary Education System

77

Internet

GPRS network

Data trans mission terminal Mobile phone

Computer

Fig. 1. Model structure of mobile learning system

in the development and use of enterprise strategic financing risk management auxiliary education system is also increasing. The development of enterprise strategic financing risk management auxiliary education system continues to produce new technologies from the earliest static page to dynamic page, so as to better meet the needs of enterprises for the system [5]. The emergence of Java EE provides a lot of middleware for software development, which can better reduce the cost of software development and maintenance. At the same time, JavaEE also supports technologies such as JsP and Java Servlets, and further optimizes the system application architecture based on this, as shown in Fig. 2.

J2EE application I

Application client

EJB

Dynamic HTML page

JSP page

Database

Database J2EE application II

EJB Customer layer

Client

Business layer

Enterprise information system layer

J2EE server

Database server

Fig. 2. System application architecture

The teaching system of enterprise strategic financing management based on data mining contains a comprehensive business process. According to the actual situation, after fully investigating teachers, students and various departments involved in teaching, sort out and form functional requirements, including data entry, modification, query, generation and so on. The system provides a visual window and has a powerful function

78

X. Zou and J. Zhu

of data query [6]. The development system of enterprise strategic financing risk system adopts Windows2007 and above operating system, XP operating system, the server side adopts ux server, and uses Myeclipse as the background development tool. Through the collection and analysis of remote connection, the goal of auxiliary teaching is achieved. During the course of enterprise strategic financing risk teaching, students log into the virtual instrument control interface from the warehousing of the network system, and through corresponding operations on the PC, the D/A converter converts the operation information into analog signals. That is, the digital signal is converted into an analog signal, and the controller performs corresponding operations on the physical instruments to complete the operation of assisting the teaching of corporate strategic financing risks. Then the sensor will convert the analog signal to the digital signal through the AD converter for the operation result of the auxiliary teaching of enterprise strategic financing risk, and then display the operation result on the PC. In this way, the auxiliary teaching of physical equipment is completed. The hardware structure of the enterprise strategic financing risk auxiliary teaching system is shown in Fig. 3.

Computer network

PC

Virtual instrument

D/A

A/D

Controller

Sensor

Phys ical instrument

Fig. 3. System hardware structure optimization

Students can access a large number of teaching resources of corporate strategic financing management through computers and mobile Internet. Among them, the lack of practical textual research and lack of scientific basis for the auxiliary learning materials of corporate strategic financing management, once students come into contact with such teaching resources, it is very easy to make learning mistakes. The main reason for this problem is that the teaching managers of corporate strategic financing risk in colleges and universities and the teaching teachers of corporate strategic financing management have not been able to effectively classify the resources on the campus learning network system, which directly leads to the fact that the actual corporate strategic financing teaching resources are relatively low dispersion. The lack of systematicness makes it difficult for students to form a systematic professional knowledge structure of corporate

Enterprise Strategic Financing Risk Management Auxiliary Education System

79

strategic financing management in the actual learning and utilization process, which naturally greatly reduces the overall quality and efficiency of the teaching of corporate strategic financing management in colleges and universities. 2.2 System Software Function Optimization In data mining, the research problem of corporate strategic financing risk early warning is positioned as a classification problem, and its goal is to classify different listed companies through the analysis of corporate strategic financing indicators, so as to give early warning. Therefore, this paper uses the technology in data mining to predict the listed company to get whether the company may have a corporate strategic financing crisis, and finally select the optimal model through comparison [7]. The use of the enterprise strategic financing risk management auxiliary education system to establish the learning knowledge points, and the combination of the curriculum knowledge material resources and the enterprise strategic financing risk management auxiliary education system can ensure its long-term use and continuous updating. A course teaching system based on data mining for virtual business social enterprise strategic financing management specialty [8]. Carry out the teaching course construction of virtual business society practice enterprise strategic financing management, and formulate the course construction plan according to the professional talent training plan; The second is to formulate an appropriate curriculum implementation plan; Based on the current teaching management system, combine the curriculum construction and implementation scheme, and form a reasonable syllabus. Finally, mobilize the resources and links of relevant knowledge points according to the syllabus to carry out each learning unit of auxiliary teaching. Based on the previous system functional requirements, the overall planning of the system is carried out, which is divided into three functional modules: teaching function module, training function module and teaching management function module, as shown in Fig. 4. In the above four sub modules, the classroom interaction function is initiated by teachers and done by students. Students can timely and effectively ask questions and interact with teachers in the learning process, which is conducive to teachers to find students’ deficiencies in class and improve them. Through the course selection function, teachers can establish courses, add course information and manage courses, and modify and delete courses. Through the course selection function, students can add courses to join the study, and obtain credits after completing the corresponding study. The teacher issues the homework and notice, and the students submit the completed homework and get the corresponding evaluation. After the significance test of the index, the dimension of the index is reduced. Principal component analysis integrates many original variables into a few component factors, and replaces the original variables with the newly formed factors for modeling. Assuming that Fm is the component factor formed by the first linear combination of the original variables, the information extracted by each principal component factor of am1 , am2 , ..., amn can be measured by variance. The larger the component variance Fm of the principal component factor, the greater the amount of

80

X. Zou and J. Zhu Initiate problem

Classroom interaction

Ask ques tions Question answering Establish curriculum

Course selection

Add course information Join the course

Teaching function

Job notification Job communication

Submit job Evaluation operation Courseware upload

Courseware upload and download

Courseware download

Browse courseware

Fig. 4. Functional structure of system software

information contained in Fm . Fm and other principal component factors are independent of each other, where Xn is the n principal components of the original variable. Its relationship is as follows: ⎧ ⎪ ⎪ F1 = a11 X1 + a12 X2 + . . . + a1n Xn ⎪ ⎨ F2 = a21 X1 + a22 X2 + . . . + a2n Xn (1) .. ⎪ ⎪ . ⎪ ⎩ Fm = am1 X1 + am2 X2 + . . . + amn Xn Taking the establishment of enterprise strategic financing risk early warning model as an example to illustrate the principle of Logistic regression analysis. Set the dependent variable P0 = 0 to represent the company in crisis with corporate strategic financing, and P1 = 1 to represent the company with normal corporate strategic financing. P represents the probability that the Y = 0 listed company will have a corporate strategic financing crisis, and P = 1−P1 represents the probability that the Y = 1, the greater the amount of information in principal component factor Fm , the smaller the strategic financing crisis. Listed company will not have a corporate strategic financing crisis:   P1 − α + βk log(P1 ) = Fm − ln (2) 1 − P0 The teaching function of the system provides teachers with the functions of uploading and downloading courseware and timely updating. Teachers can play corresponding

Enterprise Strategic Financing Risk Management Auxiliary Education System

81

resources during classroom teaching, such as classroom presentation slides, etc. At the same time, teachers can also modify and delete the resource. Students can download and browse, so as to better master the classroom learning content. The sub modules involved in this function module are shown in Table 1. Table 1. Sub module division Sub module name

Function description

Classroom interaction

Initiate and ask questions, students participate and solve problems on time

Course selection

Teachers establish courses, add course information, and students join to complete learning tasks

Job communication

Teachers issue assignments and notices, and students submit completed assignments

Courseware upload and download

Teachers pass on materials and students download materials to better master the classroom content

Due to the constraints of network and bandwidth, most of the courseware produced is animation. Because it needs a large amount of storage and is subject to the network bandwidth, it can not be transmitted smoothly, resulting in its low utilization rate and can not be widely used. 2.3 Realization of Enterprise Strategic Financing Risk Management Auxiliary Education Data mining is a complete and procedural work. This process can mine previously unknown, effective and practical information from large databases, and use this information to make decisions or form rich knowledge. The whole process is in a specific data mining environment. The process of data mining generally consists of three stages: data preparation stage, data mining stage and result interpretation and evaluation stage. The data mining process of enterprise strategic financing information is shown in Fig. 5. The enterprise strategic financing management simulation system based on data mining is a system under the auxiliary teaching simulation environment of enterprise strategic financing risk management. It is not only the development of general enterprise strategic financing management simulation system, but also takes into account the integration and teaching needs of enterprise strategic financing risk management auxiliary teaching simulation system. Enterprise strategic financing management simulation system is a targeted simulation of the actual enterprise strategic financing management system according to the requirements of teaching. It has all the functions of the enterprise strategic financing management system in the existing enterprise teaching simulation system. At the same time, the enterprise strategic financing budget and cost management module are newly added to make the simulation system more comprehensive in showing the enterprise strategic financing management thought. The functional composition of the

X. Zou and J. Zhu

Data source

Data source

Data preparation

Data

Selection of data

82

Pretreatment

Target data

Knowledge

Interpretation of results

Pattern

Data mining

Preproces sed data

Fig. 5. Data mining process of enterprise strategic financing information

enterprise strategic financing management simulation system based on data mining is shown in Fig. 6. Among them, the top layer of the diagram is the interface that interacts with users. It constitutes the user presentation layer of the application. It is composed of business process management, web service components and data access components. They provide the implementation of various business rules and business logic for the user presentation layer, and conduct data interaction with various databases by calling data service components. Risk control is the last link of risk management. Risk control shall be carried out according to the results of risk identification, risk analysis and risk evaluation. For better risk control, the results of risk identification, risk analysis and risk evaluation are summarized. The phased results of risk management are shown in Table 2. The bottom layer of the system is the database, which stores the business data related to the enterprise strategic financing management. Among them, the enterprise strategic financing budget, cost management, accounts receivable management and other modules are internally integrated through data mining, so as to achieve loose coupling of the system and increase the flexibility of the system. The enterprise strategic financing risk management system adopts the service-oriented integration method, and divides the enterprise strategic financing risk management system service into two modules: enterprise strategic financing accounting service and management accounting service. The enterprise strategic financing accounting service module includes general ledger management service, accounts payable management service, cash management service, fixed assets management service and report management service. The services included

Enterprise Strategic Financing Risk Management Auxiliary Education System

83

General Accounting

Accounts receivable

Financial budget

Accounts payable management SQL Server 2000 financial database

Cost control Fund management

Fixed assets management

Business process management

Accounts receivable

Data access operation

Cost control Financial analysis

Financial budget

Other databases of SQL Server 2000 ERP simulation system

Financial analysis

Financial decision

Fig. 6. Functional composition of enterprise strategic financing management simulation system

Table 2. Phased results of risk management Stage

Achievements

Risk identification Existing problems: national policies, laws and regulations and internal management of ERP rises Risk analysis

Bank lending and financing behavior and joint default ratio, pledge ratio, bank supervision cost, etc

Risk evaluation

Among the primary indicators, the top three are: financial, reputation and risk of benefit distribution; Among the secondary indicators, the top five weights are: solvency, financing scale, distribution scheme, administrative penalty and timeliness of delivery

in the management accounting service module include enterprise strategic financing budget service, cost management service, enterprise strategic financing analysis and enterprise strategic financing decision-making service. These modules use the interface provided by web services for mutual data exchange and access. The specific process is shown in Fig. 7. Based on the analysis and architecture design of enterprise strategic financing risk management system, combined with the hierarchical architecture and overall function of the system, this paper encapsulates the function in the form of service, and designs its granularity. The data mining method is used to analyze and design the services in the enterprise strategic financing management simulation system. When designing services,

84

X. Zou and J. Zhu

Financial budget Bpfl engine Database

Client

Web Service Bus

Web server

Cost control

Accounts receivable Database WSDL

Service rule business activity monitoring exception handling

Financial analysis

Fig. 7. Data access process of enterprise strategic financing management simulation system

we should follow the principles of business and system function alignment, reusability, statelessness and so on. In the enterprise strategic financing budget module, the budget preparation service, budget adjustment service and budget review service are designed. The budget preparation service is mainly composed of budget adjustment service, budget summary service and budget balance service. In the cost management module, the cost planning service, cost accounting service, cost control service and cost analysis service are designed. The teaching system adopts B / S architecture technology for development, and realizes the interactive function of data from three levels: surface layer, business logic layer and data layer. The first layer is the user layer. Users don’t need to care about how the function is realized. Here, they can provide an intuitive, friendly and easy to operate web interface, which can be operated directly according to their own needs; The second layer is the business logic layer, which reflects the realization of corresponding functions through logical judgment; The third layer is the data layer, where the data of the second layer can be extracted and routinely edited, and the upper layer can be fed back at any time. Through the research and development of the three-tier separation principle, the development cost is greatly reduced and the development efficiency is improved. The business function of the financing risk auxiliary management simulation teaching system includes the following three levels. The overall framework structure and function of the system are shown in Fig. 8. System function module, which is a collection of execution statements, data descriptions and other elements. In the development of enterprise strategic financing risk management system, functional modularization can reduce the workload of system development and reduce the development cost. However, it is not equal to the unrestricted division of enterprise strategic financing risk management system. It requires that each independent module must maintain a certain connection and meet the overall function of enterprise strategic financing risk management system. Therefore, determining the appropriate functional modules will play a decisive role in the success of system development.

Enterprise Strategic Financing Risk Management Auxiliary Education System Presentation layer

Business logic layer

Data layer

Student

Organization management page

Organization management page control

Teaching Organizer

Customer management page

Customer management page control

Business logic of customer management page

Customer management information data

Teaching management page

Teaching management page control

Teaching management page business logic

Teaching management information data

Organizational business logic

Organization information data

ADO NET System administrator

85

Database

Fig. 8. Overall framework structure and function of the system

3 Analysis of System Test Results In order to verify the effectiveness of the designed enterprise strategic financing risk management auxiliary education system based on data mining, using windows10 system and Apache server, the functional modules of the enterprise strategic financing risk management system were tested 100 times under the same experimental environment, first explain the test objectives of the enterprise strategic financing risk management system, and then test the functions of each module. The test objectives of the system are shown in Table 3. Table 3. System functional test objectives Test scope

Each functional module

Target

Test the errors of each module of the system, modify them pertinently, and make the system achieve the expected function

Process

When the system is operated correctly and valid data is input, each functional module can operate normally; When the operation is wrong, the system will give a response error prompt

Completion criteria

The data output of each functional module of the system meets the expectation

The test cases are designed according to the test content, and the testers need to execute according to the appropriate process. The specific test is shown in Table 4. Based on the above test contents, its tests can be divided into four categories, such as basic function test, web function test, usability test and other tests. The function test of enterprise strategic financing risk management system is to compare the functions and requirements realized by the enterprise strategic financing risk management system and determine whether they are consistent. For example, whether the training of professional

86

X. Zou and J. Zhu Table 4. System test cases

Content

Serial number

Operation content

Expected results

Test

A

Administrator login and add course information

Complete the test

B

Apply for deletion function test

It can be used normally

C

Teacher login

It can be used normally

D

Student login

The addition can be completed

courses can be completed according to the user’s authority and processing order, whether the course selection link can be completed smoothly whether students’ grades can be displayed and submitted correctly, etc. Therefore, all the functions of the system are tested according to the needs of users. This paper selects the enterprise strategic financing risk management course information management for the test. The following table is the content of the course information management function test, as shown in Table 5. Table 5. Contents of course information management test Test item

User role

Test content

Add course

Administrators

Add professional courses

Course management

Teacher A

Course information management

Course selection information management

Teacher B

Manage course selection information

Add course

Administrators

Manage course selection information

Add course

Student

Manage course selection information

Set test cases for the student sub module to test whether the functions of the student sub module meet the expectations. The test cases include: course selection operation, student achievement and credit query, student status change application, course Q&A and student registration. The function test of enrollment management module is shown in Table 6. White box testing is also called logic driven testing or structure testing, and its principle is just opposite to black box testing. This test method can regard the test program as an open box. The tester mainly judges whether the enterprise strategic financing risk management system operates according to the established path and method when he has understood its internal structure. The scope of function test is shown in Table 7. Using the prediction probability and whether the industry defaults as the investigation variables, the ROC curve of the operation of the enterprise strategic financing risk management system is obtained, and the ROC auxiliary management curve is shown in Fig. 9.

Enterprise Strategic Financing Risk Management Auxiliary Education System

87

Table 6. Function test of enrollment management module Test no.: ychaign006 Test purpose: does the function of enrollment management module meet the expectations System role: recruiter Serial number

Behavior

Expected results

Does it meet expectations

A

Enter the student account, Enter the student password and select the role welcome interface “student”

Yes

B

Click the directory to select Pop up drop-down menu courses

Yes

C

Click course to learn

Pop up drop-down menu

Yes

D

Click on the list of grades and credits

Pop up drop-down menu

Yes

E

Click the directory to apply for student status change

Pop up drop-down menu

Yes

F

Click directory personal information

G

Click any drop-down menu

The management interface pops up

Yes

H

Click register

Enter the registration information interface

Yes

Yes

Table 7. Scope of function test Expected target

Test whether the relevant functional modules of the exhibition management simulation teaching platform can run in different environments

Test scope

Test whether the basic functions of each module are stable

Testing technology

Combination of black box and white box test

Test standard

The R & D of the exhibition management simulation teaching platform has been completed, and a detailed test plan has been formulated to feed back errors in time

Test focus

Exhibition organization, exhibition service, customer management, exhibition management, teaching management

It can be seen from Fig. 9 that the ROC curve is a comprehensive indicator reflecting the sensitivity and specificity of continuous variables. Generally, the area under the ROC curve is between 0 and 1. The closer the area under the ROC curve is to 1, the better the analysis effect. The area under the curve in the figure is 0.763, which is greater than the real area value of 0.5, and its significance is 0.001 < 0.05. Therefore, it shows that the

88

X. Zou and J. Zhu ROC curve 1.5

1.2

Sensitivity

0.9

0.6

0.3

0.0 0.0

0.3

0.6

1.2

0.9

1.5

Specifi cit y

Fig. 9. ROC auxiliary management curve

logistic model can well distinguish financing analysis, and the overall prediction effect of the model is good, so the prediction probability obtained by the model is effective. This paper designs a total of 30 test sub items, and then designs the use cases for each test sub item, which will not be repeated here. The final test proves that the system can be used normally, which can prove that the system basically meets the expected requirements in terms of function realization, the performance can also meet the user experience, and is more appropriate in terms of security and operability. In order to verify the effectiveness of the proposed method, when the number of users is between 0 and 500, the time taken to test the proposed method and reference [1] is shown in Fig. 10. 60 Proposed method Reference [3]

50

Time taken/s

40

30 20 10

0 0

50

100

150

200

250

300

350

400

450

500

Number of users / person

Fig. 10. Time taken to test the proposed method and reference [1]

It can be seen from the results in Fig. 10 that the time taken by the proposed method and reference [1] increases with the increase of the number of users. When the number of

Enterprise Strategic Financing Risk Management Auxiliary Education System

89

users reaches 500, the time of the proposed method is 30 s, and the time of the reference [1] is 55 s. It can be seen that the proposed method is more efficient and practical, and can fully meet the user experience

4 Conclusion With the rapid development of China’s social economy, the competition between various industries has gradually evolved into the competition of professionals. Enterprise strategic financing management is an indispensable work content in the process of enterprise operation and development. In view of this aspect, this paper selects the appropriate hardware structure according to the size of the courseware, optimizes the system application architecture through JavaEE, and improves the overall quality of teaching. Then the data mining method is used to analyze and design the services in the enterprise strategic financing management simulation system, so as to improve the transmission speed of distance teaching courseware. Based on data mining, the enterprise strategic financing risk management auxiliary education is realized, which reduces the workload of system development and research cost. The system test results show that the proposed method can improve users’ learning efficiency and has high practicability.

Fund Project

1. “Qinglan Project” of Jiangsu Universities - Training Project for Outstanding Young Backbone Teachers (RS21QL01). 2. “Excellent Project of Social Science Application Research in Jiangsu Province” - Special Project of Collaborative Innovation Base (22XTB-77). 3. Key Project of the “14th Five-Year Plan” for Educational Science in Jiangsu Province (B/2022/02/44). 4. Provincial and Ministerial Subject Cultivation Project of Wuxi Vocational Institute of Commerce (KJXJ21602).

References 1. Li, M.: Design and research of college students safety education system based on recommendation algorithm. J. Qingdao Univ. Nat. Sci. Ed. 32(04), 72–78 (2019) 2. Niu, X.: Simulation of homogeneous network intrusion detection based on threshold joint discrimination. Comput. Simul. 34(3), 309–313 (2020) 3. Safarnejad, L., Xu, Q., Ge, Y., et al.: A multiple feature category data mining and machine learning approach to characterize and detect health misinformation on social media. IEEE Internet Comput. 25(5), 43–51 (2021) 4. Klimova, S., Zhampeiis, N., Grigoryan, A.: Contemporary approaches to money laundering/terrorism financing risk assessment and methods of its automation in commercial banks. Procedia Comput. Sci. 169, 380–387 (2020)

90

X. Zou and J. Zhu

5. Zhang, Y., Chen, W.: Optimal production and financing portfolio strategies for a capitalconstrained closed-loop supply chain with OEM remanufacturing. J. Clean. Prod. 279(10), 123467 (2020) 6. Liu, C., Chen, K., Li, M., et al.: Trade credit and revenue sharing of supply chain with a risk-averse retailer. Math. Probl. Eng. 2021(6), 1–15 (2021) 7. Alola, A.A.: Risk to investment and renewables production in the United States: an inference for environmental sustainability. J. Clean. Prod. 312(2), 127652 (2021) 8. Feldman, D., Jones-Albertus, R., Margolis, R.: Quantifying the impact of R&D on PV project financing costs. Energy Policy 142(3), 111525 (2020)

Leg Posture Correction System for Physical Education Students Based on Multimodal Information Processing Lei Yang1(B) and Yueguo Jia2 1 Xi’an Medical College, Xi’an 710021, China

[email protected] 2 Tianjin Public Security Professional College, Tianjin 300382, China

Abstract. The traditional leg posture correction system has the problem that the joint points are arranged in reverse order and connected incorrectly, which affects the accuracy of posture recognition. In response to this problem, this research designed a leg posture correction system for students in physical education class based on multi-modal information processing. In the hardware part of the system, a signal conditioning circuit is used to filter, amplify, and sample the input signal, and an operational amplifier with a zero-adjusting terminal is used in conjunction with a D/A converter to realize the zero-adjustment of the circuit. In the software part of the system, after segmenting the depth image of the scene object containing the viewpoint, establish a database of the leg pose of the students in physical education class, then fuse the data vector and use the multi-modal information processing model to recognize the leg pose, and use the recognition result as the misrecognition probability matrix model The input to realize the intelligent error correction of the wrong leg posture. The experimental results show that under the test condition of 100 users, the positioning accuracy of the leg parts of the system in this paper is as high as 86.85%, which proves that it has a good application effect. Keywords: Multimodal information processing · PEclass · Leg posture · Attitude correction

1 Introduction Speaking of sports, from the educational level, it is a teaching process to promote sports, improve human quality, preach the skills and methods of physical exercise, and promote the all-round development of morality and will quality. At the same time, physical education is the experience of cultivating and shaping the body, and it is an important means to cultivate people into wholehearted development. The goal and key of physical education is to pursue standardized sports indicators and scientifically guide training and teaching. The introduction of human posture recognition into sports can provide a new idea and scheme for sports training and action evaluation. Attitude correction system is of © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 91–102, 2022. https://doi.org/10.1007/978-3-031-21161-4_8

92

L. Yang and Y. Jia

great significance to improve motion quality. Human pose recognition is defined as the localization of human joints in images or videos. It is also defined as searching for specific poses in the space composed of all joint poses. At present, the research on motion attitude error correction system has achieved some research results. In reference [1], a sprint leg posture monitoring system based on inertial sensor is proposed. The system uses wireless sensor network to design distributed sprint leg posture information acquisition nodes, installs pressure sensors and inertial sensors in athletes’ legs, and carries out information fusion and association mining according to the output oscillation amplitude of the sensors, so as to realize the intelligent monitoring of sprint leg posture data. A gait control system based on quantitative fusion of inertial attitude parameters is proposed in reference [2]. The system constructs the kinematic model of gait, models the constraint parameters of gait control, collects the attitude parameters with position and attitude sensors such as gyroscope and accelerometer, fuses the inertial attitude parameters with extended Kalman filter and inputs them into the real-time control actuator to realize stable gait control. A neural network based attitude estimation error compensation system is proposed in reference [3]. With the help of the nonlinear mapping ability of BP neural network, the attitude error compensation model between sensor output and attitude estimation error is established, which effectively improves the detection accuracy. However, the above-mentioned posture error correction systems can all realize the function of assisting students in physical training, but there is also the problem of incorrect connection of joint points in reverse order. Multi-modal information processing technology can effectively fuse the information obtained by multiple and multiple types of sensors to form a high-level information expression form. The multi-modal information fusion technology comprehensively uses signal processing, mathematical statistics and artificial intelligence and other related theories to fuse the multi-modal information provided by sensors distributed in different positions, different types and different states of the attitude correction system, and at the same time The redundancy and contradiction that may exist between multi-modal information are eliminated, and the information obtained by each sensor is fully utilized, while achieving their respective advantages and complementing each other, and finally forming an overall, complete, and consistent of their own state and the surrounding environment The perceptual description of the system provides a basis for the selfpositioning, intelligent decision-making and other related needs of gesture recognition. Based on the above analysis, this article designs a new leg posture correction system for students in physical education class based on multi-modal information processing to help students understand the difference between their current movements and standard movements in time, and make timely repairs based on the training suggestions fed back by the system., To achieve high-quality sports effects. In this study, the signal conditioning circuit is used to filter, amplify and sample the input signal, which fundamentally improves the accuracy of subsequent error correction. Secondly, the multi-modal information processing model is used to identify the leg posture, and the recognition results are used as the input of the error probability matrix model to realize the intelligent error correction of the leg posture, so as to improve the positioning accuracy of the leg posture.

Leg Posture Correction System for Physical Education Students

93

2 System Hardware Design The leg posture correction system of physical education students must measure different postures, so the signal conditioning circuit is used to filter, amplify and sample the input signal. Because each measurement node contains four measurement channels, the single chip microcomputer can only collect data by polling. In order to meet the requirements of acquisition synchronization and accuracy, each input strain measurement signal is designed with an independent bridge circuit, and each signal is successively sent to the A/D converter port by a high-speed analog switch. The driver chip selected in this article is Sn74alvc164245 from TI. The chip can amplify the drive capacity of input and output signals, increase its drive capacity, and can drive the subsequent circuit to make it work normally. When selecting the device, pay attention to the channel switching time and signal transmission delay, and try to shorten the channel switching time [4]. During the load calibration test, the system should be zeroed before measuring the strain to ensure the accuracy of the measurement. The realization of the zero-adjusting circuit can be realized by using an operational amplifier with a zero-adjusting terminal and a D/A converter. The strain bridge is realized by an unbalanced electric bridge, and considering the problem of low power consumption, it is realized by a constant voltage bridge. In the communication serial module, the upper computer is the PC, and the lower computer is the smallest DSP system. The PC transmission data level is 1.5 V, the DSP data transmission I/O port level is TTL level, and the TTL level is “1” The characteristic voltages of and “0” are 2.4 V and 0.4 V respectively, which are suitable for data transmission within the board. There is no direct transmission between PC and DSP. In order to solve the level inequality, a serial port conversion chip is needed to convert the level. Choose MAX3232 chip here to switch between CMOS level and TTL level. The 9th and 10th pins link the receiving data bit and the sending data bit of the DSP, so the DSP transmits the data directly to the upper computer. Due to the limitation of the printed circuit board area, the strain bridge adopts a 1/4-type semiequal-arm Wheatstone bridge. Secondly, choose a three-wire external sensor to weaken the bridge arm imbalance caused by the parasitic resistance of the lead cable and reduce the measurement error in the measurement process. The schematic diagram of the strain bridge circuit is shown in Fig. 1. The communication protocol adopts the RS232 protocol, which is an asynchronous communication protocol, and adopts an unbalanced transmission method, which is the so-called single-ended communication. The check method is changed to cyclic redundancy check to reduce the data packet loss rate and protect the data from loss. In the process of system communication, the enhanced feature FIFO function is used for data transmission, which basically does not occupy the cycle waiting time of the system, and write data directly to the BUFF and send it directly. The master-slave clock configuration of the signal acquisition and transmission board can be configured through the serial communication interface or the GPIO interface. When the acquisition and transmission board receives a command to configure it as the main board, the master signal acquisition and transmission board only outputs clock signals at this time. For the slave board, on the one hand, it receives the clock signal transmitted from the upper level, and on the other hand, transmits the received clock signal to the next level.

94

L. Yang and Y. Jia R AVDD

GND

VERF

GND

VERF

VCC

Fig. 1. Structure diagram of strain bridge circuit

In the FIFO mode, the transfer of information is based on multiple frames forming a transfer unit to transfer data, rather than transmitting the data frame by frame. FIFO means first-in-first-out, that is, the first received data is read first. The 16-level depth can only save the first 16-level data. If the depth is exceeded, data overflow will occur, and the data will be lost. Use software Can deal with overflow.

3 System Software Design 3.1 Establishing a Database of Leg Postures of Students in Physical Education Class The design of physical education students’ leg posture database is an interactive way based on machine vision. In this paper, the commonly used static and dynamic leg postures in line with people’s cognitive laws and behavior habits are counted. The database includes depth map, color map and position data of leg trajectory. The image acquisition process will be affected by light intensity, external conditions and equipment functions. In the image transmission process, it will be affected by noise (common thermal noise), resulting in the deformation of the collected image effect, which will affect the accuracy of the leg posture error correction system. This article will explain the process of database collection and establishment based on Kinect. The low-cost color depth (RGB-D) camera Kinect released by Microsoft has created a new revolution in gesture recognition by providing high-quality depth images to solve complex background and lighting changes. Kinect sensor integrates a variety of advanced sensing hardware. The most worth mentioning is that it contains a depth sensor, a color camera and a microphone array, which can provide full-body 3D motion capture, facial recognition and voice recognition. Kinect can directly track the position information of the leg joint points, so this article selects the leg joint point as the tracking center based on Kinect. When processing the image, we should make corresponding preprocessing to enhance the image quality, and eliminate the ineffective information in the image to

Leg Posture Correction System for Physical Education Students

95

ensure the accuracy and precision of the image. On the whole, median filtering is a nonlinear method, which is used for image recognition and smoothing filtering processing skills. Then, all pixels in the area near the image are arranged according to the order of gray values, and the middle value is found as the pixel value to be output [5]. Image segmentation is a key step in establishing a database. The segmented image is the object of leg gesture recognition and the source of the entire interactive process. Compared with color images, depth images are based on adding depth maps. The human pose tracking problem can essentially be regarded as an extension of the human pose estimation problem in static images in the time domain, so the above-mentioned method can also be directly used to solve the pose estimation in video sequence images. In 3D computer graphics, a depth map is an image or image channel that contains information about the distance to the surface of the scene object of the viewpoint. Therefore, the database of this article uses depth images. Kinect can obtain the depth map of the user’s whole body, so it is only necessary to separate the leg area from the body as a whole. The approach taken in this article is to select the area at the same depth as the leg as the destination and save the destination directly in this article file. Because the gray value of the boundary of the region will lead to discontinuous phenomenon, the method of detecting these regions can find the edge of the region to be detected, and at the same time, it also includes the segmentation of the target image range, so as to get the depth image of the leg posture [6]. 3.2 Recognizing Leg Posture Based on Multi-modal Information Processing The basic process of multimodal information fusion is to make full and efficient use of the information obtained by multiple sensors, obtain redundant or complementary information in space or time, organically combine through the fusion algorithm, form a consistent interpretation or description of the measured object or perceived object, and then obtain better performance than a single sensor system. The multimodal model not only contains the spatial information of a single frame image, but also considers the implicit time information between adjacent images, so that it can effectively extract the spatio-temporal features in the video sequence, so as to obtain a recognition effect much higher than that of convolutional neural network [7]. In the feature layer fusion, the observation information of each sensor completes the feature extraction, obtains the feature vectors from each sensor, and then fuses these feature vectors to generate joint feature vectors. Each feature dimension in the fusion feature vector is not unified. Firstly, the feature vectors of each group are associated to ensure that the feature vectors participating in the fusion come from the same target. Considering that the leg is the main moving object in the pose sequence, it often has high optical flow amplitude. Therefore, by setting an appropriate threshold, the corresponding leg image blocks can be filtered out from the image. Recognition according to the input of different modes, including sensing and recognition of sensor information, attitude information and voice information. The fused feature can be a higher dimensional feature vector connected by each group of feature vectors, or a new type of feature vector combined by each group of feature vectors. Due to the uncertainty of the proportion of the leg region, the number of elements in the set is not fixed. The intensity of leg motion in the current image is expressed by

96

L. Yang and Y. Jia

calculating the average optical flow amplitude of the leg image block [8]. The average optical flow amplitude is calculated as follows: a=

bi n

(1)

In formula (1), a represents the average optical flow amplitude; bi represents the element in the optical flow amplitude set; i represents the sequence number of the element; n represents the total number of elements in the set. In the multi-modal information fusion interaction strategy, different nodes represent different states, and each state represents the intention of a user. The multi-modal model will sample each image sequence, and then use the obtained key frame sequence as input for subsequent processing. When segmenting the sequence, the length of the sub-sequence should be reasonably controlled to facilitate the sampling operation. The default key frame sequence length is 16 frames [9], and its advantages are good real-time performance, strong fault tolerance, and small dependence on original data. After the sequence segmentation process, each video is divided into four sequences, the sequences are independent of each other, and they are sequentially sent to the classification model for classification training. SVM is a class II linear classifier that searches for the maximum interval in the feature space, and its objective function can be expressed as:  1 2 f = 2 β   min (2) Qj β T Pj + δ ≥ 1 In formula (2), f represents the objective function; β represents the geometric distance of the classification interval; Qj and Pj represent the output and input vectors of the sample data; T represents the matrix transposition; j represents the serial number of the sample data; δ represents the hyperplane parameter. According to different operations, the system enters different states, and feeds back the expression effect of state semantics, and then conducts voice navigation and broadcast according to the system’s intentions. The characteristics of various data type sequences are fully learned from the multimodal data processing model. Each data type corresponds to a trained multimodal neural network, and then the output data of a layer of the network are fused in the test stage. Finally, the human posture and action recognition is realized based on the fused data. 3.3 Design Error Correction Model of Leg Posture In order to solve the problem of leg posture confusion caused by the wrong arrangement and connection of joint points in reverse order, a leg posture error correction model based on probability statistics is proposed according to the theorem of large numbers. The leg posture error correction model has two preconditions: the first premise is that different types of joint points have different confidence, and the second premise is that the candidate joint points predicted by the network need to be located in the human body area. Confusion matrix is usually called matching matrix in unsupervised learning. The columns of the matrix represent examples of prediction classes and the rows represent examples of actual classes, so the accuracy of the algorithm can be measured by confusing some metrics in the matrix [10].

Leg Posture Correction System for Physical Education Students

97

The repeated counting of joint points means that the joint points return to the corresponding and symmetrical joint area, resulting in two kinds of symmetrical joint points in the same joint range. The inverse arrangement of joint points means that the position of lower joint points is higher than that of upper joint points. Each row represents the predicted value and each row represents the actual category. The total number of each column represents the predicted quantity of the category, and the total number of each row represents the real quantity of the category. The values in each column represent the number of actual data predicted for this class. The misjudgment probability matrix can be expressed as: Wαγ =

zαγ 3  zαγ

(3)

γ =0

In formula (3), Wαβ represents the probability matrix of misjudgment; α represents the recognition result; γ represents the source label; zαγ represents the recognition result is α, which actually comes from the number of labels γ . Misjudgment probability model is to optimize the recognition results of leg posture model based on convolutional neural network. It should not only ensure the correctness of the original recognition results, but also correct the wrong leg posture of students, and control the new category with random numbers. When the number of similar joint points is greater than two, the solution is similar to that of only two similar joint points. For example, when there are three repeated counting problems of knee joint points, first divide the knee joint points into two groups, and there are three groups with two knee joints; Secondly, the hip joint was connected with three groups of knee joint points; Then it is processed according to the principle that there are only two joint points of the same kind. When the hip joint and knee joint are accurately positioned, the knee joint and ankle joint are treated in the same way from the correct knee joint point. The structure of leg posture error correction model is shown in Fig. 2.

Input multimodal data

Convolutional neural network

Identification category number Accuracy comparison

Misjudgment probability model

Identification category number

Fig. 2. Leg posture error correction model

98

L. Yang and Y. Jia

The multi-modal image data is recognized by the convolutional neural network model, and the recognition result is used as the input of the misrecognition probability matrix model. The error correction model is also composed of a convolutional layer and a pooling layer, and each neuron is only responsible for processing the local information of the feature map of the previous layer. Assuming that the recognition result is 01, 01 is the input parameter of the error correction algorithm. After calling the model, a new category number will be generated. The number of such recognition results is counted to calculate the recognition rate. Based on the confusion matrix of the actual class and the prediction class, the model establishes the misjudgment probability function of the prediction class and the actual class, so as to realize the intelligent error correction of the wrong leg posture. So far, the design of the leg posture error correction system for students in physical education class based on multi-modal information processing has been completed. The specific system structure is shown in Fig. 3. Hardware part Input signal

Regulate circuit

D/A converters

Strain bridge circuit

Output signal

RS232 protocol communication

Leg image segmentation

Student leg posture database

Multimodal information processing

Misread the probability matrix model

Output identification result

Software part

Fig. 3. Overall system structure drawing

4 System Test and Result Analysis In order to verify the practical application performance of PE Students’ leg posture error correction system based on multimodal information processing, the following experiments are designed. 4.1 Experiment Preparation The hardware environment of this system test is as follows: CPU: i7-8550U;

Leg Posture Correction System for Physical Education Students

99

GPU: MX150, memory 8G, video memory 4G; Experimental external equipment: Kinect 2.0, smart equipment; Input data: student’s leg posture sensor signal. The system software platform is mainly built in the Visual Studio 2015 environment. In the development process, the open source computer vision library OpenCV for deep neural networks, Microsoft’s MFC interface library, CUDA architecture and GPU acceleration library cuDNN were used. The system uses C++ for development and processes the files, data types, text streams and other objects involved in the program. 4.2 System Performance Test Results and Analysis In order to verify the application effect of the proposed system, it is compared with the attitude correction system based on inertial attitude parameter quantization fusion (traditional system 1) and the attitude correction system based on neural network compensation algorithm (traditional system 2). The evaluation index selected in this experiment is the percentage of correctly positioned parts. For each joint point, when the distance between the predicted position and the true position is less than a given threshold, the joint point is correctly positioned. When the two joint points corresponding to the two ends of the human body part are correctly positioned, the part is correctly positioned. Under the condition that the number of test users is 10, 100 and 500, the positioning accuracy of the leg parts of each posture error correction system is tested respectively. The results are shown in Tables 1, 2 and 3. Table 1. Test results with 10 users Number of experiments

Accuracy of leg posture positioning/% System of this paper

Traditional system 1

Traditional system 2

1

91.44

85.74

81.04

2

92.88

84.87

82.45

3

90.66

86.95

84.58

4

89.55

85.66

85.66

5

88.27

84.52

83.22

6

91.04

87.83

82.05

7

90.10

85.41

83.13

8

92.55

86.85

81.52

9

93.22

84.53

80.81

10

92.94

84.02

82.90

Under the test condition of 10 users, the maximum positioning accuracy of leg parts in this system is 93.22%, which is 5.39% and 7.56% higher than that of Traditional System 1 and Traditional System 2, respectively. Therefore, when the number of users

100

L. Yang and Y. Jia

is 10, compared with the two traditional systems, system of this paper can locate the leg posture more accurately. Table 2. Test results with 100 users Number of experiments

Accuracy of leg posture positioning/% System of this paper

Traditional system 1

Traditional system 2

1

88.41

81.49

78.04

2

87.84

82.56

79.48

3

85.55

83.67

80.58

4

86.68

81.88

79.67

5

85.26

80.26

78.24

6

84.51

81.35

77.58

7

85.82

80.54

78.85

8

87.66

80.01

79.66

9

88.53

82.48

80.22

10

88.20

82.82

78.34

Under the test condition of 100 users, the maximum positioning accuracy of leg parts in this system is 88.53%, which is 4.86% and 7.95% higher than that of Traditional System 1 and Traditional System 2, respectively. Therefore, when the number of users is 100, compared with the two traditional systems, system of this paper can effectively improve the positioning accuracy of leg posture. When the number of users is 500, the accuracy of leg posture positioning of the system is up to 85.56%, that of traditional system 1 is 78.15%, and that of traditional system 2 is 74.64%. In contrast, the system in this paper can obtain the leg posture localization results more accurately. Based on the above results, it can be seen that the leg posture correction system designed in this paper has obvious advantages in the positioning accuracy of the key points, and can better achieve high-precision human posture estimation and error correction tasks.

Leg Posture Correction System for Physical Education Students

101

Table 3. Test results with 500 users Number of experiments

Accuracy of leg posture positioning/% System of this paper

Traditional system 1

Traditional system 2

1

85.44

77.44

72.62

2

84.71

76.27

74.64

3

83.50

75.58

73.81

4

82.63

74.65

71.56

5

84.25

75.36

72.32

6

85.56

76.02

72.25

7

83.82

78.15

73.52

8

83.95

77.24

72.10

9

84.67

76.52

71.36

10

83.21

75.81

72.08

5 Conclusion Human pose estimation in natural images has always been a hot topic in the field of computer vision. It has a good application prospect in many fields, such as behavior recognition, human-computer interaction, motion analysis and so on. This paper designs a leg posture error correction system for physical education students from the perspective of multimodal information processing. The system can accurately locate the leg joints and parts, and improve the accuracy of pose recognition. The research of this paper mainly aims at the problem of single person pose estimation in natural images, but in fact, it is more common to have multiple human objects in natural images at the same time. Therefore, further research on multi person human pose in natural images is more in line with the needs of the market.

References 1. Yang, L.: Monitoring method of leg posture of sprinters based on inertial sensor. J. Hebei North Univ. (Nat. Sci. Ed.) 36(5), 42–47 (2020) 2. Lu, X., Huang, W., Huang, P.: Bionic gait control of robot leg based on quantization fusion of inertial attitude parameters. Transducer Microsyst. Technol. 38(3), 43–46 (2019) 3. Sun, Y., He, S., Xu, X., et al.: Application of neural network compensation algorithm in MEMS-based attitude measurement. Appl. Res. Comput. 36(9), 2696–2699 (2019) 4. Wang, L., Zhang, Z., Niu, Q., et al.: Design of human body posture detection system based on MPU9250 and MS5611. Chin. J. Electron Dev. 42(4), 978–983 (2019) 5. Liu, S., Liu, D., Muhammad, K., Ding, W.: Effective template update mechanism in visual tracking with background clutter. Neurocomputing 458, 615–625 (2021) 6. Liu, Y., Hui, H., Lu, Y., et al.: Remote human body posture monitoring system based on smart phone terminal. J. Chin. Inert. Technol. 27(6), 713–718 (2019)

102

L. Yang and Y. Jia

7. Liu, S., Wang, S., Liu, X., et al.: Fuzzy detection aided real-time and robust visual tracking under complex environments. IEEE Trans. Fuzzy Syst. 29(1), 90–102 (2021) 8. Zhu, H., Yin, J., Feng, W., et al.: Research and application of a lightweight real-time human posture detection model. J. Syst. Simul. 32(11), 2155–2165 (2020) 9. Liu, S., et al.: Human memory update strategy: a multi-layer template update mechanism for remote visual monitoring. IEEE Trans. Multimedia 23, 2188–2198 (2021) 10. Lu, J.: Attitude recognition simulation of data glove based on feature point set matching strategy. Comput. Simul. 38(4), 403–407 (2021)

Online Interactive Teaching System of Sanda Course Based on P2P Network Yueguo Jia1(B) and Lei Yang2 1 Tianjin Public Security Professional College, Tianjin 300382, China

[email protected] 2 Xi’an Medical College, Xi’an 710021, China

Abstract. In order to improve the online teaching effect of sanda, this study designed an online interactive teaching system of sanda based on P2P network. The microcontroller is taken as the core of the hardware part of the system, and the processor, CPU sub-board and compatible motherboard are designed. After debugging, the normal transceiver of the hardware environment is realized. Then, on the basis of designing system function modules and database structure based on P2P network, the index function of the system is designed to complete the construction of system software environment. Experimental results show that the system improves the interactive efficiency, teaching performance and content collection accuracy, and effectively improves the teaching effect. Keywords: P2P network · Sanda course · Online interactive teaching · Controller

1 Introduction Network teaching has become an important way of education and teaching reform in Colleges and universities, and has been applied in various courses of education. Network teaching system is a virtual education and teaching environment based on P2P network on the Internet or campus network. Through this platform, the functions of teaching, communication, resource sharing and autonomous learning can be realized. Most of the current network teaching assistance systems usually use B/S or C/S access mode. Among them, education and teaching resources are usually concentrated on the server side of the system, and each user’s PC side cannot realize the exchange of learning resources. Although this method can meet the needs of the classroom, there are still problems in the following aspects: the first network, the single point of failure problem. Since the learning resources are provided on the server side, if there is a problem on the server side of the P2P network, the system cannot run; secondly, the cost is high. In order to better meet the needs of the system, it is necessary to construct various teaching services, and at the same time, with the increase of clients, higher requirements are placed on the server and data; third, the scalability is poor. When the number of users increases, the number of expensive servers needs to be increased to provide more stable services. © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 103–117, 2022. https://doi.org/10.1007/978-3-031-21161-4_9

104

Y. Jia and L. Yang

P2P technology, as a kind of application technology of P2P network, was proposed in the 1990s of the last century. It mainly uses the nodes in the entire network to perform peer-to-peer computing, so that the space resources existing in the network can be fully mined. It has great advantages in fault tolerance, utilization and scalability. Therefore, in view of the above-mentioned problems, this paper proposes an online interactive teaching system for Sanda courses based on P2P network.

2 System Hardware Design The hardware structure of the system is shown in Fig. 1. Compatible motherboard

SD card

Memory expansion

Microprocessor

OPC interface

CPU daughter board

Button

Graphics card

Fig. 1. System hardware structure

In this study, the microprocessor is taken as the core of the system hardware, as shown below. 2.1 Processor Design The three embedded processors selected by this system are all ARM9 series processors. S3C2440 is a 32-bit RISC embedded processor based on ARM920T core produced by SAMSUNG Company [1], mainly for handheld devices and applications with high cost performance and low power consumption. The ARM920T core is composed of three parts: ARM9TDMI, storage pipeline unit MMU and cache. Among them, the MMU can manage virtual memory, and the cache consists of an independent 16 KB address and 16 KB data high-speed cache. The ARM920T has two internal coprocessors: CP14 and CP15. Among them, CP14 is used for debugging control, and CP15 is used for storage system control and test

Online Interactive Teaching System of Sanda Course

105

control. In order to reduce the system cost, the following components are integrated inside the S3C2440 chip: 1 LCD controller (independent DMA channel, supports STN and TFT) [2]; External memory controller (SDRAM controller and external chip controller); 4-channel UART; 4-channel DMA controller; 2-channel SPI; IIC bus interface; IIS audio interface and AC97 audio interface; SD/MMC interface; 2 USB host interfaces and 1 USB device interface; 8-channel 10-bit ADC; Touch screen interface; 4 timers with PWM function and an internal clock PLL; Watchdog counter; Camera interface. On the clock side, the S3C2440 integrates an RTC with calendar function and a clock generator with PLL (MPLL and UPLL). The MPLL generates the master clock, which enables the processor to operate at a frequency of up to 400 MHz. This operating frequency enables the processor to easily run WinCE, Linux and other operating systems and perform more complex information processing [3]. In addition, the S3C2440 adopts an independent power supply method for each component on the chip: (1) Core: When the main frequency is set to 300 MHz or below, use 1.2 V power supply; (2) When the main frequency is set to 400 MHz, use 1.3 V to supply power. The storage unit adopts 3.3 V independent power supply, 3.3 V can be used for general SDRAM, and 1.8/2.5 V can be used for mobile SDRAM. The memory controller of S3C2440 is used to provide control signals to the external memory. The memory controller has the following characteristics: Support big-endian/little-endian modes (selected by software); Address space: 1 GB in total, divided into 8 banks, each bank is 128 MB, S3C2440 uses 8 general chip select signals to select these banks; The start address and size of Bank0 to Bank5 are fixed for ROM or SRAM; Bank6 and Bank7 are used for ROM, SRAM or SDRAM, the start address and size of these two groups are programmable; The access cycle of all Banks is programmable; Supports self-refresh and low-power modes of SDRAM. In addition, the S3C2440 adopts an independent power supply method for each component on the chip: (1) Core: When the main frequency is set to 300 MHz or below, use 1.2 V power supply; when the main frequency is set to 400 MHz, use 1.3 V power supply. (2) The storage unit adopts 3.3 V independent power supply, 3.3 V can be used for general SDRAM, and 1.8/2.5 V can be used for mobile SDRAM [4, 5]. 2.2 CPU Daughter Board Design The CPU daughter board mainly includes the following modules: ARM chip, power supply and crystal oscillator module, SDRAM module, NANDFlash module, daughter motherboard interface module.

106

Y. Jia and L. Yang

The power supply and crystal oscillator module, on the daughter board, need to have two different voltages for the chip and each peripheral to use. 3V3 voltage, that is, 3.3 V voltage, the memory module of the system needs 3V3 voltage; 1V2 voltage, that is, 1.2 V voltage, the voltage used by the three ARM chips. Because there is only one input voltage from the motherboard to the daughter board, that is, the 3.3 V power supply voltage, a level conversion circuit should be designed in the hardware design for 3V3 to 1V2. The crystal oscillator circuit is an indispensable part of the embedded hardware circuit. The ARM chip obtains the main clock by multiplying the frequency of the external crystal oscillator. According to the requirements of the ARM chip S3C2440, this design provides two crystal oscillators with different frequencies, 16.9344 MHz and 32.768 MHz, for the ARM chip, which are respectively used in normal working mode and sleep mode [6]. SDRAM module, the memory controller of S3C2440 is used to provide control signals to the external memory. The memory controller has the following characteristics: Support big-endian/little-endian modes (selected by software); Address space: 1 GB in total, divided into 8 banks, each bank is 128 MB, S3C2440 uses nGCS [0–7]8 general chip select signals to select these banks; The start address and size of Bank0 to Bank5 are fixed for ROM or SRAM; Bank6 and Bank7 are used for ROM, SRAM or SDRAM, the start address and size of these two groups are programmable; The access cycle of all Banks is programmable; Supports self-refresh and low-power modes of SDRAM. 2.3 Compatible Motherboard Design The compatible motherboard mainly includes: power supply and reset module and peripheral interface module. The main principle of compatible motherboard design is to ensure compatibility with the three CPU daughter boards [7]. Next, we mainly introduce the core modules on the compatible motherboard. The structure block diagram of compatible motherboard is shown in Fig. 2. On the mother board of the power supply and reset module, two different sizes of voltage are set for the chip and each peripheral. The input voltage from the outside to the embedded platform is 5 V, and the voltage from the USB interface and LCD interface is 5 V. The voltage of the peripheral circuit is 3.3 V. Because the external input voltage is fixed, a level conversion circuit should also be designed in the hardware design. The reset circuit is an essential part of the embedded hardware circuit. The reset circuit in this design can be used for power-on reset or reset by manual button. At the same time, because the network interface chip CS8900A requires high-level reset, a 74HC14 inverter is added to invert the negative pulse of the system reset signal to obtain a positive pulse reset signal. Ethernet interface module. The three ARM chips do not have an integrated Ethernet controller, so in this design, an Ethernet control chip CS8900 is externally connected to the control signal provided by the memory controller. Among them, the chip select signal of CS8900 selects NET_CS, that is, CS8900 The internal address space is mapped

Online Interactive Teaching System of Sanda Course

External expansion bus interface

SD card interface

Motherboard interface

Digital Tube

107

Daughter board

Keyboard

Infrared interface

Audio port

Network Interface

Fig. 2. Compatible motherboard structure block diagram

into the memory space pointed to by NET_CS. The choice of NET_CS is determined by the specific CPU daughter board [8]. LCD and touch screen interface circuit. The LCD selected in this design is the product LTM035A776C of TOSHIBA Company. LTM035A776C is a TFT LCD with touch function. For the S3C2440 daughter board, the touch screen is controlled directly through the dedicated interface of the touch screen provided by the chip; for the two daughter boards AT91SAM9263 and MC9328MX21, the touch screen is controlled by connecting an external AD chip ADS7843 through the SPI interface. Among them, the jumper is used to select which method is currently used to control the touch screen [9]. Audio interface module. The audio interface of the system is realized through the AC97 interface provided by the ARM chip and the external audio codec AD1981. The AC97 controller provides two serial data lines, clock lines and control lines to the external AD1981. AD1981 has three external interfaces, namely: headphone interface, microphone interface, and audio conversion interface. JTAG interface module. All three ARM chips integrate a JTAG controller, providing data lines and control lines that conform to the JTAG standard, so the JTAG circuit of the system is relatively simple. Keyboard and digital tube circuit, keyboard and digital tube circuit are the key circuits on the teaching experiment platform. Most of the experimental courses include the experimental operation of keyboard and digital tube. There are no keyboard and digital tube controllers in the three ARM chips. This design uses a chip ZLG7290 designed by Zhou Ligong Company as the keyboard and digital tube controllers. ZLG7290 can directly drive 8-bit common-cathode nixie tubes (or 64 independent LEDs), and can also scan and manage up to 64 keys. At the same time, ZLG7290 is connected to the microcontroller by I2C bus, and the interface only needs two signal lines. Considering the board area, this design only expands 16 buttons and 8-bit digital

108

Y. Jia and L. Yang

tubes. The 16 buttons cut out are managed by the signals DIG0-3 and SEGA-D, and the button values are 1–4, 9–12, 17–20, 25–28 respectively.

3 System Hardware Debugging This section mainly introduces the system debugging method and part of the debugging process. When the printed circuit board is completed, you should not rush to solder the components, but first carefully check the connection of the printed circuit board against the schematic diagram, and check whether the power supply and the ground are shortcircuited with a multimeter to ensure that there is no error before welding [10]. After the system is powered on, you should first check whether the circuit works abnormally. It is normal for the chip to have a certain amount of heat during operation. However, if any chip is particularly hot, there must be a fault, and it needs to be powered off and checked to confirm that it is correct before continuing. Power-on debugging. Debugging tools require oscilloscope, multimeter, electric soldering iron, etc., and ARM debugging development software and corresponding emulators. In the process of hardware debugging of this system, the development and debugging software RVDS provided by ARM company and its JTAG emulation software H-JTAG are used. The power supply circuit, crystal oscillator circuit and reset circuit are the basis for the normal operation of the whole system, and these three modules should be tested first [11]. This section mainly takes the S3C2440 platform as an example to introduce the hardware debugging method. When debugging the power supply circuit, the circuit board after soldering is likely to be short-circuited and soldered [12]. If you directly connect the power supply at this time, it is likely to cause the main chip to heat up and burn. This situation has been encountered during debugging. After soldering the circuit, directly connect to the power supply, and the S3C2440 will start to heat up until it is hot. The correct debugging method is: before the system is powered on, first unplug the jumper near the power circuit on the motherboard to prevent the power circuit from supplying power voltage to the motherboard, and then use a multimeter to test whether the output voltage of the motherboard power conversion circuit is 3.3 V, the jumper can be plugged in under normal circumstances to debug other modules [13], and the test method of the power circuit of the daughter board is the same. Reset circuit debugging, after the system is powered on, check the working conditions of each interface module, and find that the LED light representing the normal operation of the network card chip CS8900 is not lit normally. After checking the datasheet of CS8900, it is found that CS8900 needs positive pulse to reset, so the positioning problem occurs in the reset circuit. Add a 74HC14 inverter to the reset circuit. In this way, the negative pulse of the system reset signal can be inverted to obtain a positive pulse reset signal and provided to the CS8900. After power-on reset, CS8900 can work normally. Crystal oscillator circuit debugging, the S3C2440 platform uses 16.9344 MHz and 32.768 kHz crystal oscillators. After power-on, use the oscilloscope to capture the waveforms of the above two crystal oscillators, and observe whether the output waveform parameters of the crystal oscillator are correct and whether the waveform is stable. During the debugging process, it was found that the crystal oscillator circuit of some daughter boards did not start to vibrate. This problem can be solved from two aspects:

Online Interactive Teaching System of Sanda Course

109

➀ The size of the crystal capacitor; ➁ Whether the crystal is broken. The following experience was obtained during the debugging process. Based on the stable operation of the crystal oscillator circuit, it is best to use an active crystal oscillator for the crystal oscillator circuit. JTAG module debugging, on the premise that the basic circuits such as power supply circuit, crystal oscillator circuit and reset circuit work normally, the JTAG interface should be tested first, because as the dedicated DEBUG interface of the ARM chip, the development and debugging software RVDS is downloaded to the ARM chip through the JTAG interface Debug programs, which in turn debug other peripherals. Connect the development board to the PC through the JTAG interface, and run the H-JTAG software on the PC. For peripheral circuit debugging, after the JTAG interface test is completed, other interfaces of the hardware platform can be debugged through the RVDS debugging tool [14]. RealView® Development Suite (RVDS) is a new-generation development tool mainly promoted by ARM after SDT and ADS1.2. Through RVDS, we can easily view and modify the internal registers of the ARM chip and the values in the memory, and we can also write test programs for each peripheral device to debug through RVDS. When testing other peripheral interfaces, first test the power supply and ground corresponding to the peripheral interface, then test the connectivity of each signal line, and finally download the test program corresponding to the peripheral to the ARM chip through the JTAG interface, and pass the RVDS interface. Debug to test that the peripheral is functioning properly. The following takes serial port circuit debugging as an example to introduce the method of debugging through RVDS. Using the software RVDS, download the program compilation link to the development board to run through the JTAG interface, and found that there are characters displayed on the hyperterminal. The first few characters are correctly displayed, but the characters displayed after 4 or 5 are not, characters entered. Use an oscilloscope to measure the data on the data line of the UART on the development board. The data sent is sent according to the RS232 protocol. The waveform display is correct, but the display on the HyperTerminal is wrong. It means that the data sent by the development board is correct. The sampling frequency of the PC when receiving is inconsistent with the sending frequency of the development board, which causes this error. After debugging, it was found that the root cause was the inaccurate setting of the main clock frequency of the development board, resulting in the baud rate factor not being the expected 115200. To accurately set the main clock, it is necessary to read the exact frequency of the main crystal oscillator. After modification, the frequency of the main crystal oscillator is 16.365 MHz, and the main clock is configured as 88.192 MHz. After the baud rate is set correctly, run the program, and the result is normal transmission and reception.

110

Y. Jia and L. Yang

4 System Software Design 4.1 System Function Module Design Based on P2P Network The server and client of the network teaching interactive system based on technology adopt the client server model in management. The teaching resources are classified, sorted and distributed on the central platform and transmitted to end users through the Internet. The central platform provides users with a variety of service functions and integrates various teaching resources. All service functions in the network teaching system are provided by the service center, including online live classroom, courseware download and on-demand, online communication, online examination, etc. The server center platform is the core part of the network teaching system. The backup center dynamically backs up the key data of the service center. In case of data loss of the center platform, the backup center can quickly restore the service capacity of the center platform. All remote clients need to pass the server authentication and work together under the unified management of the server. Therefore, the server and client adopt the client server model in management. However, in order to prevent all remote clients from obtaining streaming media data from the data server, resulting in the emergence of system bottleneck problems, the point-to-point model is used in streaming media data transmission. Remote clients can obtain streaming media data not only from the data server, but also from other clients. This design can obviously reduce the data load of the server and improve the speed of users obtaining streaming media data. On this basis, using the technology combined with streaming media, it is divided into two parts: client and server. The client function provides users with basic “system functions”, such as user login, user registration, system help and online update. At the same time, it can realize online live classroom, courseware download and on-demand, online homework, online examination, exercise practice, online communication and other functions. The server side includes system management and course management. First, the online live classroom applies the technology to the streaming media field and breaks the traditional client server mode. Clients can not only obtain data directly from the server, but also connect with each other to obtain streaming media data. Therefore, only some clients in the system need to obtain streaming media data from the server for playback, and the other clients can play through the connection established with each other. In this way, the service can be decentralized, so as to reduce the server load and support a wider range of streaming media release, support large-scale students to listen to classes online, and the investment cost is very low. At the same time, students can put forward questions to teachers online, teachers answer questions on site, correct errors in time and improve the quality of teaching. Second, courseware download and on-demand. With the advantages of network information sharing, the courseware downloaded by nodes from the server can be shared with other nodes. With the expansion of the network scale, the more nodes, the greater the probability of finding the required courseware, so that the network scale can be expanded infinitely without server bottleneck. Third, for online homework, the teacher provides some questions or specifies some contents to be placed in the “homework” folder of the local machine, which is set

Online Interactive Teaching System of Sanda Course

111

to be shared when necessary, and learners can complete it on the local machine by downloading. After completion, use technology to send e-mail to teachers to verify students’ learning. This is an effective supplement to “online learning” and “discussion”, because the discussion is more about staying in the mind, fragmented and broken, not a whole. Therefore, it is necessary to take some discussion questions, of course, not limited to the discussion questions as homework, which should be actually done by students, corrected and counted. Fourth, online communication, students find teachers to answer questions and solve doubts outside classroom time, or discuss and exchange problems among students. After students log in, they no longer need the support of the server. In this way, after the introduction of technology, compared with the restrictions of the server status on the communication between teachers and students through the chat room and other tools of the server under the mode, the communication of users under the system will not be limited, and will be more perfect. Fifth, online examination, there are many methods of learning evaluation in the process of network teaching, among which online examination is an effective main means to test learners’ learning effect. In network teaching, online examination ensures the authenticity of examination results to the greatest extent by effectively controlling the examination time. Its practice is similar to online homework, but it requires the time for learners to submit their answers. Sixth, exercise management. When students log in and study, they can choose their own exercises. The exercise library contains the exercises of all online courses, so that students can choose and practice the learning content to the corresponding exercises after learning some courses. The content in the exercise library can be added, modified and deleted according to the change of course content. The research system of this subject takes the basic computer course as the experimental object, so the exercises include two parts, theoretical problems and practical problems. There are multiple-choice questions and fill in the blanks. The actual operation questions are directly taken out from the database, and the questions are provided by the computer test management interface. The management of exercises is limited to teachers who have the right to modify such courses. Seventh, the specific process of user-on-demand process design based on P2P hybrid technology is as follows: 1) The user logs in to the P2P server; 2) Publish the content of P2P video-on-demand to the service webpage, and after finding the relevant ideological and political course resources required by users, such as Marx’s philosophy, Marx’s economics and other video resources, the P2P streaming media client program will perform a unique search on the video resources. The identification of the P2P network to distinguish other videos. 3) The on-demand system connects to the resource management server, finds the resource list of the content node through the identifier of the P2P network only, and schedules the content node through the attributes of the content node (such as client ID number, IP address, etc.). 4) After receiving the connection between the content server and the responding node, the video resource is stored by means of buffering through the Socket connection.

112

Y. Jia and L. Yang

4.2 Database Structure Design The E-R diagram, also known as the entity-relationship diagram, mainly represents the relationship between entities and attributes. Through the E-R diagram, developers can intuitively observe the attributes of each entity. The establishment of E-R diagram helps developers to design the content in the database more simply and clearly, which greatly improves the development efficiency of developers. Combined with the logical design of the database and the E-R diagram, the design of the physical structure of the database is realized, and the geographic name information table Ginfo is mainly designed to store the related information of different geographic names. The geographic name information table Ginfo is shown in Table 1. Table 1. Geographic name information table Ginfo Field name

Type of data

Remark

ID

Int

Database sequence is automatically generated

Pinyin

Varchar

Pinyin of geographical names

Text information

Mediumtext

Introduction to the relevant text of geographical names

Image storage path

Varchar

Geographic name image information storage path

Video storage path

Varchar

Geographic name video information storage path

Coding

Varchar

Encoding of geographic names

In the database design stage, the design of the logical structure and physical structure of the database has been completed, and the specific design and development work of the database program is realized in this stage. The system established a MySQL database intelsp, and established the geographic name information table Ginfo in the database intelsp. Next, the database development process is described in detail. First install MySQL5.5.28 and Navicat Premium software. Navicat Premium is a visual database management tool. After the installation is successful, first establish a connection with the MySQL database. After the connection is successful, you can create a database intelsp in the visual database management tool Navicat Premium, and then create a geographic name in the database. In the information data table Ginfo, enter all the geographic information data on the sand table in the Ginfo table. After the input is successful, you can see the content of the geographic information table Ginfo on the software interface. The main task of the background server is to wait for the response to the client request. After receiving the client request, it first performs data analysis to determine whether it needs to be retrieved. The data that needs to be retrieved is subjected to a Lucene-based database full-text retrieval. The retrieval results are as follows: On the other hand, it is sent to the bottom embedded sand table controller through WiFi, Bluetooth or serial communication, and the microprocessor on the controller parses the data received by the serial port, so as to control the specific LEDs, light cards and light strips on the sand table; The server-side application program directly plays the multimedia information related

Online Interactive Teaching System of Sanda Course

113

to the retrieval result according to the parsed instruction. The multimedia information includes text, pictures, and videos. Considering that there may be multiple images for a location, the playback rule for images is one every 2 s. 4.3 System Index Function Design In order to deal with the massive online interactive teaching resource set and interactive search problem, this study uses domain index method to reformulate online interactive teaching resource set by borrowing domain. Definition 1. Domain: The combination of attributes of a data set is called a domain. Use D to describe the online interactive teaching resources. If  represents the complete set of field values, then: D = {x|x ∈  }

(1)

The resource set S reformulated by the domain is described by the following formula: S = {D1 , D2 , · · · , Dn }

(2)

Definition 2. Domain correlation degree: For domain Q, I , the correlation degree of domain can be expressed as: t(Q, I ) = |Q ∩ I |/|Q|

(3)

where |·| stands for the basis of the set. Generally, Q represents the query domain, I represents the index domain, and the correlation degree of the domain is t(Q, I ) ∈ [0, 1]. The larger its value, the better the link between the domains. Definition 3. Domain search: according to the proposed threshold value t ∗ ∈ [0, 1] of domain Q, domain set I and correlation degree, the process of searching the correlation degree beyond t ∗ from domain set I is described as domain search, and its formalization can be expressed as: X : t(Q, X ) ≥ t ∗ , X ∈ I

(4)

Definition 4. Domain index: An index consisting of hash values of domain I is a domain index. Generally speaking, the signature vector will be divided into b-type partitions, and there will be r rows in all partitions, so the correlation between the probability of becoming a pre-selected domain and the Haccard phase velocity s can be described as follows: b  (5) P(s|b, r ) = 1 − 1 − sr As for the asymmetry of pre-correlation degree, this paper uses the mutual conversion between domain correlation degree and Jaccard similarity degree s to solve the problem, and the correlation is shown in Formula (6) and (7). sˆx,q (t) = t/(x/q + 1 − t)

(6)

114

Y. Jia and L. Yang

ˆtx,q (s) = (x/q + 1)s/(s + 1)

(7)

where, x = |X |, q = |Q| and X , Q ⊂ D represent the bases of fields X and Q respectively. To sum up, on the basis of designing system function modules and database structure based on P2P network, the index function of the system is designed and the system software environment is completed.

5 Experimental Comparative Analysis To verify the effectiveness of the online interactive teaching system of Sanda Course Based on P2P network, the following experiments are designed. In order to avoid the singleness of the experimental results, the traditional teaching system is compared with the system in this paper. 5.1 Comparison of Interactive Response Time The response time comparison results between the system in this paper and the traditional system are shown in Fig. 3.

Interactive response time/min

5 Historical system

4 3 Research system

2 1 1

2 3 4 Interactive project/piece

5

Fig. 3. Interactive response time comparison

According to Fig. 3, compared with the historical system, the response time of the system in this paper is less in each operation, and the interactive response time is always below 2 min, indicating that the system has a strong timeliness.

Online Interactive Teaching System of Sanda Course

115

Teaching score/point

100 Research system

90 80

Historical system

70 60 1

2 3 4 Number of experiments/times

5

Fig. 4. Comparison of teaching achievements

5.2 Comparison of Teaching Performance After applying the two systems, the comparison results of students’ scores in the school are shown in Fig. 4. Through the analysis of the results shown in Fig. 4, it can be seen that after the application of the traditional system, the teaching score fluctuates between 80 and 85 points. However, after the application of this system, the student’s score is higher, and always keeps above 97 points. It shows that the system can obviously improve the teaching effect. This is because the system in this paper improves the efficiency of interaction in the online teaching process, and the information obtained is more accurate, thus improving the teaching effect. 5.3 Content Search Accuracy Comparison Two systems are used to find the content in the system respectively, and the accuracy comparison results of content search are shown in Fig. 5. By analyzing the results shown in Fig. 5, it can be found that compared with the traditional teaching system, the system in this paper has a higher accuracy in searching for teaching resources and interactive teaching content, and its accuracy rate is always above 95%, thus fully proving the reliability of the system in this paper. This is because the system in this paper solves the traditional mode of server is easy to become the system bottleneck problem, is the system has high flexibility and scalability.

116

Y. Jia and L. Yang

Content search accuracy/%

100

90

Research system

80 70

Historical system

60 1 2 3 4 Number of experiments/. times

5

Fig. 5. Content search accuracy comparison

6 Conclusion The innovation of this paper is the application of P2P technology in the construction of network teaching system, which can well solve the problems of network teaching resource sharing and user interaction, improve the interaction, real-time and personalized, reduce the burden of the server. At the same time, this paper also studies and analyzes the technical theories related to streaming media, and designs and implements a low-cost powerful network teaching interactive system. The research work of this paper is summarized as follows: (1) This paper studies the principle of P2P network related technology, analyzes the advantages and disadvantages of the technology in detail, and expounds the P2P network structure and application field of the technology; (2) The network structure and logical structure of the system are constructed, which not only solves the problem that the server is easy to become the system bottleneck in the traditional mode, but also has strong flexibility, fault tolerance and expansibility. The verification shows that after the system is put into practical application, the interactive response time of the system is shorter, the content search accuracy is higher, and the teaching performance is significantly improved, indicating that the system has a high application advantage.

References 1. Si, Q., Ma, Z., Liu, F., et al.: Performance analysis of P2P network with dynamic changes of servers based on M/M/C queuing model. Wirel. Netw. 27(5), 3287–3297 (2021) 2. Zhang, C., Cho, H.H., Chen, C.Y., et al.: Fuzzy-based 3D stream traffic lightweighting over mobile P2P network. IEEE Syst. J. 14(2), 1840–1851 (2020) 3. Har, L., Xia, Z., Hsu, C.: Non-interactive secure multi-party arithmetic computations with confidentiality for P2P networks. Peer-to-Peer Netw. Appl. 14(2), 722–728 (2021)

Online Interactive Teaching System of Sanda Course

117

4. Rahmadika, S., Noh, S., Lee, K., et al.: The dilemma of parameterizing propagation time in blockchain P2P network. J. Inf. Process. Syst. 16(3), 699–717 (2020) 5. Meng, X., Zhang, G.: TrueTrust: a feedback-based trust management model without filtering feedbacks in P2P networks. Peer-to-Peer Netw. Appl. 13(1), 175–189 (2019). https://doi.org/ 10.1007/s12083-019-00742-2 6. Lim, J., Bok, K., Yoo, J.: An efficient continuous range query processing scheme in mobile P2P networks. J. Supercomput. 76(2), 1–15 (2020) 7. Zhou, S., Meng, X.: A location and time-aware resource searching scheme in mobile P2P ad hoc networks. J. Supercomput. 76(9), 6809–6833 (2020) 8. Han, D., Cho, M.: Selective switching dual-transmission scheme in multi-led hybrid VLC-P2P networking system. Peer-to-Peer Netw. Appl. 13(6), 1–12 (2020) 9. Liu, S., Liu, D., Muhammad, K., Ding, W.: Effective template update mechanism in visual tracking with background clutter. Neurocomputing 458, 615–625 (2021) 10. Hong, S.: P2P networking based Internet of Things (IoT) sensor node authentication by blockchain. Peer-to-Peer Netw. Appl. 13(2), 579–589 (2020) 11. Gao, P., Li, J., Liu, S.: An introduction to key technology in artificial intelligence and big data driven e-Learning and e-Education. Mob. Netw. Appl. 26(5), 2123–2126 (2021). https://doi. org/10.1007/s11036-021-01777-7 12. Ceptureanu, E.G., Ceptureanu, S.I., Herteliu, C., et al.: Sustainable consumption behaviours in P2P accommodation platforms: an exploratory study. Soft. Comput. 24(18), 13863–13870 (2020) 13. Liu, S., et al.: Human memory update strategy: a multi-layer template update mechanism for remote visual monitoring. IEEE Trans. Multimedia 23, 2188–2198 (2021) 14. Xu, N., Fan, W.: Research on interactive augmented reality teaching system for numerical optimization teaching. Comput. Simul. 37(11), 203–206+298 (2020)

Intelligent Computer Aided Instruction System Based on Cloud Computing Shiliang Liu(B) and Caifeng Gao Henan Medical College, Zhengzhou 451191, China [email protected]

Abstract. The traditional teaching assistant system can only count the teaching data. The teaching assistant function is single, and it can not support the large concurrent demand. To solve the above problems, an intelligent CAI System Based on cloud computing is designed. Taking embedded microprocessor as the core design, 5G communication and other functional modules are added as the system hardware. In order to meet the high concurrent operation requirements of the system, the cloud computing module is arranged to solve the problem of single teaching auxiliary function. By establishing students’ knowledge map, Bayesian inference network is used to recommend learning resources. According to the auxiliary teaching function, the system database is designed to meet the needs of supporting a large amount of concurrency. Through the comprehensive performance test of the computer-aided instruction system, it is verified that the response rate of the designed system is increased by nearly half, and it has the corresponding function to meet the modern teaching assistance. The practical application of the computer-aided instruction system has a good effect. Keywords: Cloud computing · Intelligent computer system · Auxiliary teaching system · Microprocessor · Knowledge atlas · Bayesian inference network

1 Introduction With the increasingly significant role of computer in teaching, the development and application of traditional multimedia CAI system are confined by its own limitations and closure, resulting in less application of CAI System in practical teaching. Computer assisted instruction is to simulate the teacher’s behavior with computer, and achieve the teaching purpose through the interaction between students and computer [1]. At present, the computer-aided teaching system commonly used can not update the teaching content at any time, students can not feed back the difficulties encountered in learning to teachers when using the teaching assistant system, and the traditional computer-aided teaching system ignores the gap between different students, resulting in unsatisfactory teaching effect. Therefore, it is necessary to improve and optimize the CAI system. Reference [2] studies the teaching and training platform of marine supply virtual equipment under Web3D technology by using web technology, and streams and downloads the compressed © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 118–128, 2022. https://doi.org/10.1007/978-3-031-21161-4_10

Intelligent Computer Aided Instruction System

119

data through HTTP protocol. At the same time, the software seven layer switching strategy is used to deal with the high concurrent access on the network. This teaching platform has good practical teaching effect, but it has high requirement for hardware support. Reference [3] uses augmented reality technology and uses interactive numerical method to optimize it, and designs augmented reality teaching system. The system can build a more realistic teaching scene for students, but there are still some limitations in practical use. With the deepening of network technology and the popularization of network applications, the combination of computer-assisted instruction system and it has made great progress. Cloud computing has become a global hot spot. It has developed rapidly and continuously affected people’s life, study and work. Cloud computing has become the third wave of technological reform since the emergence of the Internet [4]. With the accelerated development of cloud computing services and the continuous expansion of the scale of parallel jobs, cloud service providers have expanded the number of underlying computing, storage and communication components in their data centers to ensure that the performance, reliability and cost-effectiveness of applications reach the expected level. According to the above analysis, this paper will design an intelligent CAI System Based on cloud computing to help teachers teach according to students’ cognitive ability and interest characteristics, so as to greatly improve the teaching effect.

2 Hardware Design of Intelligent Computer Aided Instruction System Based on Cloud Computing Before building the hardware development environment, we should plan the overall architecture of the whole system, and use the module design method to subdivide the system into several modules, and finally implement them one by one. The main parts of the hardware environment of the intelligent computer aided instruction system designed in this paper are 5G module design and embedded microprocessor. Modular design of system hardware is adopted, which mainly consists of these modules: memory unit,

Power Supply

Reset circuit

USB

Teaching network interface card

Crystal oscillator circuit

5G communication module

S3C2440A

JTAG LCD FPGA

Flash

SDRAM

Fig. 1. Overall framework of intelligent computer aided instruction system

120

S. Liu and C. Gao

wireless network support, I/O port, embedded processor and peripheral devices [5]. Figure 1 is a schematic diagram of the overall framework of the intelligent computeraided instruction system designed in this paper. 2.1 Processor Module Design The hardware controller of the teaching assistance system designed in this paper uses S3C2440A microprocessor, which has ARM920T core and implements MMU, AMBA, BUS and Harvard cache architecture. S3C2440A internally integrates rich system resources, including: 4-channel DMA with external request pins, 3-channel UART, 2-channel SPI, 1-channel IIC interface, USB interface, LCD controller, external storage controller, 4-channel PWM timer and 1-channel internal timer, RTC, PLL on-chip clock generator, ADC and touch screen interface, etc. S3C2440A has 130 pins. In this design, 67 of them are used, including the pins of power supply, crystal oscillator and reset circuit used in the minimum system, as well as the pins needed for storage, network card, USB and external teaching auxiliary hardware. The supply voltage of the chip is 3.3 V and crystal oscillator of 12 MHz [6]. LDATA0 interface between microcontroller and NOR Flash, flash memory chip is K9F1208, the operating voltage of the memory is 2.7–3.6 V, internal use of paging mechanism for storage, K9F1208 read and write in page unit, erase data in block unit. Code is usually run directly in NOR Flash rather than having to be copied to RAM for execution, which increases execution efficiency. S3C2440A microprocessor is connected to FPGA chip, and the transmission and processing of RF communication signal are realized by FPGA. By using the overlapping and adding characteristics of rf signals, the serial output data is not so long, which makes full use of the subcarriers added in the frequency domain expansion. The FPGA control sliding window for signal sampling, sliding window step is 32, so the first all the way with enabling signal transmission, behind each data transmission delay in turn 32 length, and because the complete IFFT output for 256, eight times that of step, and no parallel match exactly, the next cycle of symbols and transmission according to this rule. The high-speed signal acquisition module mainly consists of signal conditioning and high-speed sampling. Signal conditioning mainly includes differential conversion circuit and anti-aliasing filter circuit. High - speed sampling mainly includes sampling circuit and clock circuit. The AD sampling chip AD9517, under the control of FPGA controller, collects the communication signals of each functional module in the operation process of the auxiliary teaching system and the students’ teaching feedback with the teachers by using the system terminal to realize the communication in the teaching process. 2.2 Communication Module Design If signal sent from rf front-end module is single-end signal, which is converted into differential signal through differential conversion circuit first, which improves the antiinterference ability of signal in transmission process, and then carries out anti-aliasing filtering. The data at the transmitter end after framing is sent to the antenna through the operation of the RF part of FPGA, and the receiving antenna of NI USRP-RIO and the clock module control antenna receive the data. The received data is converted into baseband data through down conversion, and the first step is to do synchronous processing

Intelligent Computer Aided Instruction System

121

to remove the noise field. Although one of the features of FBMC is that it abandons orthogonality between subcarriers and does not require strict synchronization, the first synchronization discussed is air-port synchronization in a wireless system, allowing the receiver antenna to separate valid data from noise. Signal transmission in the analog channel is bound to be mixed with some unknown high frequency interference signals, after turn single-ended difference signal and the unwanted signal components and noise, and can’t directly to the signal sampling, require the use of anti aliasing filter for filtering processing, maximum limit to restrain or eliminate the spectrum aliasing phenomenon, Avoid in-band aliasing and dynamic performance degradation [7]. The anti-aliasing filter used in this design is a bandpass filter. The ADS5474’s 400 MHz LVCMOS sampling clock is produced by the AD9517. The AD9517 is a clock management chip with four pairs of configurable parallel clock outputs. Includes two pairs of 1.6 GHz LVPECL outputs (each pair of clock outputs shares a 1 to 32x crossover) and two pairs of 800 MHz LVDS/CMOS outputs (each pair of clock outputs shares two serial 1 to 32x crossover); On-chip VCO adjustable frequency 1.45 GHz to 1.8 GHz; The AD9517 chip requires FPGA to configure its internal registers through SPI, and the signal line includes serial data (SDIO/SDO), clock signal (SCLK) and enable signal (CS_N). The SPI interface of the clock chip AD9517 is compatible with most synchronous transmission protocols, allowing the register of AD9517 to read/write operations, supporting single or multiple byte transmission, supporting MSB or LSB before transmission mode; The SPI interface can be configured as either a single wire bidirectional mode I/O interface (SDIO) or two unidirectional mode I/O interfaces (SDIO/SDO). By default, it is in single-line bidirectional mode. Serial shift clock input signal SCLK is used to synchronize SPI interface read/write operations, such as write operation on rising edge and read operation on falling edge [8] (Fig. 2). CLK

Loop filter

CP LF PASS

RESET CS_N

AD9517

SCLK SDIO

FPGA

Signal acquisition

Fig. 2. Signal sampling clock circuit

When the FPGA and AD9517 are powered on, the FPGA first gives the AD9517 reset pin (RESET_N) a low pulse, and its internal register is set to the default value. When the enable signal (CS_N) is pulled down, the INTERNAL register of AD9517 can be configured through SPI to set its output clock size, level standard, etc. The output data D[13:0] and the accompanying clock DRY± of the ADC chip selected in this paper are LVDS differential signal standards, and the data transmission mode is DDR (double data rate data transmission, data transmission at the same time of

122

S. Liu and C. Gao

the rising edge and falling edge of the clock) transmission mode. The LVDS level of the data is connected to the HR I/O bank of the FPGA, and the 100  terminals required for LVDS signals are added to the FPGA reception logic. Low power management of gateway boards is controlled by MOSFET. The main control chip can control the power supply of THE 5G module through 5G EN control signal. The student terminal and the system realize the end-to-end data transmission through TCP/IP network protocol. In the system, the TCP/IP receiver program of the server and client will always run in the background, waiting to receive the data from the edge and server. When the edge end responds to the request, the TCP/IP sending function in the edge end is automatically called. After the connection is established, the related data can be sent to the server. After receiving the data and saving the data to the corresponding directory, the server runs the TCP/IP receiving function again and waits to receive the data. With the hardware part of the intelligent computer-aided teaching system designed above as the support frame, cloud computing and other technologies are used to realize the intelligent assistance to the teaching process, so as to provide the functions required by teaching activities.

3 The Software Design of Intelligent Computer Aided Instruction System Based on Cloud Computing 3.1 Cloud Computing Module Design The establishment of the auxiliary system supported by cloud computing helps learners to extend and expand their learning time and teaching space. The whole virtual cloud architectures need to build the cloud cluster environment as a whole system of data storage and operation center, through virtualization software integration server resources to form a resource pool and multiple virtual one unified allocate these resources installed a different operating system virtual desktop, students end USES a terminal connected to the network optical switches and access their virtual desktops. Six servers, including the Active Directory domain controller, SQL server, vcenter server, Composer server, and Connection server, are required to set up enterprise-class virtual desktops. When deploying the cloud computing nodes of the teaching assistance system, the problem of energy consumption under large-scale and high-concurrent operations needs to be considered. Provide fault tolerance for task failure while minimizing energy consumption. In this paper, an energy saving and reliable task replica deployment algorithm, RER, is proposed. The algorithm takes a task replica running at low speed and allows multiple task replicas to share the same server resources. Different servers have different free time, and different task copies have different execution time sizes. Assume that the task copy scheduling process will use N servers to execute all |TS | task copies, with N = |TS | initially, and each master task deployed separately on a master server. Before the algorithm is implemented, the server and the task copy are sorted in non-ascending order according to the idle time and the task copy execution time respectively, so as to avoid repeated sorting during the algorithm [9]. The optimization objectives are as follows:  (1) F = min us

Intelligent Computer Aided Instruction System

123

In the formula, us indicates server dk yes (=1) No (=0) is used. The optimization constraints of the above formula are as follows:  1, 1 ≤ k ≤ |TS |  Ns = (2) min 1, Ns , k > |TS |  Q, k > |TS | (3) μk = Q − S(Tk ), 1 ≤ k ≤ |TS | In the formula, μk is the idle time of server dk ; Q is the estimated completion time of the operation; Ns is the number of servers used in cloud computing; S(Tk ) is the time at which a copy of task Tk is executed at the maximum processing rate of the server. If the response speed of each main task is known when the cloud computing server provides corresponding services, tasks can be allocated according to the idle degree of the cloud computing service to improve the concurrent processing efficiency. 3.2 Recommended Design of Teaching Resources For the teaching assistant system, in order to obtain better teaching effect, the auxiliary system needs to be able to personalized teaching resources according to the teaching feedback of students. To this end, knowledge graph is established and bayesian network is used to recommend learning resources for students. According to the information collected when students use the system functions to learn, the corresponding entities are extracted to establish the knowledge map. BiLSTM+CRF model is used for entity recognition in this paper, which consists of three layers. The first layer of the model is the look-up input layer, which converts sentences into word vector and word vector. The second layer is the double-layer LSTM layer, which is to input word vector and word vector in BiLSTM, and then output the scores of all labels of each word in the sentence. The third layer is the sequence annotation layer, which takes the output of the bidirectional LSTM layer as the input and finally obtains the probability value of the tag sequence. The bidirectional GRU model is used for relational extraction of the entities extracted in the previous step. According to the different requirements of the education syllabus for teaching knowledge, the weight of knowledge points is divided into understanding, understanding, mastery and application, and the weight is given from 1 to 4 respectively [10]. Calculate the weight of knowledge nodes in the knowledge graph:  (4) w = (1 −  )C +  P(Wk ) In the formula, (1 −  )C represents that the current node is accessed by other nodes with its intermediate centrality value, that is, the knowledge node with higher intermediate centrality has higher access value; C represents the mediality centrality of different knowledge points. P(Wk ) is the proportion of the weight of knowledge node k in the weight of its brother nodes with the same direct precursor as k node. Therefore, the knowledge map of different students is established, and according to the map, bayesian inference network is used to recommend personalized resources for students. The bayesian network is transformed into a joint tree, and then the global consistency of the joint tree is achieved through message transfer between nodes, and

124

S. Liu and C. Gao Bayesian network

Good figure

Recommend ed resources

Triangular figure

Global uniform union tree

Associative tree structure

Globally inconsistent union tree

Student knowledge map

Fig. 3. Resource recommendation flow chart

then the probability calculation is carried out. Figure 3 shows the flow chart of learning resource recommendation. In the process of teaching, students’ knowledge graph is constantly updated and obviously differentiated. The process shown in the figure above can be used to recommend appropriate learning resources for students. 3.3 Auxiliary Teaching System Database Design Whether the various functional modules of the teaching assistance system can be closely combined and how to combine hinge on whether the structure design of database is reasonable, especially the design of data table structure. Through the demand analysis of the teaching assistance system, and then guided by the principle of database, we get all the entities in the teaching assistance system, and each entity and its attributes belong to a basic table. The corresponding relationship table of some system entities is shown in Table 1. Table 1. Correspondence table of some entities in the database of teaching assistance system Field

The data type

Field size

Whether is empty

Key constraint

The teacher

Int

11

N

Primary key

The name

varchar

58

Y

–

Password

varchar

64

Y

–

Student student id

varchar

20

N

Primary key

The student’s name varchar

58

Y

–

Grade

int

11

Y

–

The class

int

11

Y

–

Teaching text resources

Longtext

200

Y

– (continued)

Intelligent Computer Aided Instruction System

125

Table 1. (continued) Field

The data type

Field size

Whether is empty

Key constraint

Teaching video resources

varchar

150

Y

–

Teaching picture resources

char

180

Y

–

Teaching evaluation

Longtext

300

Y

–

When using the database to operate the data in it, it is inevitable to encounter performance problems such as access efficiency. Especially when the amount of data information is large, some unnecessary or unreasonable additional operations are likely to lead to the collapse of the database system and even the application. When the MySQL database accesses the table, the life cycle of “query” is roughly from the client to the server, then parse and execute the task on the server, and finally return the execution result. The stage of executing the task includes not only the call of retrieving data to the storage engine, but also data processing such as sorting and grouping. These calls need to operate memory and CPU, and may also produce certain context switching, so they are the core of the “query” life cycle and a waste of time. In order to avoid the participation of irrelevant data when accessing data, it is required to optimize at the code level and database design level. On the premise of meeting the needs of users, find out the bottleneck of the system and improve the overall performance of MySQL system services. Since the database used in this teaching assistant system is MySQL and the stored data are common, some common methods such as selecting reasonable field attributes, using foreign keys and indexes, optimizing query statements and so on are used to optimize data access. The above software design content provides personalized and intelligent auxiliary functions for CAI system. Put the teaching assistant software function designed above in the hardware running environment, that is to complete the design of intelligent computeraided teaching system based on cloud computing.

4 System Performance Test After the completion of system development, the system needs to be tested to realize the quality assurance of software. This project will test the system from two aspects: function test and performance test. By constructing test cases and adopting appropriate test tools, the correctness of software functions is verified, and the problems existing in the system are detected and found to ensure the normal delivery of software. The whole CAI system includes several functional modules. In order to confirm that the CAI system can be used in actual teaching activities, its function can meet the daily teaching needs. In order to improve the persuasiveness of the experimental results, the experimental environment of the control group and the experimental group was consistent.

126

S. Liu and C. Gao

4.1 Experimental Content Choose the traditional teaching assistant system to compare with this system. Firstly, the logic unit of the software is tested separately to test the correctness of the function of the logic unit of the system; Integration testing is to integrate the logical units of the system through the interface and test the problems existing in the integration between functional modules. When users use the system, if the response time of the system is too long, it will cause a poor use experience. Generally, the tolerance of system users to the response time of the system is within 5 s. The speed of teaching has a great impact on the normal teaching process. Therefore, the system uses response time to test the performance of the system. The performance of the system is evaluated by simulating the number of users of different orders of magnitude and accessing the system concurrently. Then carry out the function black box test of the system: according to the system function test case, check whether the program can receive data and produce correct output information, and maintain the integrity of the database or file. Determine test cases and infer the correctness of test results according to the program functional requirements specification. Black box test is to test whether the system function operates correctly on the basis of knowing the function of the system. 4.2 Experimental Result Figure 4 shows the changes of students’ academic performance after the two teaching assistant systems are applied to teaching activities.

Proportion of excellent students

0.7 0.6 0.5 0.4 0.3 0.2

10

20

30

40 50 Tes t times

Unused s ystem Traditional system

60

70

80

Paper system

Fig. 4. Comparison of application of teaching assistance effect

It can be seen from Fig. 4 that after the application of the teaching assistant system, the students’ scores are improved to a certain extent, but the students who apply the system in this paper have better academic scores, that is, the practical application effect of the system in this paper is better. Table 2 shows the comparison of the response time of the two teaching assistant systems.

Intelligent Computer Aided Instruction System

127

Table 2. Comparison of response time of teaching assistant system Concurrent visits

Paper system

Traditional system

Minimum system response time/ms

Average system response time/ms

Maximum system response time/ms

Minimum system response time/ms

Average system response time/ms

Maximum system response time/ms

1000

106

118

112

128

168

148

2500

121

134

127

185

234

206

5000

154

162

143

236

287

260

10000

189

203

158

287

345

316

15000

233

257

174

369

402

374

20000

267

289

189

452

536

494

25000

293

335

205

491

588

538

50000

364

390

216

624

673

643

According to the test results, it can be seen that the system in this paper can still maintain a good response time without increasing the concurrent access of the system. Compared with the traditional teaching system, the response time of the teaching assistant system designed in this paper is shorter, which shows that the response rate of the system designed in this paper is faster. Numerically, the response time of this system is shortened by about 1/2, which can meet the auxiliary function requirements of teaching activities under the condition of high concurrency. Summarizing the above system performance test data analysis, it can be seen that the intelligent CAI System Based on cloud computing designed in this paper has a high response rate and can meet the functional response under the condition of high concurrency. Compared with the traditional teaching assistant system, the application of this system in teaching can greatly improve the students’ academic performance and their learning enthusiasm, which has high practical application value.

5 Conclusion The development of information technology is also making profound changes in the field of education, providing a new stage for modern education, and promoting a new great leap in educational technology, educational system and teaching model. Computer aided education is an important embodiment of the application of electronic information technology in education. The computer does promote the teaching reform, improve the quality of teaching, lighten the burden of teachers and improve the ability of students. Cloud computing has infiltrated into all fields of today’s society. A number of colleges and universities have joined the program abroad, and cloud computing has entered the category of education research in China. Considering the current teaching environment and social background for the continuous improvement of the demand for teaching assistant technology means, in order to properly handle the perfect application of computer

128

S. Liu and C. Gao

aided instruction system in the future teaching reform, this paper studies and designs an intelligent computer aided instruction system based on cloud computing. For the next step of work, can increase the tutorial guidance module, the module can control the students’ learning situation, grasp the students’ learning level and make timely feedback, determine the next step of teaching content and should adopt the teaching methods, strategies, different levels of students with different methods, to make the students master the learning initiative.

References 1. Dai, T., Lu, Z., Zhu, L., et al.: Experiment teaching assistant system of laser resonant cavity based on extended reality technology. Res. Explor. Lab. 40(08), 190–194 (2021) 2. Li, N., Zhang, S.: Teaching and training platform for virtual equipment of replenishment based on Web3D. J. Syst. Simul. 31(06), 1136–1141 (2019) 3. Xu, N., Fan, W.: Research on interactive augmented reality teaching system for numerical optimization teaching. Comput. Simul. 37(11), 203–206+298 (2020) 4. Lu, P., Lu, L.: Design of distributed network mass data processing system based on cloud computing technology. Mod. Electron. Technol. 43(18), 36–39 (2020) 5. Wei, X.: Teaching reform exploration of embedded systems and principle of network devices based on UAV systems. Softw. Guide 20(12), 216–220 (2021) 6. Pesce, M.: Peering into the pandemic end game: before COVID–19 fades, we’ll see a flurry of advances in contact tracing, cloud computing, surveillance, and online gaming. IEEE Spectr. 58(1), 22–25 (2021) 7. Devi, K., Muthusenthil, B.: Intrusion detection framework for securing privacy attack in cloud computing environment using DCCGAN-RFOA. Trans. Emerg. Telecommun. Technol. 33(9), 4561–4583 (2022) 8. Zhu, Y.: Systematic design and systematic mode construction of blended teaching from the perspective of deep learning. China Educ. Technol. 11, 77–87 (2021) 9. Toonder, S., Sawyer, L.B.: The impact of adaptive computer assisted instruction on reading comprehension: identifying the main idea. J. Comput. Assist. Learn. 37(5), 1336–1347 (2021) 10. Lejeune, L.M., Lemons, C.J.: The effect of computer-assisted instruction on challenging behavior and academic engagement. J. Posit. Behav. Interv. 23(2), 109–118 (2020)

Construction of Online Ideological and Political Education Platform Based on Artificial Intelligence Technology Huijuan Li(B) and Xiuying Dong College of Marxism, Changchun University of Finance and Economics, Changchun 130122, China [email protected]

Abstract. Because the current online ideological and political education platform is difficult to build a virtual learning environment, and it is difficult to ensure that users can access the network regardless of time and place through the mobile network, resulting in the decline of teaching quality, this paper puts forward the construction method of online ideological and political education platform based on artificial intelligence technology. Artificial intelligence technology is used to collect and store massive ideological and political information, build a database, and optimize the functional structure of the teaching platform in combination with mobile technology, so as to build a virtual learning environment based on artificial intelligence, so as to ensure that users can access the network without time and place restrictions through the mobile network and realize the goal of educational resource sharing, complete the construction of online ideological and political education platform. The experimental results show that the average score of students’ curriculum knowledge test after application has increased by 47.1 compared with that before application. The online ideological and political education platform is very dissatisfied with 3.2%, dissatisfied with 8.9%, generally 18%, satisfied with 61% and relatively satisfied with 8.9%. It shows that the online ideological and political education platform designed in this paper has higher practicability in practical application, can effectively improve the teaching quality and fully meet the research requirements. Keywords: Artificial intelligence · Ideological and political · Education platform

1 Introduction With the rapid development of Internet technology, the large-scale popularization of smart terminal devices such as smart phones and tablets and the gradual enrichment of mobile network resources, the digital and mobile online learning methods are more and more accepted by people. The market scale of online education industry is increasing year by year, and the impact of online education on users is also expanding [1]. At the © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 129–144, 2022. https://doi.org/10.1007/978-3-031-21161-4_11

130

H. Li and X. Dong

same time, there are still many problems in online education, such as poor learning effect, difficult enterprise profitability and so on. Based on this, this study takes the learning platform of artificial intelligence technology as the research situation. On the basis of combing relevant research and theories, through interview and social network analysis methods, this study explores that platform design, interaction design, teaching resources and teachers are the factors affecting the learning effect of users of online education platform. On this basis, this paper constructs the theoretical model of influencing factors of user learning effect of artificial intelligence technology online education platform. The structural equation model method of confirmatory factor analysis is used to empirically test the theoretical hypothesis, and further explore the influence degree of different influencing factors. The influence of interaction design platform design, teaching resources and teachers on the learning effect of online education platform users decreases in turn. Finally, based on the above research, this paper puts forward the strategies to improve the learning effect of online education users. This research has innovation in research methods and research contents. After the experimental research on the theoretical model, it further explores the influence degree of various influencing factors, and explores and expands the online education theory to a certain extent. Because the current online ideological and political education platform is difficult to build a virtual learning environment, and it is difficult to ensure that users can access the network regardless of time and place through the mobile network, resulting in the decline of teaching quality, this paper puts forward the construction method of online ideological and political education platform based on artificial intelligence technology.

2 Online Ideological and Political Education Platform Construction Based on the analysis of the functional structure of the online ideological and political education platform, this paper uses the analytic hierarchy process to design the online ideological and political education quality evaluation algorithm, and realizes the construction of the online ideological and political education platform by designing the information screening process of political teaching resources, the functional activity process of curriculum management stage, and the functional relationship of the online ideological and political education platform. 2.1 Function Structure of Online Ideological and Political Education Platform Taking the online education platform as the research object, this paper discusses the theoretical research on the interactive relationship among users, platform and continuous use intention. With the development of the Internet and the progress of science and technology, artificial intelligence online education is constantly changing, and the use needs of users are constantly improving and changing with the mature application of user experience research, and new influencing factors are constantly introduced into the research. The overall design of Online Ideological and political education platform is the first stage of the software design process. The main task of this stage is to divide the whole platform into reasonable modules according to the demand analysis of the platform and the module division idea of high cohesion and low coupling, so as to facilitate the later

Construction of Online Ideological and Political Education Platform

131

detailed design. At this stage, we also need to design the overall architecture of the platform, and finally complete the design of the database. The specific structure is as follows (Fig. 1):

Application layer

Business layer

Embedded layer

Django structure

nginx

Model layer

View layer

Template layer

Functional module

Web Browser uwsgi

Course browsing User related

Teacher module Course recommendation

Curriculum organization module Background management system

Fig. 1. The overall architecture of an online ideological and political teaching platform based on artificial intelligence

The online ideological and political teaching platform based on artificial intelligence is divided into three layers: application layer, embedded layer and business layer. The application layer takes the browser as the core, the embedded layer takes uwsgi and nginx as the support, and the business layer is divided into Django structure and function modules, so as to provide support for the good operation of the online ideological and political teaching platform based on artificial intelligence. Good design is the premise of developing a good platform. The ultimate purpose of platform development is to provide users with a scientific and efficient network teaching platform. More and more online education platforms begin to carry out knowledge sharing mode. Initial adoption is the first step to obtain potential users. The influencing factors of initial adoption have a direct correlation with whether users are willing to experience and try, but initial adoption can not benefit the platform. Whether users can insist on using and voluntarily manage the course is the most critical step, It depends on whether the user has the intention of continuous use. User related functions are divided into login registration module and personal center module because of the difference of functions. By combing the information service content, several classifications of online education related information content are obtained, as shown below:

132

H. Li and X. Dong

Online education related information content

Platform information

It mainly includes platform background, teaching form, teaching advantages, curriculum planning, etc

Teacher information

It mainly includes the list of subject teachers, personal information of subject teachers, teaching characteristics, etc

College Information

It mainly includes the list of colleges, students' personal information, learning situation, learning problems and learning objectives

Course feedback

Teaching content and course status records generated in the process of customer supervision, comments and feedback of students and subject teachers, etc

Class information

It mainly includes course scheduling time, subjects, class hour management information, etc

Fig. 2. Ideological and political online education related information content

It can be seen from the analysis of Fig. 2 that the information related to ideological and political online education includes platform information, teacher information, college information, course feedback, class information, etc. The two functional groups of course institution and course lecturer are responsible for the information processing of institutions and lecturers. Because lecturers and institutions are nested internally, and their basic functions and implementation methods are basically the same, in order to reduce the coupling between modules, the two functions are combined into one module, namely institution and lecturer module. The course module is used to process the business functions of the relevant functional groups of the course, It realizes the functions of shopping cart and order management. The course recommendation module is responsible for realizing the relevant functions of statistical based course recommendation and personalized recommendation. The relevant processing logic of the background management sub platform converges to the background module. Finally, the platform is divided into seven modules: login registration module, organization and lecturer module, course module, single transaction module, center module, course recommendation module Background management module [2]. The platform function module structure is shown in the figure below (Fig. 3):

Construction of Online Ideological and Political Education Platform

133

Login registration module

Organization and lecturer module

Course module

Online education platform

Order transaction module

Personal center module

Course recommendation module Background management module

Fig. 3. Platform module functional structure

The homepage of the online education platform provides courses personalized and recommended by the platform for users. The curriculum recommendation function is mainly based on the user’s behavior data, so to achieve personalized recommendation, first obtain the user’s behavior data, and then build a suitable recommendation engine based on the data. Obtain the user’s participated courses, favorite courses and course scoring data from the MYSQL database. When the user selects courses or collects 3 courses in total, the personalized course recommendation process is triggered. This article uses artificial intelligence technology as the recommendation algorithm for personalized courses [3]. All recommended courses are directly saved in Redis. When a user logs in, the platform randomly selects 5 courses from Redis as personalized recommendations and displays them on the platform’s homepage. Course recommendation based on artificial intelligence technology is obtained by executing SQL calculations through timed tasks every day. 2.2 Ideological and Political Online Teaching Quality Evaluation Algorithm In order to improve the performance of the online ideological and political education platform based on artificial intelligence technology constructed in this paper, the analytic hierarchy process is used to design the ideological and political online teaching quality evaluation algorithm, which is mainly based on the expert judgment matrix and according to the score deviation between courses and the user’s historical score, predicts the score of the courses that are not scored after the user’s learning, so as to evaluate the ideological and political online teaching quality. The algorithm is applied to the function evaluation model of information teaching platform to improve the efficiency and quality of online ideological and political education quality evaluation, and promote the further improvement of the performance of online ideological and political education platform.

134

H. Li and X. Dong

The scale in the judgment matrix of ideological and political online teaching quality evaluation refers to the comparison of the superiority or importance of every two evaluation elements between each level in the criterion layer and the measure layer. The degree of comparison between the importance of the element “i” and the latter element “j” is mainly divided into the 19 (and its reciprocal) scale method, which is a pairwise comparison of the first-level evaluation index (user experience dimension) in the evaluation system model, and the second-level evaluation Pairwise comparison of indicators, 1–9 (and its reciprocal) specifically refers to the relative importance of the two indicators that are compared. When the current element is more important than the latter element, it is represented by an integer of 1–9, and vice versa. The reciprocal of 1–9 is shown in the table for details (Table 1). Table 1. Judgment matrix scale table Scale (aij)

Meaning

1

The importance of the former and the latter is “equal”

3

The degree of importance of the former and the latter is “slightly important”

5

The importance of the former and the latter is “obviously important”

7

The importance of the former and the latter is “very important”

9

The importance of the former and the latter is “absolutely important”

2, 4, 6, 8

The importance of the two is between the adjacent quantitative standards

Reciprocal

The comparison between the latter and the former is 1/(aij)

According to the index elements in the expert judgment matrix questionnaire, the comprehensive judgment matrix analysis is carried out. The expert only needs to score one of the triangles in the lower left or upper right corner of the questionnaire, which is marked as α, and the corresponding other Half is the reciprocal β of the expert scale. Assuming that the relative weight vector of the element is A = (a1 , a2 , ..., an ), the calculation method of the relative weight is as follows: si =

n  j

n(β + 1) n(α − 1) − n  1 n     n A n βij − 1 nj=1 αij + 1

k=1

(1)

i=1

The slopone collaborative filtering algorithm in the recommendation algorithm is further used to make personalized course recommendations to users. The advantage of this algorithm lies in its simple structure, easy implementation, high execution efficiency, and small amount of storage required [4]. At the same time, the accuracy of the recommendation is relatively high. The algorithm also has a relatively high-quality recommendation accuracy. Even if the user’s participation on the platform is small, they should be able to obtain effective recommendations. This paper uses the user’s score k on the course after learning as the calculation data recommended by the slopone algorithm.

Construction of Online Ideological and Political Education Platform

135

Essentially, this method uses a simpler form of regression expression and a single free parameter, such as the formula:  f (x) = k si (x + b) − n (2) If user r’s ratings of course i and course j are 1 and 1.5, respectively, j is 0.5 points more than i. When the user learns course i and scores it 2, the platform refers to user A’s ratings of courses i and j. It is derived that the user’s possible score for the course j is 25 points. Calculate the average score difference between courses, the calculation result is recorded as the score deviation between courses, the calculation formula is as follows:   rui − ruj − ra (3) R(x) = |f (x) − si | where ra is the rating of user r on course i, and rui is the rating of user u on course j. Tij is the user who has overrated the object, Hij is the user who has rated both the course i and the course j, N (u) is the number of users who have rated both the course i and the course j, according to the score deviation between courses and the user’s history scoring, predicting the score of unrated courses after users study. The calculation process is shown in the formula.  i∈N (u) |Tij − Hij | − R(x)  −γ (4) Puj = i∈N (u) |N (u) − 1| Among them, γ is the course rated by user u. The information platform success model is mainly applied to the evaluation research of various information platforms, including library retrieval platforms, e-government, competitive intelligence platforms, social networking sites, online document sharing platforms, etc. In the article, the online education platform has its own characteristics compared with the general information platform.

System quality

Information quality

Service quality

Use or willingness to use

User satisfaction

net profit

Fig. 4. Improved function evaluation model of information teaching platform

136

H. Li and X. Dong

According to the data in Fig. 4, the improved information teaching platform function evaluation model mainly evaluates the system quality, information quality and service quality from the perspective of user satisfaction, net profit and use intention. The most important thing of online education platform is to transmit educational information resources. Therefore, information quality, that is, curriculum quality, is the focus of evaluating online education platform [5]. Secondly, the platform provides online education services, and its responsiveness, functionality and other platform quality factors also affect the development of the platform. Finally, the online education platform mainly provides services through two aspects: one is the video, notes, retrieval and other functional services provided by the platform itself; the other is that teachers provide course content services through class, homework and communication. The functional service quality can be reflected by the platform quality, and most of the content service quality can be measured by the information quality. However, combined with the characteristics of education, the interaction between students and teachers and the interaction between students are important influencing factors, which can not be included in the platform quality or information quality. To sum up, this paper will evaluate the online education platform from the perspective of platform quality, course quality and interaction. 2.3 Realization of Online Ideological and Political Education The user enters the course list page through the navigation bar, and arranges the courses in a pagination method for a large number of courses. The courses on each page show 3*6 courses. In the right column of the course list, there will be a popular course recommendation based on the number of clicks of the course, showing three courses ranked at the top. The page is also sorted according to the number of new favorites and the number of learners. The course list process is shown in the figure:

start

Course list page

Course search

Enter keywords

search result

Select sort method

New course list

Select course

Popular course recommendati on

Select recommended courses

Course details page

end

Fig. 5. Ideological and political teaching resource information screening process

By analyzing the data in Fig. 5, it can be seen that after users search keywords indepth on the online ideological and political education platform, a course list will be generated in the platform, and a new course list will be generated by selecting and sorting. The course details page will be presented to users through the selected courses and course recommendations. The right column of the page is mainly composed of two parts: introduction to the course organization and relevant course recommendations. The course

Construction of Online Ideological and Political Education Platform

137

institution column mainly displays the institution name, number of courses, number of lecturers and region, and can collect institutions here [6]. The course scheduling stage is the stage of arranging formal courses for formal students, and following up the formal return visit and daily maintenance of students. In this stage, the key needs of students and subject teachers are the interaction of course arrangement, pre class reminder and post class evaluation feedback. Enterprise staff need to reward students who have made progress in learning and continue to urge them to learn. The specific course management process is as follows: Online education platform products

Students and subject teachers

Acquired information system

Independent course arrangement

By arranging classes

Student parents

Online head teacher

Course reminder

Course teaching

Course mutual evaluation feedback

Sorting progress

Fig. 6. The functional activity process of the course management stage

By analyzing the data in Fig. 6, it can be seen that the curriculum management is related to the roles of students, subject teachers, students’ parents and head teachers. The curriculum arrangement is obtained through the online education platform, and the teachers and students are reminded of the curriculum. The students give the curriculum feedback, and send the curriculum feedback results to the teachers. The head teachers can obtain the sorting progress, and feed it back to the parents, students and teachers. The freedom of online education class changes the course arrangement from passive to active The online head teacher informs the students and subject teachers of the course scheduling notice. The students and subject teachers respectively choose the class time on the products of the online education platform. The course schedule is recorded on the background information platform. The course scheduling academic affairs will remind the class 20 min before the class after screening the overlapping time and confirming the course schedule. After the course, students and subject teachers evaluate each other,

138

H. Li and X. Dong

and the evaluation feedback of both parties is recorded in the course records of both parties. The online head teacher shall store the course supervision records in the course records at the same time, and regularly sort them out and send them to the students’ parents The online head teacher collects the feedback on the progress of students, and can also view the relevant courses recommended by the platform. The relevant courses in this department are also based on feature statistics. Mainly by adding course labels to the course model, three courses with the same labels as this course are searched in the database in the form of labels [7]. Online education information service refers to providing information services for students or enterprise staff in the process of online education service experience. Specific information services include the dissemination of information on the online education platform, the collection, storage and feedback of students’ information, and the exchange of course related information. In the existing online education information service platform, the functions related to students are mostly information storage functions, there are few activities in information interaction. The online education information service platform optimized and designed by the author is active and interactive, which increases the interactive operation between students and enterprise staff and promotes the mutual relationship, so as to improve the enthusiasm of students to actively obtain information and improve the work efficiency of enterprise staff. Information management

Information acquisition

Teacher management

Teaching and research management

Class management

Enterprises recruit subject teachers and record the personal information of subject teachers. Match students' teaching through relevant records

The teaching courseware of online courses shall be independently uploaded or arranged by subject teachers, and developed and produced by the educational administration

The course arrangement and class information of the college need to be confirmed repeatedly

Student information

Understand students' personal information through communication with students' parents, and record students' information to the background information system

Information interaction

Course arrangement

Confirm the lag of course arrangement through instant communication tools, and students obtain course arrangement information through online education platform

Curriculum evaluation feedback

The mutual evaluation between students and subject teachers after the course, and the evaluation results are recorded in the background information system

Learning achievement sharing

Xueyan's school examination results are improved, which is shared by the students' parents to the enterprise staff, and then shared with the help of the online communication slip

Fig. 7. Basic concepts of online education information services

According to Fig. 7, the basic concepts of online education information service include information management, information acquisition and information interaction. Information management includes teacher management, teaching and research management and class management; Information acquisition includes extracting student information and course arrangement; Information exchange includes course evaluation feedback and learning achievement sharing. Online education information service platform can help students and enterprise staff establish good information transmission and communication. The relevant teaching information, student information, learning

Construction of Online Ideological and Political Education Platform

139

situation information, learning objectives, curriculum planning information and other information involved in this topic will be integrated and unified in a complete information service platform. The design of online education information service platform is to meet the information services of stakeholders in the product service platform. In order to achieve the goal, it is necessary to fully understand the information function requirements and platform process of stakeholders in the platform, as well as the content and form of each contact. The architecture of the education and teaching integrated platform is designed into a multi-layer development model according to the needs. Its main modules are: user presentation layer, business logic layer, data persistence layer and database layer [8, 9]. In order to develop the architecture of the platform in more detail and show the functions and connections of each layer more clearly. The following figure is obtained by analyzing the functional relationship of the education and teaching platform in the platform.

Communication data interface

Data persistence

Database

Business logic

User management User registration Front desk management

Login module

Competition management module Teaching resource module

back-stage management Tutor Q & A

Online question module

Resource upload

Resource download module

Management module

Forum foreground module

Background module

Teaching curriculum knowledge system

Media platform

Key and difficult points

Learning state

Course management

Curriculum construction

Learning space

Fig. 8. Ideological and political online teaching platform functional relationship

As shown in the Fig. 8, the educational teaching platform provides many convenient conditions for computer teaching. Any application or learning service provided through the teaching platform is a service from the educational teaching platform. The education and teaching platform can mobilize various resources in the network according to

140

H. Li and X. Dong

the user’s real-time location, course time, learning purpose and other information, and provide customers with convenient and fast services to meet customer needs [10]. From the perspective of platform functions, it can integrate scattered resources to meet the various needs of different users. Integrating various educational information resources together, using an educational teaching platform can provide conditions for the smooth implementation of online teaching activities, and the platform can complete educational and teaching activities.

3 Analysis of Experimental Results In order to test the actual use effect of this platform, students log on to this platform through mobile devices equipped with WiFi to learn and participate in periodic tests of related courses. The traditional online ideological and political education platform design method is to develop a small database of medical students’ Ideological and political education knowledge competition based on python, use the small program online knowledge competition answer platform to infiltrate the ideological and political education in stages, and guide students to actively scan and participate in answer learning with their usual grades as rewards, so as to improve students’ autonomous learning ability. The crawler technology is used to obtain the data of the education platform, and the data is integrated to remove the noise in the data. Finally, the obtained data is 4.56 GB. The obtained data is divided into experimental set and test set. The data in the test set is input into the simulation platform for testing to obtain the optimal operating parameters and ensure the authenticity and reliability of the experimental results. In this paper, the functional testing mainly uses the black box testing method to carry out testing and test case design. Functional testing is mainly to test the specific functions that each functional module of the platform should achieve. By selecting different test data corresponding to it, it is judged whether the program has reached the expected goal, and finally a complete test report is formed according to the test results. Part of the test the results are shown in the table (Table 2): Table 2. Some functional test results of the platform Test items

Function description

Test result

User login

Check whether the user name, password and verification code are correct

It can normally verify user login, and has prompt function to respond immediately

Enter course

Enter the new course information It can increase courses to check whether the information normally and respond is correct immediately (continued)

Construction of Online Ideological and Political Education Platform

141

Table 2. (continued) Test items

Function description

Test result

Modify course information

Modify course information

Can normally query the progress of corresponding courses and respond immediately

New test paper

Enter the new test paper information and check whether the information is correct

It can add test papers normally and respond immediately

Score query

Query the examination results of The query of scores runs students in each course normally and responds immediately

According to the process of module function, test cases are used to verify the platform function. The following will start with the four main functions of course browsing, personal center, background management and order transaction. The design of test cases mainly focuses on the main function points of the module and divides the test points in the function points. According to the results of the normal operation of the platform, the expected results of the test point are obtained, and finally the test results of the test point are obtained to complete the test. In order to determine that the course browsing module can operate normally, the test cases of the course browsing module are designed. According to the test results, the course browsing module can provide services normally. Course browsing module test table (Table 3): Table 3. Test cases of course browsing module Function point

Test point

Expected results

Test result

Course search

Invalid character entered

Search failed, no course list returned

Adopt

Enter valid characters

Search succeeded, return to the course list

Adopt

Course ranking

Can courses be sorted by most collections

In order

Adopt

Course details page

Can I enter the course normally from the course list

Enter the course details page and display the full function

Adopt

Introduction to teaching institutions

Whether it can be displayed normally

Can collect parity stocks

Adopt

In order to make sure that a background management module can run normally, corresponding test cases are designed. According to the test results, the background

142

H. Li and X. Dong

module can provide services normally. Example table for back-end management module test (Table 4): Table 4. Test cases of platform background information management Function point

Test point

Expected results

Test result

Administrator account

Can the administrator get permission

The administrator can operate the background system

Adopt

Operation of user information

Can you add, delete, modify and check completely and normally

Be able to manage the Adopt basic information of users and return results successfully

Operation of course information

Can you complete the Be able to manage basic addition, deletion, course information and modification and query of return results successfully the course

Adopt

Operation of institutions Can you complete the Be able to manage basic and lecturers addition, deletion, information and return modification and query of results successfully institutions and lecturers

Adopt

Authority unit management

Allocation of different permissions

Use different accounts to set different permissions

Adopt

Log function

Is the corresponding log generated

Log record successfully displayed

Adopt

Single-factor inter-subject room design, independent variable is the teaching method, including two cases, the control class adopts the conventional teaching method, and the experimental class adopts the F online education platform learning method. The duration is one semester, and the closed-book examination will be conducted after the end of the course. A one-way analysis of variance was performed on the test scores, and the results showed significant differences in test scores after participating in learning, as shown in the table (Table 5). Table 5. Means and standard deviations of students’ curriculum knowledge test scores under different teaching methods Pre test scores

Post test results

Control class

48.85 (6.98)

89.65 (5.59)

Experimental class

49.55 (6.25)

96.65 (6.35)

The average score of students’ curriculum knowledge test after application has increased by 47.1 compared with that before application. It can be seen that the scores of

Construction of Online Ideological and Political Education Platform

143

students who use online education platform are significantly better than those of students who use traditional teaching methods. In addition, the results of the investigation on the learning effect of the two classes after the course are shown in the table (Table 6). Table 6. Comparison of learning effect tracking Satisfaction object

Extremely dissatisfied

Dissatisfied

Commonly

Satisfied

Quite satisfied

Control team (38 persons)

6.2%

15.6%

30.5%

42.5%

5.2%

Experimental class (38 persons)

3.2%

8.9%

18%

61%

8.9%

The online ideological and political education platform is very dissatisfied, accounting for 3.2%, dissatisfied, 8.9%, generally 18%, satisfied, 61%, and relatively satisfied, accounting for 8.9%. It can be seen from the test and survey results that the learning effect of students in the experimental class is significantly better than that in the control class. In other words, the data mining algorithm used to track students’ learning behavior and effect has played a role. The platform tracks and records students’ learning behavior data, analyzes students’ online learning status, and uses the decision tree method to let students predict and understand their cognitive ability, so as to help students improve their learning ability, strengthen learning management, and finally enhance the effect of learning. This platform basically realizes mobile, intelligent and personalized teaching, and gradually teaches students according to their aptitude.

4 Conclusion After the online education platform of Ideological and political course based on artificial intelligence, the resource sharing and teacher-student interaction of the course become convenient and fast. Students practice and self-test after class through the course practice and test platform, which stimulates their learning enthusiasm and autonomy. The teaching work is greatly facilitated based on website survey and automatic statistics. The online ideological and political education platform based on computational thinking makes full use of the advantages of network resources, reasonably designs and implements the online education platform, and improves the teaching effect to a certain extent. Fund Project

1. National Social Science Foundation Ideological and Political Course Special Project for Colleges and Universities (No. 21VSZ78).

144

H. Li and X. Dong

2. Research project of higher education teaching reform in Jilin Province: Research and practice of halving classroom teaching mode of ideological and political courses in private colleges and universities (project number: JLJY202192348402). 3. Jilin Province Higher Education Society Project: Research on the Path of Improving Undergraduates’ Sense of Ideological and Political Courses Based on the Perspective of Students (No.: JGJX2021D514).

References 1. Dama, C., Langford, M., Dan, U.: Teachers’ agency and online education in times of crisis. Comput. Hum. Behav. 121(3), 106793 (2021) 2. Sun, L., Tang, Y., Zuo, W.: Coronavirus pushes education online. Nat. Mater. 19(6), 1 (2020) 3. Ht, A., Xiang, Y.B., Hty, C.: Learning-related soft skills among online business students in higher education: grade level and managerial role differences in self-regulation, motivation, and social skill. Comput. Hum. Behav. 95(3), 179–186 (2019) 4. Trussell, H.J., Gumpertz, M.L.: Comparison of the effectiveness of online homework with handwritten homework in electrical and computer engineering classes. IEEE Trans. Educ. PP(99), 1–7 (2020) 5. Heyden, E., Küchenhof, J., Greve, E., et al.: Development of a design education platform for an interdisciplinary teaching concept. Procedia CIRP 91(3), 553–558 (2020) 6. Kaw, A., Clark, R., Delgado, E., et al.: Analyzing the use of adaptive learning in a flipped classroom for preclass learning. Comput. Appl. Eng. Educ. 27(3), 663–678 (2019) 7. Liang, C., Fan, R., Lu, W., et al.: Personalized recommendation based on CNN-LFM model. Comput. Simul. 37(3), 6–12 (2020) 8. Xu, Y.H., Jin, G.Q.: Artificial intelligence matching simulation of multi feature cascaded image database. Comput. Simul. 38(3), 437–441 (2021) 9. Liu, J., Sun, H., Guo, D., et al.: Design and implementation of online education platform based on spring MVC and mybatis framework. J. Shenyang Norm. Univ. (Nat. Sci. Ed.) 37(3), 268–273 (2019) 10. Peng, X.H., Li, K.L., Zhong, L.H., Liao, P.: Design of an online education platform for multi concurrent high-speed communication. Mod. Electron. Technol. 44(18), 92–96 (2021)

Intelligent Interactive Mobile Teaching Platform in Colleges and Universities Based on Artificial Intelligence Network Chaojun Zhu(B) Department of Judicial Information Management, Sichuan Vocational College of Judicial Police, Deyang 618000, Sichuan, China [email protected]

Abstract. Under the condition of today’s network technology, the realization of online teaching has become a reality and accepted by most colleges and universities. It has become a popular technology for online teaching platform to complete online teaching tasks. However, the traditional online teaching platform does not give enough consideration to the interaction between teachers and students. Under this background, an intelligent interactive mobile teaching platform in colleges and universities based on artificial intelligence network is designed. This paper analyzes the design objectives and key technologies of the platform, and designs a platform framework based on B/S three-tier architecture. With the enhanced 16bit MCU (MC9S12DG128) as the core, four hardware modules are designed. The platform database is designed with reference to SQL Server 2000 database technology, and five functional modules are designed. The test results show that the average delay of the platform is 1866 ms when there are 1000 concurrent users, and it has good application performance. Keywords: Artificial intelligence network · Platform frame · Hardware · Function module · Mobile teaching platform

1 Introduction With the advent of the era of knowledge economy, the traditional teaching mode has been unable to meet the needs of modern society for higher education, and the educational resources are wasted to a certain extent due to the limitation of time and space. Information technology, especially computer network and multimedia technology, has penetrated into all fields of modern society, and provided a broad development space for the optimization and sharing of educational resources. Under this background, modern distance education shows its great superiority and feasibility. At the same time, with the popularization of computers and the development of Internet technology, artificial intelligence network technology has developed rapidly, and its application in the field of education has become more and more extensive, and the way of educational information

© ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 145–157, 2022. https://doi.org/10.1007/978-3-031-21161-4_12

146

C. Zhu

dissemination has also changed, which has led to great changes in educational models, concepts and methods. The application of artificial intelligence network in the field of education not only provides students with plenty of learning opportunities, but also provides students with rich teaching resources, which makes learning activities more autonomous, breaks the limit of traditional teaching time and space, and accelerates the pace of updating teaching content and system. At present, network teaching is mainly conducted through virtual classroom and wireless network. In virtual classroom, the network can provide students with a man-machine interface with pictures, texts, audio and video, and provide a knowledge base that is more in line with students’ thought expansion and a large-scale information base organized by hypertext structure. Therefore, it is easy to stimulate students’ interest in learning. Network teaching can not be limited by time and space, as long as there are network conditions in any place, we can learn independently through the network. Using traditional computer-assisted instruction usually presets all the teaching contents in the system on a single computer according to the programming method. Although the computer-aided instruction system can teach online with its intelligence, the lack of necessary communication means leads to the inability of interaction between teachers and students. Some scholars have suggested that using network platform in PBL teaching of immunology can break through the limitations of time and space and teachers. Students can interact with tutors anytime, anywhere to find PBL case information, which is helpful for teaching [2]. Some scholars have proposed to use data mining method to extract the relationship between students and teachers in art teaching, and apply it to multidimensional analysis of data, which enhances the structural integrity of teaching system [3]. Therefore, an intelligent interactive mobile teaching platform for colleges and universities based on artificial intelligence network is designed.

2 Design of Intelligent Interactive Mobile Teaching Platform in Colleges and Universities With the development of computer and network technology, the space of education is no longer limited to traditional classrooms, and the form of education has also broken through the traditional teaching mode. The teaching platform system based on artificial intelligence network has the characteristics of unlimited time and space, various forms, flexibility and convenience, etc., and has been gradually applied to practical teaching activities as a brand-new modern education method, which has become a powerful supplement to traditional education forms. How to effectively carry out network teaching activities has gradually become an important topic in the construction of teaching informatization in higher vocational colleges, and the construction of network teaching platform has become the foundation and key to solve this problem. Based on the work practice of “Intelligent Interactive Mobile Teaching Platform for Colleges and Universities Based on Artificial Intelligence Network” designed and developed by the author, this paper discusses the design pattern, overall structure and system implementation scheme of the network teaching platform, and explores the application of educational informatization in teaching through the realization of examples.

Intelligent Interactive Mobile Teaching Platform

147

2.1 Platform Design Goal In order to know what kind of technical support the network teaching platform needs, we first need to know the design goal of the platform. The design of a network teaching platform should reflect the following objectives: make full use of the software and hardware resources of the existing network, apply artificial intelligence network, establish an interactive, open and easy-to-use environment, and facilitate the use and communication of three users. On this basis, the platform design must also meet the following objectives: • Content-oriented subjective goal when combining teaching content with technical means; • We should not only base ourselves on today’s advanced technology, but also consider the forward-looking goal of future technology development; • Enable the teaching of different disciplines and the use of different user groups to achieve a unified universal goal; • The economic goal of obtaining the most efficient teaching benefit with the lowest possible investment. 2.2 Key Technologies of Platform ASP Technology ASP (Microsoft Active Server Pages) is the abbreviation of active server page, and it is a scripting environment on the server side developed by Microsoft, which can be used to create dynamic WEB pages or generate powerful WEB applications. ASP can combine HTML pages, script commands and ActiveX components to create dynamic and interactive WEB pages and WEB-based applications. ASP is a WEB page technology on the server side, which runs on the server side (web server) rather than on the client side (visitor’s browser). When a customer requests an ASP file, the server first interprets the file as a standard HTML file and then sends it to the customer. There are two advantages to running on the server side: first, it can be free from the limitation of the client browser; Second, it is very convenient to exchange data with the data server. Strictly speaking, ASP is not a language, it just provides a running environment to run Script. The language it uses is still Vbscript or JavaScript, or it can be a combination of both. Database Technology Using SQL Server 2000 database technology as the DBMS of this system, it has high efficiency data analysis performance, flexible business expansibility, security of integration with operating system and ease of use of customers and management tools, which improves the efficiency of management and reduces the cost of system operation and maintenance. SQL Server 2000, as a database developed by Microsoft on Windows series platform, is a fully functional database management system. SQL Server 2000 also has the function of rapidly developing new Internet systems. Especially, it can directly store XML data and output search results in XML format, which is conducive to the interoperability of heterogeneous systems and lays the cornerstone of Internet-oriented enterprise

148

C. Zhu

applications and services [4]. In the case of using the relational database engine of SQL Server 2000, XML data can be stored in relational tables, and queries can return relevant results in XML format. In addition, XML support simplifies the integration of back-end systems and realizes seamless data transmission across firewalls. You can also access SQL Server 2000 by using HPE HperText Transfer Protocol (HTTP), so as to realize secure Web connection for SQL Server 2000 database and online analytical processing (OLAP) cube without extra programming. SQL Server 2000 adds OLAP online analytical processing function, which allows many users to use some features of data warehouse for analysis. Streaming Media Technology With the rapid development of the Internet age, streaming media technology is being widely used and gradually known by people. With the popularization of broadband network, people can even play movies or watch live programs online, and these applications can not be realized without the support of technology, and streaming media technology plays an important role in it [5]. At present, the streaming media technologies adopted in the market are mainly composed of two series: Real Media technology developed by the most representative Real Network company and Windows Media technology launched by Microsoft company, and a small amount of Ouick-Time technology of Apple company is also adopted. Compared with the traditional multimedia technology, they have the following common features in technology. 1) 2) 3) 4) 5)

Adopt audio and video encoder with high compression rate and high quality; Coding mode with multiple bit rates; Have intelligent flow control technology; Support script command transmission mode; Different from the WEB server mode, it makes up for the deficiency of Web server function;

2.3 Platform Frame Design At the end of 1990s, with the in-depth application of network technology in all walks of life, a new architecture, three-tier B/S network architecture, appeared in IT industry with low cost, low management overhead and the advantages of Client/Server computing mode. If the C/S architecture is called “thick client/server” computing mode, then the three-tier B/S architecture can be called “thin client/server” computing mode [6]. The technical feature of the three-tier (or multi-tier) B/S architecture is that one (or more) middleware layers are added to the two-tier architecture. It moves the application program originally running on the client in the C/S architecture to the middleware layer, and the client is only responsible for displaying the interface for interaction with users and a small amount of data processing (such as data legitimacy check). The client submits the collected information (requests) to the middleware server, which performs corresponding business processing (including database operation) and then feeds back the processing results to the client. When the number or demand of teaching content and clients changes greatly, it is easy for the system to be overloaded and its performance will be greatly

Intelligent Interactive Mobile Teaching Platform

149

reduced. The three-tier architecture can greatly improve the performance of the system, and the database server is only responsible for storing and managing data, which can reduce the amount of data transmission and the load of the network [7]. Especially, the three-tier structure system designed by technology can achieve interoperability and cross-platform operation. The whole teaching platform logically adopts a three-tier architecture, and the system framework is designed, as shown in Fig. 1.

student

teacher

administrators User layer

browser

Job module

Interactive module

Management module

Teaching application module

Data transmission and synchronization

Interactive control

Archive information base

Artificial intelligence network communication module

Communication equipment

Teaching resource database

Funct ional layer

Network commun ication layer

Data layer

Fig. 1. Platform framework

User Layer The user layer is equivalent to the window of the whole system. Users can directly access the system through this layer to realize the interaction with the system, thus completing the work that needs to be realized. In order to facilitate users to enter the system with different identities, the system is divided into three different user interfaces, namely, teacher interface, student interface and administrator interface. Different users have different permissions, and the permissions from low to high are students, teachers and administrators, as shown in Fig. 2 below. Students are faced with what they want to learn. Teachers enter the background to manage teaching contents through certification, and administrators have the highest permissions. After certification, they can manage all contents and system settings. The user layer adopts browser mode, and the interface is as friendly as possible for the convenience of users [8]. This layer is mainly implemented by Web dorms of ASP. NET. Functional Layer The function layer is the connecting part between the user layer and the data layer, but

150

C. Zhu Manage user information

Maintain database

Management system information

administrat or

Manage personal information

Student management

Managing teaching materials

Analyze the learning situation

Management announcement

teacher Manage ment marking

Manage jobs

Answer questions

Manage personal information

online learning

Download teaching resources

online exchange

Manage jobs

Query results

Online detection

student

Fig. 2. User management authority of teaching platform

it is not a simple connection, but a detailed classification of users’ needs. It consists of many modules, which can be stored on the Server side according to their different functions. The functional layer in this system includes three parts: teacher subsystem, student subsystem and administrator subsystem, and each part is composed of several modules. The function layer realizes fast data access to the data layer through ADO. NET. Communication Layer The communication layer is similar to the CLL layer in CSCL model. Its main function is to provide reliable communication connection for interactive teaching system and a unified cooperative communication link for the upper layer. This layer uses XML-based message format for transmission. Data Layer The data layer is the foundation of the whole system. It consists of three parts: user information database, teaching resource database and knowledge structure database. The user information database includes the basic information of students, teachers and other managers. The teaching resource database mainly provides data support for the teaching and auxiliary subsystems, including lesson plans, courseware, videos, homework, test questions and so on. Knowledge database mainly organizes information resources rationally. The data layer all uses stored procedures to operate the underlying data. 2.4 System Hardware This platform takes the enhanced 16-bit MCU (MC9S12DG128) as the core, and mainly consists of the following four modules. Each module is a relatively independent small

Intelligent Interactive Mobile Teaching Platform

151

practical experimental system, and the control function of each module is realized through the expansion of I/O interface. (1) The experimental control module realizes the control of stepping motor or DC motor. (2) The analog-to-digital conversion module of the system realizes the A/D conversion and D/A conversion functions. (3) The experimental clock module realizes the function of calendar clock stopwatch. (4) Routine experiment module and extended experiment module, realizing routine experiment and extended experiment of single chip microcomputer. MC9S12DG128 is an enhanced 16-bit single-chip microcomputer in S12 series single-chip microcomputer introduced by Freescale Company, which is rich in on-chip resources and widely used. It integrates 16-bit central processing unit HCS 12 CPU, 128-kbyte Flash EEPROM, 8-kbyte RAM, 2-kbyte EEPROM, 2 asynchronous serial interfaces SCI, 2 synchronous serial interfaces SPI, 8-channel enhanced capture timer (ECT) with IC/OC function, 2 8-channel 10-bit ADCs, 1 8-channel PWM and 1 BDLC module. Two CAN 2.OAB softwares are compatible with the CAN controller MSCAN, one Byteflight module, one IIC module and rich IO ports. MC9S 12DG128 has all 16-bit external data channels, and can run in 8-bit narrow mode, so that 8-bit wide memory modules can be used to reduce costs. In addition, MC9S12DG128 also contains PLL circuit, which allows adjustment of power consumption and performance to suit specific applications. MC9S12DG128 can run at the highest S OM crystal oscillator, that is, 25M bus speed, and has three low power consumption modes: stop, pseudo stop and wait. MC9S 12DG128 is available in 80-pin and 112-pin packages, which gives users great freedom in design. 2.5 Database Design To design an effective database, we must consider the problem from the viewpoint of system engineering. Because this website is an artificial intelligence network teaching platform, there are many kinds of course resources, so there are many corresponding data tables, including: article system data sheet; Download system data sheet (downloading system data sheet); Electronic lesson plan data sheet (electronic lesson plans data sheet); The FLASH system data sheet; Classified information sheet; Image data sheet (photo system data sheet); Exam Text, etc. Exam Text is a collection of test records, which is used to store detailed information of all test questions in various subjects. It includes the main test number (primary key), the subject number of the test questions, the type of questions, the difficulty of the test questions, the stem content of the test questions, the alternative answers of the test questions, the standard answers of the test questions, the entry time of the test questions, the storage path of the picture files and video files if the test questions contain pictures and video data, the answer times and the correct answers of the questions, which have nothing to do with the specific test papers.All the data tables stored in the database are classified and stored in the form of codes. Considering that the corresponding technologies are extensive, I will not repeat them here.

152

C. Zhu

2.6 Main Function Module Design Management Module Website management: user management and course management constitute the three main modules of the platform management module. Website management module: Users with management status operate the website management module and effectively manage the settings of the whole platform, making the whole platform more reasonable. User management module: learners log on to the platform as ordinary users, and use some modules and resources of the user management module platform to help them learn. Course management module: users with the status of course teachers can manage the course, which is beneficial to their management of the whole course. Moreover, teachers can fully control all the settings of the curriculum including other teachers through the curriculum management module. Teaching Application Module The main function of the teaching application module is to provide interactive teaching environment for teachers and students and facilitate teaching activities. In order to achieve the above goals, the teaching application module needs to provide a variety of different types of teaching application tools to realize multi-dimensional interaction with users. In order to make each teaching application independent and display in different areas of Activity, it is necessary to separate the view of teaching application from its specific functions. Teaching tools are designed based on the View control provided by Android application framework. The View control is the basic control provided by Android system for user interface, which represents a rectangular area on the display screen and is responsible for image drawing and event handling in this area. The View control provides the onDraw () method to redraw the area and the invalid () method to update the area. Teaching application tools are implemented based on View controls, and event listeners are also needed to capture the user’s operations on the teaching application tools and inform the controller of the user’s operations, so as to realize the response of the teaching application to the user’s operations. There are two kinds of user operations for teaching application: touching the View control and clicking the function button. Android application framework provides OnTouchListener interface to monitor touch events. Artificial Intelligence Network Communication Module In order to enter the virtual classroom of the network, accept the guidance of teachers from different places and cooperate with students from different places, users of the interactive teaching system must require them to use mobile terminals as computing devices, access the Internet and communicate with other users on the basis of the network model designed by the system. Considering that the application scenario of the teaching system belongs to small-scale teaching, the system adopts the C/S framework, and the star connection is used between teachers and students.

Intelligent Interactive Mobile Teaching Platform

153

The network module adopts TCP connection and XML message format. In order to facilitate receiving and sending messages and improve communication efficiency, the network communication module does not open a thread for each connection to realize communication, but uses Selcet mode of Socket to communicate. Android provides ServerSocketChannel and Selector to complete the non-blocking communication mode of Android terminal. The network module first registers the Selector Selector, and after monitoring is started, it enters the select loop. Every time there is a new Sele ctio nKey, it may come from the newly established connection with new readable data, or a new connection may be established. At this time, it is necessary to judge and handle it according to the event of Selectio nKey. If there is no new SelectionKey and no stop notice is received, the selection will continue. After the interactive teaching system is started, when students want to join a teacher’s class, they need to apply to the teacher server for connection. If the teacher server has started the service and is running normally, the teacher server will accept the student’s connection application and add the student to the current student list. When the first connection fails, the system will try again several times. If all attempts fail, the system will report to the user. During the teaching process, either the teacher or the student can terminate the communication. For the mobile terminal of the interactive teaching system, if the teacher terminates the class normally, the mobile terminal of the interactive teaching system will disconnect after receiving the disconnection message of the teacher. If the teacher disconnects abnormally, the interactive teaching system will close the current communication session and try to reconnect after detecting the disconnection. If many attempts fail, the system will report to the user. When the mobile terminal of the interactive teaching system actively terminates the connection, the mobile terminal of the interactive teaching system will send a notice of disconnection to the teacher server, and after receiving the reply from the teacher server, it will actively disconnect, and the teacher server will delete the student from the current student list. Homework Module As we all know, the main body of interactive teaching is students, and it is very important to establish rich homework modules to improve the teaching quality. In the system setting, front-line teachers can set the highest score of each course and the completion time of homework, and students apply for late homework, etc. based on the actual situation of the course. For students’ homework, the system also comes with a set of question bank suitable for students. Teachers can select questions from the question bank and then assign them to students to do. At the same time, teachers can limit the time of their own questions. When students log in to the system to do homework, they log in to the interactive teaching information system of colleges and universities, and then finish the specified questions within the specified time. Finally, the system will automatically judge the answer and return the completion of the test questions and the final score to the teacher. In addition to the automatic correction of homework by the interactive teaching information system in colleges and universities, after the teachers’ evaluation and examination papers are completed, the failed homework will be returned. After the students

154

C. Zhu

log in to the system, they will be prompted on the homepage of the system for the failed homework until the students confirm it. Interactive Evaluation Module In the teaching process, after students log in to the interactive teaching information system of colleges and universities, they first evaluate the documents specified by teachers. After submitting the evaluation correctly, teachers can open the evaluation content and then manage the evaluation of students. The flexible option setting of interactive teaching information system in colleges and universities can greatly improve the quality of teaching interaction. The system gives students the right to evaluate teaching. When the teaching process is over, students can evaluate teachers’ teaching activities. Teachers can only view the evaluation contents, but can’t view the evaluation sources. At the same time, teachers have the right to explain the evaluation contents.

3 System Test Through the above contents, an intelligent interactive mobile teaching platform in colleges and universities has been constructed. In order to verify the application performance of the platform, this paper adopts a comparative experiment to test the reliability of the platform in the laboratory LAN environment, taking the teaching achievement and the delay of concurrent user information transmission as experimental indicators. 3.1 Testing Environment The mobile terminal of the interactive teaching system is based on Android operating system, while the teacher terminal uses Windows operating system, and the test is carried out with multiple Android mobile terminals. The test is carried out in the LAN environment of the laboratory. It should be noted that the PC uses the wired network, while the mobile terminal uses the wireless network. 3.2 Functional Test With the support of big data, the teaching effects of traditional system and distance teaching system based on artificial intelligence network are compared and analyzed. On the basis of a total of 300, compare the average scores of the two kinds of distance education, as shown in Table 1.

Intelligent Interactive Mobile Teaching Platform

155

Table 1. Comparison of average scores Number of students

The designed teaching platform

Traditional teaching platform

100

263.25

221.21

200

254.21

201.52

300

235.69

198.65

400

222.57

188.44

500

217.52

178.42

From Table 1, it can be seen that with the growth of students, the application of the designed teaching platform for teaching, students’ achievements have been higher than the application effect of traditional teaching platform. Therefore, using the distance teaching platform based on artificial intelligence network is more conducive to teaching and improving students’ achievements. 3.3 Performance Test LoadRunner is a performance test software. By simulating the real user behavior, realtime monitoring of load, concurrency and performance and the finished test report, the possible bottlenecks of the system are analyzed. One of the most effective means of LoadRunner should be concurrency control, which can simulate the operation of thousands of users in the same business at the same time by setting in the console. Scenario description-user retrieval concurrency delay. Explanation: Under the condition of strict concurrency, whether the interactive teaching platform resource retrieval can support the number of concurrent users is 1000. Simulation scenario description: The number of users simulated by the tool is 200, 400, 600, 800 and 1000 respectively (all users operate concurrently), and each user initiates several retrieval processes to test the concurrent delay of user retrieval. The test results are shown in Table 2 below. Table 2. User retrieval concurrency delay subscriber number 200

Average delay/ms 536

Maximum delay/ms 1225

Minimum delay/ms

Average concurrency delay index

862

2.26

400

869

1562

862

3.11

600

1022

1658

827

3.52

800

1526

2284

1433

4.48

1000

1866

2144

1563

5.69

156

C. Zhu

Average delay calculation formula: n 

Y =

xi

i=1

n

(1)

where Y represents the average delay; xi represents the delay of the i rd user; n represents the number of users. Concurrent delay index: n    xmax − Y

Y =

i=1

xmax − xmin

(2)

where xmax represents the maximum delay; xmin represents the minimum delay. As can be seen from Table 2, the delay and delay index are both small, reaching the research goal.

4 Concluding Remarks To sum up, when the current information age marked by multimedia has become a powerful supporting technology for the revolutionary change of pedagogy, people pay more and more attention to the advantages of distance learning based on artificial intelligence network. The characteristics of distance education system such as complexity, concurrency and individualized teaching are all very suitable for artificial intelligence technology. Therefore, artificial intelligence teaching technology has broad application prospects in distance education. The use of this system has a certain learning ability of artificial intelligence. By applying artificial intelligence technology to teaching data analysis, by setting up a multi-level teaching platform and using multi-modules to coordinate the information interaction between teachers and students in teaching, the auxiliary education system can provide a convenient learning platform for students. In the design of distance education system, it is still in the primary stage of system design, and the real realization of the system still needs to be optimized in many aspects. I hope that through the design of this system, it can provide new ideas for the research and help the next research work. In the future design of distance learning system of artificial intelligence network, it can still be improved from the following three aspects: 1) The introduction of computer simulation technology into the system can provide technical support for the application of distance education system; 2) Set up a collaborative interactive learning environment for system simulation verification, so that the system is more suitable for the actual education mode; 3) Under a large number of different information environments, teachers should supervise students’ learning with the support of network data, and evaluate their learning results in time.

Intelligent Interactive Mobile Teaching Platform

157

References 1. Farias, G., Fabregas, E., Peralta, E., et al.: Development of an easy-to-use multi-agent platform for teaching mobile robotics. IEEE Access 1(7), 55885–55897 (2019) 2. Wang, S., Zhang, F., Gong, Q., et al.: Research on PBL teaching of immunology based on network teaching platform. Procedia Comput. Sci. 183(11), 750–753 (2021) 3. Gang, L., Fang, W.: Research on art innovation teaching platform based on data mining algorithm. Clust. Comput. 22(2), 13867–13872 (2019) 4. Ramos, D.B., Ramos, I., Gasparini, I., et al.: A new learning path model for e-learning systems. Int. J. Distance Educ. Technol. 19(2), 34–54 (2021) 5. Sun, J.: Research on resource allocation of vocal music teaching system based on mobile edge computing. Comput. Commun. 160(2), 342–350 (2020) 6. Yu, X.: Resource scheduling for piano teaching system of internet of things based on mobile edge computing. Comput. Commun. 158(99), 73–84 (2020) 7. Bhat, M., Zaman, M., Butt, M.: An intelligent prediction system for educational data mining based on ensemble and filtering approaches. Procedia Comput. Sci. 2(167), 1471–1483 (2020) 8. Min, Z.: Data-driven interactive network group intelligence-aware task allocation simulation. Comput. Simul. 38(01), 476–480 (2021)

Construction of Multimedia Online Education Platform Based on Fuzzy Neural Network Dan Yu(B) Media College, Hulunbuir University, Hulunbuir 021008, China [email protected]

Abstract. The current multimedia online education platform has low resource utilization and response rate. In order to effectively solve this problem, this paper will study the construction of multimedia online education platform based on fuzzy neural network. The hardware framework of online education platform is built with FPGA as the control core, and the platform resources are mined and processed by fuzzy neural network. Use cloud computing to design platform resource scheduling and improve platform operation efficiency. Through experiments, it is verified that the recommendation accuracy of the platform is higher than 90%, the response rate and processing speed of the platform are significantly improved, and the practical application effect is better. Keywords: Fuzzy neural network · Multi-media · Online education platform · Data mining · Resource scheduling · Cloud computing

1 Introduction In the information society where knowledge is updated rapidly, the aging speed of information and knowledge is unprecedented. In recent years, with the rapid development of Internet technology and the large-scale popularization of smart terminal devices such as smart phones and tablets, and mobile network resources such as 3G and 4G are no longer scarce, digital and mobile online learning methods have become more and more popular for people. Accepted, at the same time, more and more traditional education institutions and Internet companies are involved in the field of online education. The construction and use of online education platforms have promoted profound changes in the field of education. Universities, middle schools, and large educational institutions have established modern education information platforms such as digital book information systems and Internet education platforms. Learners can not only use traditional classroom teaching Learning in the form of learning, and as long as you have a networked computer or other handheld terminal, you can enter the network online platform to achieve the purpose of learning [1]. Online education has the characteristics of fragmented learning time, unlimited learning locations, and targeted learning content. Online education has the advantages of being able to choose learning content according to one’s own needs, and being able to watch and study repeatedly for many times. It not only affects the learning © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 158–169, 2022. https://doi.org/10.1007/978-3-031-21161-4_13

Construction of Multimedia Online Education Platform

159

effect of users, but also promotes the transformation of traditional education mode. The development of multimedia communication has always been the driving force of modern distance education systems. The role of an online education platform is indispensable. It not only needs to collect, edit, and store abundant learning resources, but it also needs to respond to massive user requests efficiently in real time. However, there is still a big gap between the development of modern distance education in my country and the distance education in developed countries. Generally speaking, my country’s distance education platforms generally lack innovation. Reference [2] analyzes the current situation and difficulties of cyberspace security offense and defense practice teaching, designs a virtual and real cyberspace security offense and defense platform that integrates traditional information security and industrial control system security, and gives a specific construction plan. The experimental platform can meet the cyberspace security practice training needs of professional postgraduates, and has strong practicability and scalability. Reference [3] aims at the defect that the online teaching platform based on the ADDIE model of classical teaching design theory fails to reveal the complexity of online practical teaching activities, and uses advanced ubiquitous technology to build an ubiquitous practical teaching activity model based on activity theory and situational perception. Based on the Internet of things and sensor technology, the state evolution diagram method of experimental process is adopted to create an ubiquitous practical teaching platform with the ability of dynamic generative data acquisition in experimental process. Through reasonable teaching activity design and practical evaluation, it is proved that the online platform can complete the teaching task, avoid the result oriented experimental evaluation, and effectively improve the online teaching effect of practical courses. However, the recommendation accuracy of this platform is low, and the practical application effect is not good. Reference [4] summarizes the problems of users’ online learning experience on online learning platforms by analyzing the current status of multimedia online education platforms and online learning behavior characteristics of online learning users, combining quantitative and qualitative analysis of users’ questionnaires, interviews and surveys, and according to the user journey map of online learners. Then combined with the FBM behavior model proposed by Stanford University behavioral science professor Fogg, the interactive design method of multimedia online education platform under the FBM behavior model is obtained from the three aspects of enhancing user motivation, improving user ability and increasing trigger mechanism. However, the response rate of the platform is low after its application. Reference [5] focuses on the key technologies of Android development, database design and personalized recommendation algorithm based on the design and analysis of the overall architecture, functions and execution process of the multimedia online education platform using Android platform and C/S structure, combined with SQLite database, this paper realizes the construction of multimedia online education platform. However, the platform has the problem of low processing speed, and the actual application effect is not good. Neural network has parallel processing ability and strong fault tolerance. However, there are obvious deficiencies in the expression of knowledge and the interpretation of the rules learned. Fuzzy logic and neural network are related and complementary in many aspects. Therefore, fuzzy logic system and artificial neural network are combined

160

D. Yu

to learn from each other. The combination of fuzzy logic and neural network produces fuzzy neural network. It has the advantages of fuzzy logic system and neural network, has the function of universal approximator, has clear physical meaning, fast convergence speed and higher processing efficiency. Therefore, according to the above analysis, this paper will build a multimedia online education platform based on fuzzy neural network.

2 Construction of the Hardware Part of the Multimedia Online Education Platform When put into large-scale actual use, there will be higher hardware requirements, and the software must be supported by the hardware, and the combination of software and hardware can complete various functions, so the hardware part must also be considered. The material foundation supporting the multimedia online education platform is an actual computer network, which is similar to the current network center structure in schools at all levels. The functional modules of the software are placed on the public network. And as a general-purpose online education platform, depending on the number of students to be supported, the scope, the way students access the Internet and many other factors, its hardware structure will change greatly, and the organizational structure may also be different, but generally speaking, they all have the following functional modules: access module, exchange module, server module, two-way interactive synchronous teaching module, courseware development module, etc. According to the implementation requirements of the functional modules, the overall framework of the hardware part of the multimedia online education platform designed in this paper is shown in Fig. 1 below. Power supply module

Memory

Flash DDR

Upper machine

Ethernet interface FPGA XC65LC85 Interface circuit

Fig. 1. The hardware framework of the multimedia online education platform

The hardware core controller of the multimedia online education platform designed above is FPGA chip. The structure of FPGA is flexible. Its logic unit, programmable internal wiring and I/O unit can be programmed by users, which can realize any logic function and meet various design requirements; It has the advantages of high speed, low power consumption and strong universality, and is especially suitable for the design of

Construction of Multimedia Online Education Platform

161

complex systems; FPGA can also realize dynamic configuration, online system reconfiguration, hardware softening, software hardening and other functions; At the same time, FPGA can allow the function of the circuit to be changed as needed at different times of system operation, so that the system has a variety of space related or time-related tasks [6]. The platform communicates with the host computer of the platform through Ethernet interface or JTAG structure. The design content of the hardware part will be described in detail below. 2.1 Hardware Core Control Module Design The core control circuit board uses a 50 MHz single-ended CMOS crystal oscillator, and the clock signal is connected to the global clock pin, which can be multiplied to a higher operating frequency through the DCM or PLL in the clock management module; the memory system includes 1 GB of DDR2 SDRAM Chip, 128 Mbit SPI Flash chip, 16 Mbit Nor Flash and 128 Mbit Nand Flash. Among them, SPI Flash is used as the configuration memory when FPGA is powered on and used to store FPGA configuration files; DDR2 SDRAM is mainly used to execute the code of the main program when running the embedded operating system; Nand/Nor Flash is used to store embedded operations System programs such as the kernel, drivers, and file system of the system. The interfaces on the core control board are mainly Ethernet interfaces based on the DP83848 network card chip, and USB based on the FT2232H chip. JTAGAJART interface. According to the FPGA chip configuration mode scheme in Table 1 below, select the hardware FPGA configuration mode pins that meet the education platform. Table 1. FPGA chip configuration mode selection scheme Pin P1 level

Pin P2 level

Composite clock

Output bus width

Configuration mode

0

0

FPGA output

8/16

Main Select MAP configuration mode

0

1

FPGA output

1/2/4

Main serial port/SPI configuration mode

FPGA input

8/16

Configure mode from Select MAP

1

FPGA input

1

Configure mode from serial port

1

TCK input

1

JTAG test configuration

1 1

—

This platform mainly adopts two configuration methods, JTAG and SPI Flash. The combination of these two methods can not only ensure the characteristics of the “soft” and hardware design of the experimental platform, but also ensure the ease of operation in the experimental link. Configure the FPGA through SPI Flash. SPI Flash is a non-volatile

162

D. Yu

storage device. After the configuration mode of the FPGA chip is set, the configuration file will be automatically loaded from SPI Flash after the system is powered on. 2.2 Serial Port Module Design The circuit uses STM32 serial port to supply power through MAX232 level conversion chip. MAX232 is a charge pump chip that can complete two TTL/RS232 level conversions. Two serial ports are used in the system. USART1 is used to collect the serial data of the device, and USART2 is used as the operating system terminal output, which is convenient for viewing and debugging internal programs. The running situation. This platform uses the USART multiplexed I/O ports PA9 and PC10 as serial port sending pins, configured as push-pull output with a speed of 50 MHz; USART multiplexed I/O ports PA10 and PC11 are used as serial port receiving pins, configured as floating input. In addition, the asynchronous communication interface of FPGA supports the connection of 8- and 16-bit wide asynchronous memory devices, such as NOR Flash, NAND Flash and dual-port RAM. It has two main operating modes: WE trigger and selective trigger. The difference between the two trigger modes is listed in Table 2 [7]. Table 2. Trigger mode of asynchronous communication interface Trigger mode

Trigger function

Asynchronous access operation

WE trigger mode

Write trigger

Always active during asynchronous access

Select trigger mode

Byte enable

Only activated during the trigger phase of a visit

Using the address latch of CPLD, expand 7 high address lines AA11–AA17 through I/O port to connect with A11–A17 of FLASH respectively, thereby defining refresh rate, CAS delay and many SDRAM timing parameters. 2.3 Storage Module Design The external storage module, that is, off-chip SRAM, uses IS61LV25616 SRAM chip from ISSI. Its storage capacity is 16 * 256K, and it has high and low selection signals. Features are as follows: access time is 10 ns, 12 ns, supports tri-state output, fully static operation does not require clock or refresh, interface level is compatible with TTL standard, independent 3.3 V power supply, high byte data and low byte data can be controlled separately. This module is mainly used to buffer the collected data. Since the system communication cycle is one second, it only needs to buffer the data for two seconds. The size and access speed of the SRAM meet the performance requirements of the platform. On the above platform hardware, the fuzzy neural network is used to design the platform software part to complete the construction of the overall multimedia online education platform [8].

Construction of Multimedia Online Education Platform

163

3 Partial Construction of Multimedia Online Education Platform Software Based on Fuzzy Neural Network 3.1 Fuzzy Neural Network Mining Multimedia Online Education Resources Fuzzy neural network is the product of the combination of fuzzy theory and neural network. It combines the advantages of neural network and fuzzy theory, and integrates learning, association, recognition and information processing. In this paper, the network is applied to the process of multimedia online education resource mining. The input is a large education database resource, which is obtained by crawler technology, and the input is the mining result of multimedia online education resources. The performance of the fuzzy neural network is improved by using the fuzzy inference layer. The fuzzy neural network designed in this paper consists of a antecedent network and a consequent network. The antecedent network is used to match the antecedent of fuzzy rules, and the consequent network is used to generate the consequent of fuzzy rules [9]. The front network consists of three layers, of which the first layer is the input layer; In the second layer, each node represents a language variable, which is used to calculate the membership degree of the input component belonging to the fuzzy set of each language variable; The third layer is used to normalize the excitation and calculate the activation degree of each rule. The latter network is composed of parallel subnetworks with the same structure, and each subnetwork produces an output. The consequent network is generally divided into two layers. The first layer calculates the consequent of each rule, and the second layer calculates the network output, using the back-propagation learning algorithm. The adjustable parameters in fuzzy neural network are mainly concentrated in the first layer and the fifth layer. The parameters of the first layer are the membership function parameters, and the parameters of the fifth layer are the subsequent parameters of the rule, that is, the parameters of the part of the if-then rule. The functions of each layer are described in detail below. The first layer of the fuzzy neural network converts the input vector into the membership degree of the corresponding fuzzy set, and the node i in the layer is an adaptive node composed of membership functions [10]: Outi1 = uQi (x), i = 1, 2 Outi1 = uPi−2 (y), i = 3, 4

(1)

In the above formula, Outi1 is the degree of membership of fuzzy set F = {Q1 , Q2 , P1 , P2 }, which determines the degree of a given input or satisfaction. The membership function of neural network in this research is Gaussian function. The second layer of the network is a paradigm operator that performs fuzzy AND on the input variables. Outi2 = uQi (x)uPj (y), i = 1, 2; j = 3, 4

(2)

The third layer performs normalized excitation processing on the input, and the output of the third layer is the ratio of the excitation intensity of the first and second rules to the sum of the excitation intensities of all rules calculated for the i node. Outi3 = ui (u1 + u2 )−1 , i = 1, 2

(3)

164

D. Yu

The fourth layer is the output of calculating fuzzy rules, where ui is the standardized excitation intensity output by the third layer. Outi4 = ui (ai x + bi y + ci ), i = 1, 2

(4)

Among them, (ai , bi , ci ) is the parameter set of the fuzzy rule consequence, which is called the rule consequence parameter. The fifth layer sums up each output to get the total output. The fuzzy neural network uses a combination of backpropagation algorithm and least square method for learning and training. Figure 2 below is the training flowchart of the fuzzy neural network. Start

Structure identification

Determine input and output

Input space partition Membership function calculation

Parameter identification

Training fuzzy neural network

Minimum output error?

N

Y Network simulation

End

Fig. 2. Flow chart of fuzzy neural network training

Among them, structure identification is to set the structure of the network, mainly determine the input and output variables of the model from the following aspects, obtain the optimal combination of input and output variables, and determine the division of input and output space, the number of rules, the number of membership functions and the initial parameters of membership functions. Parameter identification refers to the identification of a group of parameters under the determined structure, adjusting each parameter in the model to obtain the optimal model parameters of the system. In the process of parameter identification, it mainly depends on network training to judge the training error. The parameters that meet the minimum training error are used to mine multimedia educational resources.

Construction of Multimedia Online Education Platform

165

Firstly, the resource extraction of statistical analysis is carried out for the large database of multimedia educational resources. Make statistics on the preferences of different educational objects for educational resources, and conduct generalization operations to form a data warehouse. After data cleaning, the multimedia education resource data is integrated, and the mapping method is used to transform the non numerical data into numerical data. After the maximum and minimum normalization method is used to normalize the input data, the multimedia education resources are mined by using the fuzzy neural network established above, and stored in the education platform database according to the established data labels. 3.2 Platform Resource Scheduling Model Design The users of multimedia teaching platform include teachers and students of the school and visitors outside the school. In some time periods, the amount of concurrent visits is huge. In order to improve the service performance of the platform, cloud technology is used for platform education resource scheduling. Based on the idea of cloud computing and computer resource virtualization, the cluster server module of the platform is designed, and its structure is shown in Fig. 3.

Client layer

Control layer Virtual cluster

Cluster Resource scheduling & Virtual monitoring management allocation machine

Fig. 3. Virtual resource service cluster architecture diagram

The virtual resource service cluster is divided into three levels from top to bottom: consumption layer, control layer and virtual cluster layer. The consumption layer refers to various client resource access requests forwarded by the Web master server; The control layer consists of two parts: one is the basic resource data distribution server, which realizes the multimedia teaching function of the system; the other is the load balancing control server, which realizes the scheduling and task allocation of the data distribution server according to the load balancing algorithm; The virtual cluster layer is based on the virtualization technology in cloud computing, combined with high-speed resource transmission technology, high-capacity storage technology, dynamic scheduling technology, etc., uses various existing resources to build a large number of schedulable virtual machine resources, and provides resource support for multimedia teaching according to needs [11].

166

D. Yu

Considering the indicators of its parallel operation; In the process of resource service operation, not all resources are called evenly. In many cases, most teaching on demand focus on a small number of resource files. Therefore, in the case of limited service resources, if most resources are allocated to more intensive service requests, the service efficiency will be improved on the whole. Predict the virtual CPU utilization according to the following formula: −1 ηvcpu (t) = 1 − tf · tto

(5)

Among them, tf is the idle time of the education platform; tto is the total time of the overall platform for scheduling resources. Using the collage theorem and separate interpolation method to analyze and process the above formula, the general prediction formula can be obtained as follows: ηvcpu (t) =

t−1 

Ci (D)

(6)

i=1

Among them, Ci is the affine transformation obtained by the principle of statistics; D is the collection of all platform resource scheduling data on the time axis. The impact of resource data measurement and statistics in a small time interval on the working state change of the CPU itself and the process state change caused by the platform scheduling tasks based on the process queue. In order to eliminate these effects, the virtual CPU occupancy can be calculated by weighted average, as shown in the following formula:  η¯ vcpu (t) = (1 − λ)η¯ vcpu (t − 1) + ληvcpu (t) (7) ηvcpu (0) = 0 In the above formula, λ is the weight, and its value range is [0, 1], which is used to control the proportion of the historical virtual CPU occupancy rate value in the current occupancy rate prediction. The value of λ can be dynamically adjusted according to actual needs. When the virtual node is running, the tasks of the virtual central processing unit need to be allocated to each thread of the actual physical central processing unit to run, thereby realizing efficient scheduling of platform resources. Combining the above design content of platform hardware and software, completed the research on the construction of a multimedia online education platform based on fuzzy neural network.

4 Experimental Research A multimedia online education platform based on fuzzy neural network is constructed above, and the feasibility and application of this online education platform will be studied in an experimental manner.

Construction of Multimedia Online Education Platform

167

4.1 Experiment Content When CPU is The 10 core Intel Xeon e5−2640 CPU, the operating system is tested in the window 10 computer, and the simulation software is Matlab 7.0. Conduct comparative experiments on multimedia online education platforms to obtain more objective experimental results. The multimedia online education platform based on fuzzy neural network constructed in this paper and the online education platform constructed by ADDIE model are compared with the three indicators of response rate, transmission rate and personalized resource recommendation accuracy rate of the platform, so as to comprehensively analyze the performance of the education platform. 4.2 Experimental Results Figure 4 shows the comparison of personalized resource recommendation accuracy of the platform. Accuracy of recommendation/%

100 90 80

70 60 50

Platform of this paper Contrast platform

40 30

0

1

2 3 4 5 Number of service object types

6

7

Fig. 4. Accuracy rate of personalized resource recommendation

Analyzing Fig. 4, it can be seen that with the continuous enrichment of platform service object types, the resource recommendation accuracy rate of the comparison platform has decreased, and the recommendation accuracy rate of the platform in this article is still higher than 90%. Table 3 below shows the test results of the response rate and transmission rate tests on the two educational platforms. Among them, the change of experimental variables is controlled by changing the service request accepted by the platform.

168

D. Yu Table 3. Comparison of platform response rate and transmission rate

Service request volume

This article platform

Comparison platform

Response time/ms

Transmission time/ms

Response time/ms

100

306.4

98.8

518.2

119.8

150

317.9

98.5

541.5

114.3

200

328.6

99.1

603.2

117.9

250

338.3

98.9

633.6

117.5

300

366.5

99.3

643.7

120.8

350

370.9

97.8

665.3

116.6

400

504.1

98.6

865.1

119.2

450

530.7

98.9

898.5

114.5

500

665.9

99.2

942.4

127.9

600

689.6

98.7

972.9

128.9

700

845.9

99.4

1053.8

128.4

800

859.2

99.0

1069.1

137.7

900

901.5

99.4

1301.4

144.6

1000

979.8

98.1

1378.6

145.5

Average value

571.8

98.8

863.4

125.3

Transmission time/ms

Analysis of the data in Table 3 above shows that with the increase in service requests, the response time of the two education platforms has shown an increasing trend, but the response time of the education platform in this article is far less than that of the comparison platform. Calculate the average service request response time of the two platforms in this experiment. The average response time of the platform in this paper is 571.8 ms, and the average response time of the comparison platform is 863.4 ms. The response time of the platform in this paper is shortened by about 33%. Analyzing the transmission rate of the platform, the transmission time consumption of the platform in this article has remained relatively stable, while the comparison platform has been increasing. Therefore, it can be determined that the education platform constructed in this article has a higher response rate and transmission rate. In summary, the fuzzy neural network-based multimedia online education platform constructed using the research in this article has the advantages of responsiveness, fast transmission speed and accurate resource recommendation.

5 Conclusion The online education platform will cover a variety of terminals with a variety of emerging communication channels. It combines multimedia technology, computer technology and communication technology to transmit audio, video, graphics, animation, text and

Construction of Multimedia Online Education Platform

169

other media information to different places. Modern distance education has rich teaching resources, flexible teaching methods, and is not limited by time and space, so that students can quickly obtain knowledge in a relaxed and happy atmosphere. The continuous enrichment of multimedia teaching technology promotes the development of education and teaching. In order to meet the needs of multimedia education, this paper studies and constructs a multimedia online education platform based on fuzzy neural network. FPGA is used as the control core to build the hardware framework of the online education platform, and fuzzy neural network is used to mine the platform resources. Use cloud computing to design platform resource scheduling and improve platform operation efficiency. Through experiments, the educational platform is tested, and it is verified that the platform has the advantages of high transmission rate, high response rate and good resource provision, and can meet the needs of multimedia education. However, there are too few indicators verified during the experiment, which leads to the decline of the reliability of the multimedia online education platform. Therefore, in the future, it is necessary to use more indicators to verify the performance of the multimedia online education platform, and introduce advanced technologies to optimize the platform, in order to improve the effect of multimedia online education and promote the further improvement of modern education level.

References 1. Wang, J., Zheng, H., Li, W., et al.: Online education governance in the era of intelligence: connotation, dilemma and breakthrough. E-educ. Res. 42(07), 54–60 (2021) 2. Zhu, C., Sun, B., Jin, X., et al.: Construction of the experimental teaching and training platform for cyberspace security attack and defense. Res. Explor. Lab. 40(06), 265–267+275 (2021) 3. Wu, Z., Zhong, C., Yin, J., et al.: The construction and evaluation of the ubiquitous practical teaching platform based on activity theory. Res. High. Educ. Eng. (06), 194–200 (2020) 4. Tan, Z., Jiang, X.: Research on interaction design of online learning platform based on FBM behavior model. Packag. Eng. 41(4), 189–194 (2020) 5. Chen, X.J., Song, C.J.: Research and design of online learning platform based on mobile Internet. Microcomput. Appl. 35(1), 35–38 (2019) 6. Chen, J.: Research on the application of online education platform support technology in mobile terminal. China Educ. Technol. (08), 118–122 (2017) 7. Fang, J., Shi, Z., Liu, L.: Migrating to 5g online education platforms: an empirical study on the influence mechanism of leaners’ behavior. Mod. Distance Educ. Res. 31(06), 22–31 (2019) 8. Zhang, Z., Chen, J.: Construction practice of cloud experiment teaching system for information security specialty based on cloud platform. Exp. Technol. Manag. 38(09), 251–255 (2021) 9. Jiao, K., Xu, K.: Research on MPPT of photovoltaic power generation based on neural network and fuzzy control. Comput. Simul. 37(5), 71–76 (2020) 10. Zhang, J., Chen, Z., Cao, Z., et al.: Evaluation of teachers’ online learning platform based on user experience. Mod. Educ. Technol. 31(10), 59–67 (2021) 11. Li, X., Liang, H., Feng, J., et al.: Design of personalized learning resource recommendation system in online education platform. Comput. Technol. Dev. 31(2), 143–149 (2021)

Design and Implementation of Mobile Intelligent Education System Based on Cloud Architecture Dan Yu(B) Media College, Hulunbuir University, Hulunbuir 021008, China [email protected]

Abstract. Because there are many intelligent education features, and the current mobile intelligent education system uses relatively backward education feature extraction technology, the extracted education features are fuzzy, resulting in excessive CPU utilization during the operation of the system. In order to solve this problem, a mobile intelligent education system based on cloud architecture is designed. Hardware part: adopt the active serial FPGA configuration method, load the program that controls the FPGA chip; Software part: use mobile internet technology to identify the type of mobile learning, optimize the traditional format, obtain intelligent education features under the support of cloud architecture, and calculate service resources. The distance to the user’s location is used to build a course resource management model and improve the intelligence of the education system. The experimental results show that the average CPU occupancy rates of the designed system and the other two systems are 20.169%, 30.087%, and 29.987%, respectively, indicating that the performance of the mobile intelligent education system integrated with the cloud architecture is better. Keywords: Cloud architecture · Mobile Internet technology · Network media · Online education · Mobile communication · Learning resources

1 Introduction Mobile intelligent education refers to an educational model in which learners use mobile devices and mobile Internet technology to learn and communicate at different times and in different places [1]. Moreover, the business combination mode on the client side and the backend server side has also developed rapidly, and some combination modes with high business scalability have appeared, such as MVC, MVP, etc. At the same time, some server distributed frameworks that optimize the system structure also appear, such as Dubbo. The goal of mobile education is to provide learners with assimilation-competent learning anywhere and at any time. In the information age where the mobile Internet dominates the development of science and technology, new knowledge and new things are changing with each passing day, providing smarter and more convenient services for the fast-paced life. The emergence and development of these technologies make it possible to build a high-performance system platform. Mobile learning is an extension © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 170–184, 2022. https://doi.org/10.1007/978-3-031-21161-4_14

Design and Implementation of Mobile Intelligent Education System

171

of digital learning, with digital learning environment, digital learning resources and digital learning methods. The characteristic of mobile education is that it can adapt to and support more and more floating population, and at the same time has great compatibility with mobile terminals. Since the system platform can be based on content, it can be applied to different groups of people according to different content, and users can customize the content they are interested in, so the application and commercialization prospects are very considerable. Lifelong learning has become necessary and possible, and the development of mobile Internet provides a strong technical guarantee for lifelong learning. In the near future, education will move from schools to communities, families, and poor and backward places, changing the predicament of unbalanced educational resources. In the mobile online education environment, the user’s request and access to services are a dynamic change process. One is the dynamic growth of the number of users, which requires the system to be capable of capacity expansion and business expansion. The Internet makes people do not need to sit in the classroom for face-to-face education, and the mobile Internet allows people to learn even on the move, so that they can learn anytime, anywhere. E-learning will become an integral part of people’s daily life. In addition, the most important thing is that the time period and frequency of users’ access to mobile educational resources also change dynamically. In order to deal with sudden and resource shortages, it is necessary to make dynamic adjustments, which are unchanged from the user’s point of view, that is, users cannot see the process of dynamic adjustment, and they have been enjoying smooth and stable services. In order to realize the real “Anyone, Anytime, Anywhere, Anystyle” in the field of education, a research direction based on mobile communication and the Internet came into being, called mobile education. Mobile online education has developed rapidly in recent years with the advancement of technology and social concerns. Mobile education covers learning concepts and mobile Internet technology. It is a form of learning that meets students’ learning needs with the help of various mobile terminal devices and network media. Learning devices include, but are not limited to, PDAs, MP3 players, laptops, mobile phones and tablets. The mobile education system provides users with multimedia information (including text, pictures, audio and video, etc.) education services, and realizes a variety of functions, such as login authentication, browsing information, streaming media on demand, interactive live broadcast, registration, payment, sharing, community discussion etc. Mobile education has the remarkable characteristics of small size, large class capacity and strong randomness. At the same time, it has digital teaching resources, and the implementation method is also digital. Yu Shengquan et al. believe that education informatization should be transformed from focusing on platform construction and resource construction to focusing on service construction. Based on the service architecture of cloud network fusion, the collaborative construction method is adopted to design the service architecture of cloud network fusion. The system has excellent operation performance, but ignores the fuzzy characteristics of intelligent education, resulting in high CPU usage [2]. Jing Chunhui et al., based on the channel data such as smart wristband, camera and score system, conducted data optimization and split processing, and determined the user roles of all stakeholders by using the user experience role construction method, so as to improve the quality of education. In practical application, it is found that the system can effectively improve

172

D. Yu

students’ academic performance, but the problem of fuzzy characteristics of intelligent education is not described in detail, resulting in the CPU occupancy rate is difficult to meet the application requirements [3]. Li Qian designs a network distance education system based on ASP under the background of MOOC. The hardware design part of the system classifies the hardware components and divides them into three modules according to the classification standards for specific research: In the communication module of the system, the network remote monitoring camera is selected for data image monitoring to ensure the integrity of the image collection and improve the data processing performance of the system; In the transmission module, LX-VGA-3UVA optical fiber transmitter is selected to transmit image data to ensure the quality of data transmission and realize efficient data arrangement. In the receiving module, the cloud management fiber transceiver JRGT-1002M is used to receive data signals, improve the system operation, reduce the operation time, improve the operation efficiency, and achieve the purpose of system hardware design. The proposed system software design mainly solves the application problems existing in the system software, constantly integrates the design basis of hardware components, reduces the contradiction between the data system software and hardware, promotes the development of the data system, and realizes the overall operation of the system software. However, in practical application, it is found that the response speed of this system is fast, but the CPU usage is too high, and the practical application effect is not good [4]. Because more intelligent education characteristics, and the current mobile intelligent education system used by the education after feature extraction technology is more, lead to the extracted fuzzy education characteristics, lead to excessive CPU usage during the process of system operation, in order to solve the problem, in the cloud architecture to support intelligent education characteristics, combining feature extraction result to calculate the distance from the center of the service resources to the user, In this way, the curriculum resource management model is constructed to improve the intelligence degree of the education system.

2 Hardware Design of Mobile Intelligent Education System In this design, the active serial FPGA configuration method is adopted. The serial configurator is used to store the configuration information of the FPGA. After power-on, the FPGA automatically loads the data to reconfigure. Therefore, the process of updating the FPGA program in this system is the process of writing data to the serial configuration chip. The main function of the power module is to provide stable power for the entire hardware system and manage the power accordingly. Since the input power is 9 V and the power required by the hardware system is 5 V and 3.3 V, the core function of the power module is a step-down and filtering process. The update of all FPGA programs in this system is controlled by ATmega128 A, and the nCE and nCONFIG pins of FPGA are connected with PD6 and PD7 pins of ATmega 128 A, which are used to control the program loading of the FPGA chip. The most commonly used step-down methods are: transformer step-down, voltage-stabilizer chip step-down, etc. Since the volume of the transformer is too large to be applied in this project, this paper adopts the step-down method of voltage-stabilizer chip. The chip communication mainly includes four pins DATA, ASDI, nCS, DCLK. Like the MCU system design, the download and debugging

Design and Implementation of Mobile Intelligent Education System

173

interfaces are reserved for use in the development stage. The voltage regulator chip is very small in size, flexible and convenient to use, and is very suitable for providing regulated power supply for embedded systems. The circuit switch architecture is shown in Fig. 1:

L2

L1

S

Fig. 1. Circuit switch architecture diagram

As can be seen from Fig. 1, L1 and L2 are controlled by the main switch S as two branch circuits. For different application environments, this system selects two video conferencing systems based on H.320 protocol and H.323 protocol. The difference between the two is that H.320 protocol is usually used as the standard of private line video conference TV. The transmission channel can choose E1, DDN and ISD interface according to the needs, and can use the dedicated line mode or dial-up mode 0 when networking. The H.323 protocol is defined for the IP network, which uses TCP/IP, RTP/RTCP and RSTP, etc. protocol to support real-time encoding and transmission of video, audio, and data over packet-switched networks. The 5 V power supply voltage regulator chip uses 78M05, its input voltage range is 7 V to 20 V, the output voltage is SV, and the maximum current it can provide is 350 mA. These characteristics fully meet the design requirements of the hardware system. Two kinds of interfaces are reserved in the system design of this section: JTAG interface and AS download interface. JTAG is a standard interface for chip testing commonly used in the industry. The voltage regulator chip of the 3.3 V power supply is AMS 1117-3.3, its input voltage range is 4.5 V to 12 V, the output voltage is 3.3 V, and the maximum current it can provide is 1 A. These characteristics also meet the design requirements of the hardware system. ATLERA’s FPGAs basically support JTAG commands to download FPGA programs, and the JTAG configuration method has a higher priority number than any other method. Among them, the JTAG interface has four signal pins which are essential. Because the DSP chip has special requirements for the power supply, when selecting the power supply for the DSP, the chip TPS767D318, which is specially used for DSP power supply produced by TI, is selected as the power supply chip. The power supply method used by the hardware control platform is internal power supply, and a 7.4 V lithium battery is selected, which has large capacity and strong driving ability. Considering the system voltage situation comprehensively, it is found that there are 5 V and 3.3 V in the system use. The main function of the AD module is to convert the analog voltage into a digital quantity and provide it to the DSP for processing. The schematic diagram of the reset circuit is shown in Fig. 2:

174

D. Yu

RESET R1

VCC

GND I5

Fig. 2. Schematic diagram of reset circuit

As can be seen from Fig. 2, the communication pins of the serial configurator are respectively connected with the SPI function pins PBO-PB3 of ATmega128 A for writing configuration information. In this paper, the information provided by the sensor is an analog voltage, and then the AD module converts the collected voltage into a digital quantity that DSP can process. There are three working voltages of 1.5 V, of which SV is the main working voltage of the system, and 3.3 V and 1.5 V are the voltages required by some devices respectively. When the switch is pressed, the battery interface provides 7.4 V power, and the AMS 1117-5.0 is used to convert the voltage to SV, and then the AMS 1117-3.0 and AMS1117-1.5 are used to convert the SV to 3.3 V and 1.5 V, respectively. MS320C2812DSP itself integrates a 12-bit AD conversion module, but the range of this AD module can only be 0 V to 3.3 V, which cannot meet the needs of sensor signals from 0 V to 5 V, so this article chooses an external AD sampling module. Note that the output end of the AMS1117 chip needs to be connected to a single capacitor of at least 10 uF during use, thereby improving its transient response and stability of power consumption. Based on the above description, the steps of hardware design of the mobile intelligent education system are completed.

3 Software Design of Mobile Intelligent Education System 3.1 Identify the Type of Mobile Learning Mobile learning means that learners obtain learning resources through wireless and mobile devices (such as mobile phones, PDAs with wireless communication modules, etc.) and wireless communication networks at any time and anywhere they need to learn, communicate with others and learn. Mobile learning is developed on the basis of digital learning. It is an extension of digital learning. It is different from general learning. Mobile learning is the product of the combination of mobile computing technology and digital learning. It brings learners a new feeling of learning anytime and anywhere. It is an indispensable learning mode in the future [5]. Mobile learning is nothing new, because in traditional learning, printed textbooks can also well support learners to learn anytime and anywhere. It can be said that textbooks have become a tool to support mobile learning long ago, and mobile learning has always been around us. Mobile learning organically combines mobile communication, network technology and education. Compared with wired online learning, mobile learning has more characteristics, as shown in Fig. 3:

Design and Implementation of Mobile Intelligent Education System Mobility

175

Efficiency

Mobile learning characteristics

Universality

Individuality

Fig. 3. Features of mobile learning

As can be seen from Fig. 3, mobile learning has the characteristics of mobility, efficiency, universality, personalization and so on. It can be seen that as a new thing and new concept, mobile learning must be distinguished from traditional learning, otherwise it will lose its significance. Mobile Internet brings information exchange and service means of “anytime, anywhere and everywhere”. Information flows due to people’s flow. People can not rely on geographical restrictions, so as to realize the real dream of transmitting information anytime and anywhere. In addition to all the characteristics of digital learning, mobile learning also has its unique characteristics, that is, learners are no longer limited to the computer desk, and can learn for different purposes and different ways freely and anytime, anywhere. Using mobile Internet technology, learners can not only rely on computers to surf the Internet, but also use small and portable mobile terminals such as mobile phones, PDAs and smart phones to surf the Internet. In particular, the development and application of 3G technology can realize instant Internet access and always online, which makes information acquisition more convenient and information processing more real-time and efficient. The learning environment is mobile, and teachers, researchers, technicians and students are mobile. From its implementation mode, the technical basis of mobile learning is mobile computing technology and Internet technology, that is, mobile Internet technology. So far, in many countries, especially China, the number of mobile phone users far exceeds the number of wired Internet users. Through mobile Internet, mobile phone users who do not understand computers can easily obtain and process online information. The implementation tool is a miniaturized mobile terminal device. The equipment realized by mobile learning mainly has the following characteristics: portability, that is, the equipment is small in shape, light in weight and easy to carry; No linearity, that is, the equipment does not need to be connected; Mobility means that users can also use it well in mobile. This has greatly broadened the scope of education and will have a great driving force for lifelong education, democratization of education and personalized learning. Mobile learning is the product of the organic combination of mobile communication, network technology and modern education. It is the specific application of mobile communication technology in education. It represents a new direction of modern educational technology. Based on this, complete the steps of identifying mobile learning types.

176

D. Yu

3.2 Cloud Architecture Extracts Intelligent Education Features The research on learning platform under the background of cloud architecture is more conducive to meet the personalized learning needs of user groups. Using “Cloud Architecture” technology to analyze learners’ learning behavior, study learners’ e-learning, and continuously deepen the research and application of cloud architecture can bring greater development space for the construction of mobile intelligent education system [6, 7]. The concept of “intelligent education”, that is, the data mining application driven by knowledge support in the field of education, takes this opportunity to expect the application research in this field to flourish and integrate into the educational research method system. The new generation of information technologies, such as cloud computing, Internet of Things, cloud architecture, etc., continue to penetrate into the field of education and further promote the informatization of education. Because the data transmission between different networks, especially between mobile networks is no longer the same as the traditional Internet, data traffic is a more sensitive parameter. Mobile Internet requires fast data transmission and traffic saving. Therefore, the requirements for data transmission format are higher, and the traditional format needs to be optimized. The data stream transmission process is defined as: l=

(1 − φ)2 ×ε H

(1)

In formula (1), φ represents the access speed of the TCP protocol, H represents the total amount of data sent, and ε represents the response message. Specifically, the current stage of intelligent education refers to the use of information technology by educational technicians on the premise of fully understanding the educational situation, based on the existing and collected data in the educational process. The contradiction between the explosive growth of information and the slow improvement of human information literacy is prominent, and the ability of scholars with different information literacy to acquire, identify, and process information is significantly different, resulting in an information gap. Summarize data according to a certain purpose, use appropriate tools for data processing, establish models and analyze data through data mining technology, and transmit the found model content to educators or learners to assist them in educational research, decision-making or learning, and these models can be further deployed in various learning systems or educational management systems. The expression formula of transportation information flow in the system is:  (2) P= (s − η)−1 In formula (2), s represents the data transmission amount, and η represents the input coefficient. Recommend learning activities that adapt to the learner’s cognitive style, provide them with self-adaptive, personalized education services and the process of intelligent push, and provide directional tracking services for learners in the whole process of learning on the platform, and provide learners with learning context information and interests. Information such as hobbies and cognitive behaviors are captured in real time. Mobile education and training is based on behaviorism. The teaching design principles of behaviorism include: prescribed goals, frequent inspection, small steps and low

Design and Implementation of Mobile Intelligent Education System

177

error rate, self-paced, explicit response and immediate feedback. Suitable learning types are: Learn facts, define concepts, provide explanations, emulate processes, and solve problems. Provide intelligent management for educators or intelligent guidance, personalized learning and other services for learners. Intelligent education introduces data mining technology into the field of education, which belongs to quantitative research, but it should also comply with the general principles of general education research methods. Using artificial intelligence technology to mine association rules, accurately predict and understand learners’ learning needs, and then carry out push service. Quantitative research and qualitative research complement each other. On the other hand, not all problems are energy in the current level of educational development, and they will still be so in the future. Therefore, intelligent education is a knowledge driven data mining application in the field of education, and this definition is appropriate. In mobile learning, the mobile learning is the first mock exam. Because mobile devices are portable, learners can make full use of the scattered time. These characteristics of mobile learning are especially suitable for testing and practice. Based on the above description, the steps of extracting intelligent education features are completed. 3.3 Building a Course Resource Management Model The teaching process of intelligent education is a process of knowledge and information dissemination, which covers all the elements and contents of the above three communication theories. This process consists of selecting each basic component of the teaching process-the sum of tasks, contents, methods, means, forms, etc. The teacher course management function is mainly provided for teachers to set up their own online courses. Teachers can customize their own course summary and outline, and edit the chapter information under each course. The content that teachers need to customize in this chapter includes two types: courseware and standardized test questions. When the teaching objectives are determined, we should pay attention to the selection of communication information and the scheme design of information communication, including the connection between subject information, establish a reasonable information structure (including learners’ specific conditions, learning conditions, arranging learning information, etc.), find out the key points and pay attention to coordination, so as to improve the acceptability of information. Therefore, use the cloud to preprocess the service resources: calculate the distance from the service resources to the user’s location, and include this example in the distance characteristic. The idle resource characteristic is expressed as 1. The service status value is sent from the terminal. The service quality and efficiency are calculated by formula 2. Then match the weight of the service policy and resource scheduling, and the weight is expressed as: ⎤ ⎡ t11 t12 t13 t14 t15 ⎢t t t t t ⎥ ⎢ 21 22 23 24 25 ⎥ ⎥ ⎢ (3) tm = ⎢ t31 t32 t33 t34 t35 ⎥ ⎥ ⎢ ⎣ t41 t42 t43 t44 t45 ⎦ t51 t52 t53 t54 t55 In formula (3), t represents a service resource scheduling strategy, and m represents the scheduling strategy proposed by the user. The first to fifth columns are service

178

D. Yu

quality priority, closest distance priority, driving efficiency priority, taxi service status priority, and overall priority. According to the user’s service needs, the matching value is obtained by directly multiplying the resources and the n column of the weight matrix. In this system, teachers first set up their own courses and edit the relevant materials of the courses, such as course overview, course offering period, chapters corresponding to the courses, courseware under the chapters, and standardized test questions. After editing, they choose to publish the courses. The learner’s control of learning activities refers to the ability to complete learning tasks according to their actual needs and abilities, to self-check and correct their own learning activities, and to successfully complete learning tasks. Evaluating the teaching effect with effect standard and time standard is also a very important part in the dissemination of modern intelligent education. The representations, units, and data of the five characteristics of resources are not uniform, such as idle state, distance, state, efficiency, and evaluation. It is necessary to normalize each resource so that the resource characteristics can be used for subsequent calculations. The five characteristics of the resource are normalized respectively, and the normalization formula of the idle state is: G=

max(q) − cur(q0 ) max(δ) − min(δ0 )

(4)

In formula (4), q represents the number of service resources, q0 represents the characteristics of service resources, δ represents the distance of service resources, and δ0 represents the distance of service resources from the user’s location [8–10]. The purpose of evaluation is to judge whether the effect disseminated and obtained conforms to the maximum benefit. The module designs interfaces with different functions for students and teachers. The functions that students can use include: asking questions, viewing their own questions, viewing the 10 questions with the most visits, querying questions according to keywords, deleting their own questions, and viewing the existing questions in this chapter. The functions that teachers can use include: answering questions, viewing and correcting answers, deleting users, etc. The scheduling of service resources is divided into polling scheduling, distance first, idle first, quality of service first and other scheduling algorithms. This paper will form a feature first scheduling algorithm through the comprehensive use of these algorithms. The specific idea of the algorithm is as follows: the feature first scheduling algorithm first quantifies the characteristics of service resources. Through the statistics of the existing historical evaluation information, the favorable information ratio is obtained. When the evaluation information is only good and bad evaluation results, the favorable value is 1, otherwise the value is 0. Set the existing N evaluation information Yn, the score of each evaluation information is g, and the range of evaluation score is 0 ~ a, then the normalization formula of evaluation information is expressed as follows: g=

Yn  × (a − n)2 N

(5)

After the teacher publishes the course, students can log in to the system to query the optional course information, and view the course overview and course outline. If students intend to select the course, they can select the course to be reviewed by the opening teacher. The characteristic values are used to represent the service resources,

Design and Implementation of Mobile Intelligent Education System

179

and the service resources with different characteristics are normalized, so that the service resources can participate in the calculation. The expert group scores the different needs and importance of each feature of the service resources in the application processing, and obtains the resource feature weight under different user needs. The matching degree of resources is obtained by the product of weight and quantified resource characteristics, and polling scheduling is carried out according to the matching degree from high to low. After receiving the student’s course selection application, the teacher can view the basic personal information and previous learning of the student. If the teacher agrees that the students choose their own course, the topic selection is successful, and the students can start learning the course. Analyze and evaluate whether the time and amount are optimized, etal. Find out the causes according to the evaluation results, take corresponding measures, and make corrections and improvements.

4 Experimental Studies 4.1 Set up the Experimental Environment The details of the client’s development environment are as follows: Development platform: Eclipse for Android Developers; Android development plug-in: ADT C Android Development Tools. Development running environment: jdk1.8.0_60, Google Android SDK 1.16.0.0. Relational database: SQLite open source framework: Spring for Android; development language: Java, XML. The server side uses Jersey, Hibernate, Spring RELEASE and other third-party open source frameworks. Among them, Jersey is mainly responsible for the development of the resource layer, such as: registering resources, parsing and encapsulating transmission data formats, defining resource interfaces, and sending resource responses to the outside world. For the back-end server, the performance test is to consider and evaluate the time to respond to the request. Normally, the response time should be guaranteed to be within 3 s, which is generally acceptable to users. Client 1 configuration: OS: Windows 7, CPU: Intel E1260 1.8 GHz, RAM: 2G, resolution 480 * 320. Client 2 configuration: OS: Windows 7, CPU: Intel E1260 1.8 GHz, RAM: 2G, resolution 480 * 320. Client 3 configuration: OS: Android 2.1, CPU: ARM Cortex A8 800 MHz, RAM: 512M, resolution: 480 * 800. For the mobile client, the performance is mainly reflected in the smoothness of the screen when using the system functions. To investigate the reasons, it is necessary to consider the CPU usage (not too high, the system will be busy for a long time), and the memory usage (not too large, otherwise it will be blocked by the system). Optimized to kill the process) and GPU situation (drawing time should not be too long, otherwise it will cause frame drop and other situations that affect the picture display quality). Experimental tests were carried out in the above-mentioned experimental environment. 4.2 Experimental Results The experiment selects the NFC-based mobile intelligent education system, the deep learning-based mobile intelligent education system, and the mobile intelligent education system designed this time for experimental tests, respectively testing the CPU usage of

180

D. Yu Table 1. CPU usage when file size is 200 kB (%)

Number of experiments

NFC-based mobile intelligent education system

Deep learning mobile intelligent education system

The mobile intelligent education system in the text

10

4.315

4.115

1.669

20

3.667

3.648

1.267

30

4.102

4.206

1.088

40

3.589

3.776

1.758

50

4.065

4.335

1.599

60

3.114

4.007

1.623

70

4.056

3.458

1.544

80

3.746

3.866

1.729

90

4.213

4.064

1.663

100

3.558

4.213

1.728

the three systems when downloading files of different sizes. The experimental results are shown in Table 1, 2, 3, 4, 5 and 6: It can be seen from Table 1 that the average CPU occupancy rates of the mobile intelligent education system and the other two systems are 1.567%, 3.843% and 3.967% respectively. Table 2. CPU utilization when file size is 500 kB (%) Number of experiments

NFC-based mobile intelligent education system

Deep learning mobile intelligent education system

The mobile intelligent education system in the text

10

12.316

13.645

7.845

20

11.504

12.866

8.121

30

12.114

11.593

7.688

40

13.124

12.848

8.236

50

11.669

11.606

7.694

60

12.347

13.010

8.203

70

11.984

12.347

7.885

80

12.030

11.595

8.126

90

11.685

12.063

7.859

100

12.331

11.245

8.203

Design and Implementation of Mobile Intelligent Education System

181

It can be seen from Table 2 that the average CPU usage of the mobile intelligent education system and the other two systems are 7.986%, 11.998%, and 12.282%, respectively. Table 3. CPU occupancy rate of 1 GB file (%) Number of experiments

NFC-based mobile intelligent education system

Deep learning mobile intelligent education system

The mobile intelligent education system in the text

10

26.454

24.772

17.645

20

28.313

25.908

15.325

30

25.319

24.331

14.616

40

27.441

25.499

13.822

50

24.088

23.561

15.227

60

25.616

27.202

14.399

70

24.337

26.144

13.283

80

25.189

27.229

14.211

90

24.177

28.313

15.699

100

25.006

26.125

13.224

It can be seen from Table 3 that the average CPU usage of the mobile intelligent education system and the other two systems are: 14.745%, 25.594%, and 25.908%, respectively. Table 4. CPU occupancy rate when file size is 1.5 GB (%) Number of experiments

NFC-based mobile intelligent education system

Deep learning mobile intelligent education system

The mobile intelligent education system in the text

10

36.512

35.461

19.848

20

35.418

35.219

20.166

30

33.991

36.511

18.553

40

35.574

37.818

19.006

50

35.210

36.559

20.117

60

36.914

35.267

20.031

70

35.206

34.299

19.848

80

36.228

35.218

20.615 (continued)

182

D. Yu Table 4. (continued)

Number of experiments

NFC-based mobile intelligent education system

Deep learning mobile intelligent education system

The mobile intelligent education system in the text

90

35.209

36.915

21.535

100

36.778

38.220

22.160

It can be seen from Table 4 that the average CPU usage of the mobile intelligent education system and the other two systems are: 20.188%, 35.704%, and 36.149%, respectively. Table 5. CPU occupancy rate of 2 GB file (%) Number of experiments

NFC-based mobile intelligent education system

Deep learning mobile intelligent education system

The mobile intelligent education system in the text

10

47.151

46.554

29.616

20

44.369

45.031

31.005

30

45.915

44.518

32.147

40

42.612

46.906

32.066

50

46.303

44.677

31.548

60

45.918

45.223

29.288

70

46.991

46.208

31.599

80

48.205

47.566

31.206

90

49.778

48.312

31.588

100

51.007

49.522

29.008

It can be seen from Table 5 that the average CPU usage education system and the other two systems are: 30.907%, respectively. It can be seen from Table 6 that the average CPU usage education system and the other two systems are: 45.615%, respectively.

of the mobile intelligent 46.825%, and 46.452%, of the mobile intelligent 54.748%, and 55.163%,

Design and Implementation of Mobile Intelligent Education System

183

Table 6. CPU occupancy rate when file is 3 GB (%) Number of experiments

NFC-based mobile intelligent education system

Deep learning mobile intelligent education system

The mobile intelligent education system in the text

10

59.164

57.618

45.616

20

52.161

56.339

44.395

30

51.649

55.184

46.218

40

53.227

54.226

45.319

50

55.869

56.319

46.548

60

54.313

54.218

44.613

70

55.279

55.339

46.948

80

54.287

56.208

45.649

90

56.319

52.391

46.227

100

55.214

53.787

44.612

5 Conclusion At this stage, there are many intelligent education features, and the current mobile intelligent education system uses relatively backward education feature extraction technology, which leads to the fuzzy extracted education features and excessive CPU utilization during the system operation. In order to solve this problem, we take the intelligent education features under the support of the cloud architecture, and design a mobile intelligent education system based on the cloud architecture to improve the intelligence of the education system. The experimental results show that the CPU utilization of the system is low, indicating that the mobile intelligent education system with cloud architecture has better performance. The system meets the needs of people to query information anytime and anywhere in the teaching process. It not only enriches the teaching means, but also improves the teaching efficiency. It is an extension of the current intelligent education and the development direction of the future network education. In the future, the document import and synchronization functions of the system need to be continuously improved.

References 1. TLi, T.: Theoretical construction and implementation strategies of mobile intelligent terminals empowering community education. Adult Educ. (3), 25–29 (2020) 2. Jing, C., Zhi, J.: Experience design of intelligent education system based on the big data. Packag. Eng. 41(2), 21–27 (2020) 3. Yu, S., Chen, F.: Construction of smart education service ecosystem. E-educ. Res. 42(6), 5–13, 19 (2021) 4. Li, Q.: Design of network distance education system based on ASP under the background of Mu class. Mod. Electron. Technol. 44(6), 72–76 (2021)

184

D. Yu

5. Wang, P., Wang, C., Zhu, X.: Design and application of educational artificial intelligence system based on automation method. China Educ. Technol. 15(6), 7–15 (2020) 6. Yan, Y., Xie, H., Liu, G., et al.: Researching on nuclear-biological-chemical situation awareness system based on cloud architecture. Fire Control Command Control 44(1), 131–134 (2019) 7. Zhu, S., Guo, Y.: Emerging engineering education reform under the background of intelligent education: framework design and implementation approaches. Res. High. Educ. Eng. (6), 155–161 (2019) 8. Luo, Z.P., Jiang, Y.C., Hu, Z.J.: Cloud computing virtual resource enhanced multipoint secure transmission simulation. Comput. Simul. 38(1), 158–161, 166 (2021) 9. Che, J.X., Wei, W.W., Zhang, H.R., et al.: Research and design of blockchain technology applied to education system. Mod. Electron. Technol. 43(10), 92–96 (2020) 10. Hui, S.: Design and implementation of web-based remote multimedia teaching management system. Int. Educ. Forum 2(6), 31–35 (2020)

Design of Online Preschool Education Decision Support System Based on Data Mining Nan Li1(B) and Min Li2 1 Boda College of Jilin Normal University, Siping 136000, China

[email protected] 2 School of Literature, Capital Normal University, Beijing 100048, China

Abstract. Online preschool education decision support system has the problem that large-scale data can not be calculated in parallel, which affects the efficiency of random data writing and reading. An online preschool education decision support system based on data mining is designed. In the hardware part, CS5368 chip is used to realize the storage function of the acquisition node, 64K static random access memory 23LCV512 is used as the buffer, and SRAM reads and writes in byte mode. In the software part, the overall system is based on B/S architecture and combined with web technology to make the whole system application run on the server side. Using data mining technology to establish a database, the decisionmaking process of online preschool education is regarded as a classification and prediction problem. Design the system function module to complete the management operations such as data addition, deletion, modification and query. The system performance test results show that the total time and rate of random data writing and reading of the system are obviously better than the decision support system based on GA-BP neural network and artificial intelligence, so it has higher data processing efficiency and load capacity. Keywords: Data mining · Online education · Preschool education · Education data · Decision support · System design

1 Introduction Modern information technology with multimedia technology, computer technology and network technology as the core is widely used in higher education, providing a new technical means for the management and reform of discipline construction in Colleges and universities [1]. To build the discipline of online preschool education into a modern, open and international first-class discipline, we must make decisions that conform to the discipline development law. The need for scientific decision-making in the discipline construction of online preschool education in Colleges and universities provides a demand background for the application of decision support system [2] in the discipline field of colleges and universities. Zhang zhuoyou and others designed and implemented an intelligent decision system. On the basis of establishing a reasonable bid evaluation index system, they designed the © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 185–196, 2022. https://doi.org/10.1007/978-3-031-21161-4_15

186

N. Li and M. Li

fitness function with the network mean square error of BP neural network algorithm, and then established a computer automatic bid evaluation model based on GA-BP neural network by MATLAB programming [3]. Zhan Jinwu and others designed an adaptive evaluation decision support system based on artificial intelligence. In order to avoid the limitations of single index decision-making and the defects of subjective judgment, the system completes the quantitative selection of multi index intelligent decision-making by using the method of combining intelligent design theory and decision theory [4]. The system combines the evaluation model with knowledge acquisition to represent knowledge in the form of rules, and the evaluation results are consistent with the actual situation. Although the above research can integrate all kinds of data and realize the openness and convenience of decision support system, it is applied to online preschool education decision support. Because online preschool education involves many aspects of decision data, it has the problem that large-scale data cannot be calculated in parallel. Decision making is a kind of behavior activity that people decide strategies or effective schemes to achieve a certain purpose. Any type of behavior activity is also a result of relevant decisions. Data mining is the process of knowledge discovery and application (prediction). In the field of education, knowledge or rules are embodied in the rules of teaching and learning, so educational data mining is the discovery and application of teaching and learning rules (prediction) process. The reason why data mining can play a role in the learning and teaching process. Firstly, data mining is used to analyze the past learning experience of learners or others, so as to predict the learning conditions for effective learning and the future learning behavior of learners. Then, according to the above data, in order to provide learners with targeted learning resources and learning processes, teachers can guide learners to learn, so as to solve the problem of large-scale data parallel computing. Therefore, this paper designs the online preschool education decision support system based on data mining, uses advanced tools and new technologies to assist in making more effective decisions, and explores the application of the decision support system based on data mining in the discipline construction of colleges and universities.

2 Hardware Design of Online Preschool Education Decision Support System Online preschool education decision support system collects more education data, so it can not directly transmit the collected data to the host computer in real time. In order to ensure that the node has enough space to store the collected data, the measurement node must have its own storage function. Taking the sampling speed of 1 s interval as the standard, the node needs 17 bytes of space for one measurement, and 17 bytes include measurement results and time tags. The system uses CS5368 chip to realize this function. The chip is a 32 MB flash memory with SPI interface, and the flash is read and written in sectors, 4K bytes per sector. The schematic diagram of flash storage circuit is shown in Fig. 1.

Design of Online Preschool Education Decision Support System

VD_AD

R

C

C

NC

VCC

RST SO SCLK CS5368 SI SCK MOS GND GND

187

R

R

GND

Fig. 1. Flash memory circuit

The reference voltage collected by CS5368 is generated by the chip itself, and its value is half of VA. it requires three clock sources during operation, namely MCLK (master clock signal), SCLK (shift clock signal) and lrck (acquisition clock signal). Most of the energy is exchanged between the device power supply on the circuit board and the pin connecting the ground wire. Each pin is directly connected to the ground and is independent of each other. The clock management module of CS5368 can be configured through hardware. You can choose to use an external controller to provide clock source to CS5368 or an external crystal oscillator to provide clock signal to CS5368. In this paper, the crystal oscillator clock pins XT0 and XT 1 in CS5368 are grounded, and the clock source signal is provided through external clock signal, that is, through FPGA, Then it is configured through FPGA to make its working mode as pin configuration mode. The power module is responsible for level conversion and supplying other modules. Therefore, MC79FC33HT 1 step-down DC-DC regulator of mc78fc00 series is selected to realize 1.5 V to 3.3 V voltage conversion. Flash has large storage capacity, but it takes a long time to read and write by page. SRAM writes by byte, and its single write is on the microsecond level. In order to meet the storage requirements and reduce the impact of data writing on measurement synchronization, SRAM and flash are used to realize the storage function. Divide the SRAM into two blocks. After one area is filled, switch to another area to continue writing. The data in the full block is written to flash. Let CS5368 generate SCLK clock and lrck clock internally. The analog signal at the current end enters the ADC. Firstly, it will be converted into digital signal through oversampling, filtering and other operations inside the ADC, and then output through sdout pin. CS5368 has four acquisition data output pins. In this subject, it is configured as 3.3 V level output, and the sdout of each channel is to multiplex the analog signals of two front ends. Because SPI communication is adopted, pull-up resistance is adopted to ensure the reliability of communication. Using 64K static random access memory 23LCV512 as buffer, SRAM can read and write in byte mode. The isolation band and a large number of vias in the circuit board will inevitably make the current distribution uneven, and this layout will cause voltage fluctuations. In this case, the voltage fluctuation will have a certain negative impact on the power supply pin and the device connected to the ground wire. In order to avoid this negative effect, a decoupling capacitor should be added next to the chip. On the one hand, the change of current in the circuit network can be reduced, so the instantaneous overshoot of voltage on the power supply will also be reduced. On the

188

N. Li and M. Li

other hand, the decoupling circuit can effectively suppress and eliminate the radiation of the power loop on the PCB and the burr on the power supply, and better suppress the parasitic coupling between circuits.

3 Software Design of Online Preschool Education Decision Support System 3.1 System Architecture Design The structure of online preschool education decision support system designed in this subject is established on the basis of fully combining the characteristics of discipline management in Colleges and universities, the applicability of decision support system and the friendliness of users. Online preschool education decision support is actually a process of data and information exchange. All aspects involved in discipline management are usually quantified with certain data [5]. The design of the system fully considers the existing technical characteristics and future functional requirements, and uses a variety of integration technologies provided by Microsoft net platform to make the whole system not only realize the existing functions, but also adapt to the future needs and technical requirements. At the same time, in order to integrate with the third-party system, The system adopts service-oriented architecture to exchange information with the thirdparty system through standard data exchange interface to realize the integration between systems. The system combines web technology with B/S architecture to make the whole system application run on the server side, which can ensure a good running environment of the system. The system fully supports XML, soap, web service and other currently widely supported open standards, which ensures that the system can exchange data with application systems and databases of other platforms, and carry out application level interoperability and interconnection. More importantly, the B/S architecture has good scalability and communication ability, so that the system can easily communicate with other MIS systems and expert systems to achieve more intelligent decision-making ability. For example, the system can send the knowledge obtained from data mining to the expert system, so it can automatically add knowledge to the knowledge base of the expert system, In this way, the knowledge acquisition ability of the expert system is further improved, and the reasoning results can also be obtained from the expert system. Moreover, B/S architecture ensures good interactivity, strong page presentation ability and interaction ability, and displays decision-making knowledge and mode. The architecture of the online preschool education decision support system finally designed is shown in Fig. 2. The system architecture uses a business relational database as the decision information base (knowledge base), which solves the coupling of training knowledge and using knowledge, because knowledge mining is a very time-consuming operation and requires a lot of reasoning and calculation. If the system needs a long operation process every time, it is difficult to ensure the real-time performance of the system [6]. The system supports the output display of data and information. The system shall provide rich and friendly output interface and display platform through the design of Multimedia Library and its management system, so as to facilitate users, decision makers and other

Design of Online Preschool Education Decision Support System User browser

Data source Data cleaning, transformation and integration Database and server

189

Response

Application server

Decision information base

Request

Web server

Fig. 2. Architecture of online preschool education decision support system

information demanders to query, browse and make decisions. The mining knowledge is stored through the business database. When users make decisions on decision-making problems, they can directly obtain the existing knowledge from the knowledge base, so as to improve the real-time performance of the system. 3.2 Establishment of Online Preschool Education Database Based on Data Mining To make full use of the effective data resources in online preschool education decisionmaking, database is the best choice at present. The main function of database is to store, manage, provide and maintain data for decision support. The database stores various data information inside and outside the school required for online preschool education monitoring, analysis and evaluation. Using data mining technology, online preschool education decision-making process can be regarded as a classification prediction problem, and an education decision can be determined as a supported or unsupported problem [7]. Applying data mining to realize data classification is actually to construct a mathematical model or classifier to predict its class label, so as to realize parallel computing of large-scale data [8]. Support vector machine is proposed for binary classification, but many problems in preschool education decision-making are multi classification and need to be extended to multi classification. At present, there are two main ideas to realize multi classification support vector machine: the first is to optimize the objective function to construct a multi classification model, so as to realize the multi classification problem; the second is to reduce the multi classification to multiple two classification problems. Database technology is used to solve the problem that business data cannot solve - data analysis. By effectively organizing a large number of business data, the database can realize business intelligence and help managers get rid of the blindness of decision-making without data support. In order to improve the generalization ability of binary tree multi classification SVM, the key is to use reasonable strategies to generate binary tree. The distribution volume of various sample data is calculated according to the formula of the minimum class inclusion of hypercube or the minimum class inclusion of hypersphere. Assuming that a certain type of data P has s samples p1 , p2 , · · ·, ps , the center of gravity of such sample set can be expressed as: 1 p= pa s s

a=1

(1)

190

N. Li and M. Li

In formula (1), p represents the center of gravity of such sample set; s represents the total number of samples; a represents the serial number of data samples. The volume of hypersphere is calculated as: V = π rm

(2)

In formula (2), V and r represent the volume and radius of the hypersphere respectively; m represents the dimension of the sphere. Because the purpose is to compare the relative size of the distribution range of each category, radius comparison can be used to replace the volume, and only the radius needs to be calculated. Then the minimum radius of the hypersphere containing these samples is: r = max{p − pa }

(3)

Sort the categories according to the distribution volume from large to small. The system puts data analysis and data mining in an important position in the design, regards the whole decision-making process as a data mining process or a data analysis process, and provides the interface of data mining algorithm. After the data set is divided into blocks, each block calls for parallel training processing to obtain the support vector of each data block. Database management system is a software for the establishment, use and maintenance of database. Users can access data, call and calculate model library and method library through database management system. The support vector obtained from the training data block is transmitted to the reduce end for processing. After the data is transferred to reduce, it will be sorted and merged, and finally the results will be output to the specified file. For the domain problems to be solved, the decision model or decision knowledge is trained by designing or selecting a reasonable data mining algorithm to act on the historical data. These decision-making knowledge is the reference for decision-makers to make scientific decisions. 3.3 Design System Function Module To access the system, users need to be authenticated by the user login module. At present, the system provides three roles for users to access the system, which are divided into the heads of the competent departments of the Education Commission. The user selects the corresponding role and enters the user name, the password and verification code of the Department in charge of discipline construction in Colleges and universities and each student in Colleges and universities. Only after the system is verified can he log in to the system and enter the system home page. The ultimate purpose of the database is for the upper application, so it needs to have accessibility and presentation. The data application layer of online preschool education evaluation and decision support system adopts a two-tier structure, which is divided into access layer and display layer to meet different control needs. The database access layer is mainly used for comprehensive evaluation, early warning monitoring, operation analysis, coordination control and other online preschool education data query. The display layer is used for data access display of desktop operating system, large screen and other outdoor multimedia display systems. The data management function of the system is mainly completed by the data management module of key disciplines. They jointly realize the management operations

Design of Online Preschool Education Decision Support System

191

such as adding, deleting, modifying and querying the data used in the system. The addition of system data also realizes the functions of single data entry and batch data upload according to business requirements. Knowledge base management system is an important branch of artificial intelligence technology. Its intelligence is mainly manifested in that it can imitate human expert thinking to solve complex problems in a specific field. The basic principle of knowledge base management system is to transform the problem solving process into a search problem in state space. The known facts of solving the problem constitute the initial state, and the final solution of the problem constitutes the target state. There are different levels of sub target states between the initial state and the target state, The solution process is to use the inference engine to find a path from the initial state to the target state in the whole state space. The business logic layer reads the data in the cloud environment after ETL processing in the data layer, calls the mining algorithm library to perform data mining analysis, and finally returns the mining results to the presentation layer and interface service layer. Through the corresponding function buttons on the navigation bar, you can enter the data management interface, through which you can view all historical data in the system, or enter each specific record to modify all indicator attribute values of each discipline record. The business logic layer should undertake these functions: running the mining algorithm to calculate and analyze the data, executing the mining algorithm in the parallel computing environment, managing the mining process and results, and undertaking the development of business logic scalability. The business logic layer is mainly composed of these key components: mining algorithm computing module, cloud computing module, mining management module, policy design pattern module, etc. Multimedia library is a part of the decision support system designed separately in this paper. It is mainly used for the display interface in the human-computer interaction system. The display interface function design includes report display, process display, visual display, intelligent display, indicator display, event and early warning display, etc. Through the multimedia library management system, the report, analysis and evaluation model are displayed comprehensively and vividly to realize the generation and display of various graphics such as pie chart, histogram, broken line chart, two-dimensional table, honeycomb chart and dashboard, as well as the function of visual guide service. So far, the design of online preschool education decision support system based on data mining has been completed.

4 System Performance Test 4.1 Test Preparation The online preschool education decision support system designed this time is built on a Hadoop cluster composed of six computers running HDFS distributed file system. The operating system of all node machines is Ubuntu 14.04. A computer with the same configuration acts as an application server for deploying the online preschool education decision support system. The system provides services for users through browser or intelligent terminal application. After referencing the DLLs related to Microsoft Sync Framework in.Net, first create a SQL sync scope provisioning synchronization object, then create local and cloud providers, and start the synchronization clientprovision by using the file provider provided by Sync Framework.

192

N. Li and M. Li

The experimental data comes from a kindergarten. The decision-making data generated during its online teaching is taken as the source and set up a database for testing. 4.2 Test Results and Analysis Performance test is mainly responsible for testing the load capacity of online preschool education decision support system. In order to verify the superiority of the system designed in this paper, the test results of the system are compared with the decision support system based on GA-BP neural network and artificial intelligence. This experiment carries out stress test on the database of online preschool education decision support system, stores millions of test data on the resource memory, and tests the total time and rate of random writing and reading of data respectively. The comparison results of the total time length of data random writing and reading are shown in Table 1 and Table 2, and the rate comparison results are shown in Table 3 and Table 4. Table 1. Comparison of total random write time of data (s) Number of experiments

Decision support system based on data mining technology

Decision support system based on GA-BP neural network

Decision support system based on artificial intelligence

1

826

1125

1236

2

804

1208

1231

3

856

1147

1342

4

849

1354

1475

5

925

1168

1254

6

912

1239

1363

7

901

1226

1321

8

765

1312

1284

9

882

1125

1119

10

858

1150

1243

The experimental data writing and reading have a certain randomness, and its efficiency can characterize the operating pressure of the system. According to the results in Table 1, the average total time of data random writing in the online preschool education decision support system based on data mining technology is 858 s, which is 347 s and 429 s shorter than the system based on GA-BP neural network and artificial intelligence.

Design of Online Preschool Education Decision Support System

193

Table 2. Comparison of total time of random data reading (s) Number of experiments

Decision support system based on data mining technology

Decision support system based on GA-BP neural network

Decision support system based on artificial intelligence

1

204

299

324

2

216

312

326

3

207

324

315

4

214

319

328

5

199

299

325

6

198

305

317

7

207

312

299

8

215

307

311

9

208

314

304

10

206

319

298

According to the results in Table 2, the average total time of random data reading of online preschool education decision support system based on data mining technology is 207 s, which is 104 s and 108 s shorter than the system based on GA-BP neural network and artificial intelligence. Table 3. Comparison of data random write rates (rows/s) Number of experiments

Decision support system based on data mining technology

Decision support system based on GA-BP neural network

Decision support system based on artificial intelligence

1

1658

1025

997

2

1614

1047

984

3

1707

1076

978

4

1628

1154

1195

5

1759

1068

1056

6

1836

1189

1265

7

1525

993

1124

8

1764

935

1041

9

1627

929

1082

10

1631

1014

994

194

N. Li and M. Li

According to the results in Table 3, the average rate of random data writing in the online preschool education decision support system based on data mining technology is 1675 rows/s, which is 632 rows/s and 603 rows/s higher than that based on GA-BP neural network and artificial intelligence. Table 4. Comparison of data random reading rate (rows/s) Number of experiments

Decision support system based on data mining technology

Decision support system based on GA-BP neural network

Decision support system based on artificial intelligence

1

4672

3181

3034

2

4504

3044

3064

3

4647

3287

3189

4

4888

3165

3296

5

4965

3452

3365

6

4826

3354

2938

7

5055

3218

3222

8

4994

3006

3243

9

4738

3054

3371

10

4822

3135

3054

According to the comparison results in Table 4, the average data random reading rate of the online preschool education decision support system based on data mining technology is 4811 rows/s, which is higher than that of the system based on GA-BP neural network and artificial intelligence by 1621 rows/s and 1633 rows/s. Based on the above results, the online preschool education decision support system designed based on data mining technology has strong load capacity, supports the writing and reading of large-capacity data, and provides users with an efficient data storage environment. System response time is an important index to verify system performance. Compare the response times of the three systems, and the results are shown in Fig. 3.

Design of Online Preschool Education Decision Support System

195

9 Decision support system based on Data Mining Technology

8

Decision support system based on Artificial Intelligence Decision support system based on GA-BP neural network

System respo nse time/s

7 6 5 4 3 2 1 0

0

2

4

6 8 10 Number of iterations/time

12

14

16

Fig. 3. Comparison of response time of three systems

According to the analysis of Fig. 3, the response time of the system in this paper shows an increasing trend under different iteration times. Although the decision support system based on data mining technology and the decision support system based on GA-BP neural network fluctuate to a certain extent, they also show an upward trend on the whole. By comparing the specific data results, it can be seen that the shortest response time of the system in this paper is only 1.4 s and the longest is only 3.6 s. After comparison, it is found that under the same number of iterations, the system in this paper has the shortest time and the highest efficiency. Although the decision support system based on data mining technology is stronger than the decision support system based on GA-BP neural network, its time is also much higher than the system in this paper.

5 Conclusion The research on online preschool education decision support system is of great significance to improve the objectivity and scientificity of discipline construction and development decision-making. This paper designs the online preschool education decision support system based on data mining. Millions of test data are stored in the resource memory. The total time and rate of random writing and reading of data in the system are obviously better than the decision support system based on GA-BP neural network and artificial intelligence, which supports the writing and reading of large capacity data. After educational analysis and decision-making, educators need to improve and optimize teaching according to decision-making. The specific implementation of improving and optimizing teaching requires school management authority, long experimental cycle and data comparison during the experiment. It is expected to organize relevant teams to further carry out relevant research in the future.

196

N. Li and M. Li

References 1. Zhang, Z., Hu, K., Jiang, X.: Design and implementation of a tender evaluation IDSS system based on DEA-GA-BP neural network. Comput. Sci. Appl. 10(3), 541–552 (2020) 2. Zhan, J., Li, T., Li, C.: Decision support system of adaptability evaluation for TBM selection based on artificial intelligence. J. China Coal Soc. 44(10), 3258–3271 (2019) 3. Huang, Y., Zhao, C., Zhao, G., et al.: Intelligent technology of education process mining: research framework, status and prospect. E-educ. Res. 41(8), 49–57 (2020) 4. Liu, W., Li, X.: Research on terminal online education data mining technology based on model driving. Mod. Electron. Tech. 43(16), 112–114, 118 (2020) 5. Peng, Q.: Emergency decision support system demand data self-service mining simulation. Comput. Simul. 36(8), 329–332 (2019) 6. Zhang, T., Zhang, S.: Research on design and computing of learner model of educational big data mining. E-educ. Res. 41(9), 61–67 (2020) 7. Li, Y., Dai, X., Huang, X., et al.: Design and research of the decision-making support system of hospital teaching management based on data mining. China Digit. Med. 15(2), 63–65, 109 (2020) 8. Wang, W.: Simulation of software secret data loss prevention transmission based on mobile gateway. Comput. Simul. 38(11), 459–463 (2021)

Construction of Online Education Model of Marketing Specialty Based on Cloud Computing Lingli Mao1(B) and Zhichao Xu2 1 Guangzhou Huali Science and Technology Vocational College, Guangzhou 511325, China

[email protected] 2 Dalian University of Science and Technology, Dalian 116033, China

Abstract. In the process of constructing the online education model for marketing major, the actual application requirements are not considered, and the problems of low functional success rate and poor operation performance exist. In order to provide auxiliary work for the teaching work of marketing major, the cloud computing technology is used to realize the optimization design of the online education model for marketing major. Install cloud servers, modify cloud storage, communication networks and other hardware devices as the operation support of the model. Determine the online education mode and teaching content of marketing specialty. Use the web crawler algorithm to collect the online education resources of marketing specialty, and complete the sharing of online education resources according to the principle of cloud computing technology. Finally, through the design of online education activities and interactive functions, the construction of online education model is realized. Through the model test experiment, it is concluded that the functional success rate of the design model reaches 98.6%, and the response time and memory occupation meet the test requirements. Keywords: Cloud computing · Major in marketing · Online education · Construction of educational model

1 Introduction The major of marketing is to cultivate senior professionals of business administration who have the knowledge and ability of management, economy, law, planning and copywriting, market research and marketing, and can engage in marketing and management, teaching and scientific research in enterprises, institutions and government departments. It is an emerging and popular specialty, and the market is everywhere. Marketing is not only sales, but also includes the development of the market, providing needed services and help for people or things in need, and understanding can achieve correct marketing [1]. Marketing is divided into macro and micro levels. Macro marketing is a reflection of social economic activities. Its purpose is to meet social needs and achieve social goals. Micro marketing is an economic activity process of an enterprise. It produces marketable © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 197–210, 2022. https://doi.org/10.1007/978-3-031-21161-4_16

198

L. Mao and Z. Xu

products according to the requirements of target customers, and flows from producers to target customers. Its purpose is to meet the needs of target customers and achieve the objectives of the enterprise. The core of marketing activities is exchange, but its scope is not limited to the circulation process of commodity exchange, but also includes prenatal and postnatal activities. The marketing activities of products are often longer than the circulation process of products. The transaction scope of modern society is very wide, which has broken through the barriers of time and space and formed a universal market system. Education requires both theoretical teaching and practical teaching. Theory aims at “moderation and sufficiency”, and practical skill teaching should be strengthened. Marketing specialty is not only a traditional specialty, but also a specialty that needs to constantly update its teaching contents and teaching methods. It is closely related to social progress, the development of productive forces, the renewal of science and technology, the change of economic situation and the reform of economic system. Theory is closely related to practice, and has strong theoretical and practical significance. The traditional teaching plan and system of marketing major only pay attention to the teaching of the theoretical part, but do not place the practical part in its due position and pay full attention to it. This practice is closely related to the training objectives, analysis methods, teachers’ inherent concepts and ability levels of graduates in the past. Higher vocational education is a new way of education. Practical teaching is incorporated into its teaching system and regarded as a very important part. According to the reflection of the society on the talent demand information over the years, the demand for marketing professionals has always been in the forefront, but the society has low satisfaction with the job performance of marketing graduates, which is mainly reflected in the unsatisfactory practical application ability, the long time to adapt to the post, some basic businesses are difficult to carry out, it takes a long time, and it is difficult to find high-quality marketing talents, There are too few talents who can formulate scientific, reasonable and effective marketing strategies and marketing plans. In view of this situation, it is not difficult to find the reasons, and its root is the problem of education. The practical teaching of marketing specialty is a weak link. Therefore, we should reconstruct the teaching plan of marketing specialty, pay attention to practical skill education, and constantly explore the practical teaching mode, so as to cultivate excellent marketing talents. In order to provide sufficient and high-quality marketing talents for the market and break the problems existing in the traditional teaching of marketing specialty, an online education model of marketing specialty is constructed. From the current construction of the online education model of marketing specialty, we only pay attention to the teaching of knowledge, ignore the subjectivity of students in learning, seriously inhibit the sense of innovation, lack of vitality in the classroom and lack of interest in learning, which is not conducive to the cultivation of students’ overall quality. The online education model has obvious problems such as low success rate and poor performance. Therefore, cloud computing technology is introduced. Cloud computing is not only a method of sharing it infrastructure, but also an extensible software library and program library. It connects a large number of computing resources to form a “resource pool” in a secure and flexible way to provide users with cloud applications and services stored in large data centers. In the optimization design of online education

Construction of Online Education Model

199

model of marketing specialty, the application of cloud computing technology is expected to provide auxiliary tools for the teaching of marketing specialty while improving the educational function and operation performance of the model.

2 Design of Online Education Model for Marketing Specialty The online education model of marketing specialty refers to an open teaching system based on teaching concepts, teaching methods and teaching processes. For marketing related majors, good teaching results can be obtained through the teaching platform according to their own syllabus, teachers’ characteristics and students’ situation. At the same time, the teaching methods of marketing planning should include a variety of teaching methods, not specifically a certain kind of teaching methods. In the teaching platform of marketing planning, the collection of teaching methods needs to maintain long-term openness, and can continuously update their own contents and methods according to the changes of the external market. Schools and teachers can update their teaching method system according to their own resources and students’ overall knowledge reserve. The optimized online education model of marketing specialty is an organic whole composed of various elements of practical teaching activities, including target system, content system, management system, guarantee system and other elements. Its specific framework is shown in Fig. 1.

Driving level

Practical teaching objectives

Passive level

Practical teaching content system

Control level

Practical teaching management system

Guarantee level

Practical teaching guarantee system

Teaching system of Marketing Specialty

Fig. 1. Block diagram of online education model of Marketing Specialty

Under the constraints of the principles of demand orientation, system construction, focus, personality development and combination of production and learning, the online education model of marketing specialty is optimized through the application of cloud computing technology. 2.1 Install ECs and Related Hardware Devices The optimized online education model of marketing specialty based on cloud computing technology needs to be supported by ECs and related hardware devices. The optimized ECS is composed of four independent modules, as shown in Fig. 2.

200

L. Mao and Z. Xu

Core processing module

UDP

Com munic ation protocol analysi s module

UDP

Socket connection module

Dat abase m odule

Fig. 2. Internal structure of ECS

As can be seen from Fig. 2, ECS consists of socket module, communication protocol analysis module, core processing module and database module [2]. Socket module provides TCP connection with gateway or administrator, and is the communication interface between server and gateway or administrator; The communication protocol analysis module has a receiving buffer and a sending buffer. The module preliminarily processes the data reported by the socket and stores it in the receiving buffer, waiting for the core processing module to read the buffer data. At the same time, the data command issued by the core processing module is processed, stored in the sending buffer and sent to the socket module; The core processing module is the core of server data processing. It analyzes the data frame information in the original data and makes corresponding processing, or stores the data in the database or reads the data from the database for processing. After the data operation, it sends the feedback information packet to the gateway or administrator through the communication protocol analysis module [3]. The communication between modules within the server adopts UDP transmission mode. Although UDP is a connectionless communication mode, it can be considered that using

Server

Server

Server

Socket

Socket

Socket

Bind

Bind

Bind

Sendto/sendmsg

RevefRom/ recvmsg

Sendto/sendmsg

RevefRom/ recvmsg

Sendto/sendmsg

RevefRom/ recvmsg

End

End

End

Fig. 3. Schematic diagram of UDP communication

Construction of Online Education Model

201

UDP to transmit data between modules within the server has high reliability and little packet loss rate. The connectionless UDP communication model is shown in Fig. 3. The UDP transmission communication frame for communication between modules in the server is composed of five parts, in which the length represents the sum of the length of source address, destination address, load and error correction code; Each module is assigned an ID with a size of 1 byte to represent its address information; The load fills the original data collected by the sensor node, and the length is variable; The error correction code is used to check the error correction ability starting from the length. In addition to the cloud server and its communication network, in order to provide sufficient educational resources for the marketing specialty, it is also necessary to expand the storage space of cloud shared memory [4]. Finally, the modified and optimized hardware equipment is connected together through the power supply circuit to provide hardware support for the operation of the online education model of marketing specialty. 2.2 Determine the Online Education Mode of Marketing Specialty Based on the concept of online education model and preliminary research, combined with the talent training and curriculum characteristics of marketing specialty, the curriculum model of marketing specialty is determined, as shown in Fig. 4.

Teaching model

Registered courses Lear ner

Desig Reso Participa n urce tion them shari activitie e ng s

Offering courses Modi Uploa Impleme fy d nt them resour assessm e ces ent

Teac her

Teacher student educational administration system, social network, discussion group

Fig. 4. Schematic diagram of marketing education mode

The selected curriculum model takes students as the main body of education. Students can learn about the curriculum dynamics through the interactive network platform, design or participate in the theme activities of the curriculum based on their own needs and interests, and independently obtain the curriculum resources required for theme learning. Under this curriculum model, students are no longer passive learning recipients, which not only mobilize students’ learning enthusiasm, stimulate students’ in-depth research

202

L. Mao and Z. Xu

on specific problems, but also improve students’ ability to adapt to the society. At the same time, each course takes students’ learning path as the core, with clear course objectives, themes, time arrangement, assessment, interaction and resource download, giving learners a clear learning orientation. The course model of marketing specialty includes four structures: course introduction, course content, course assessment and communication and interaction. Among them, the course introduction includes the course outline, teaching plan and teaching announcement, which is the course information that learners need to know in advance to select and register the course [5]. The course content includes a large number of video courseware, handouts, guiding documents, examination question bank, case bank and related information link resources. Rich curriculum resources are the key to the education model. Course assessment includes teacher assessment and assessment between students. Teacher assessment includes in class test module and final test module with randomly generated test questions. Teachers can set the type and difficulty of test questions in the background system and extract them in groups according to the purpose of the test. Assessment among students refers to the mutual scoring of homework among students. The communication and interaction platform includes the teaching and educational administration system of teachers and students, various forms of social networks, discussion groups, etc. it can be used to provide and share resources, organize and accept assessments, communicate and discuss courses, answer homework questions and discuss freely. In addition to the basic professional education mode, it is also necessary to set up practical teaching mode for professional practical skills. The practical skills of marketing specialty can be divided into four modules: First, professional basic skills refer to the most basic basic skills required by the specialty, mainly including computer application, statistical analysis, financial analysis, economic writing, legal application, etc. Second, professional basic skills refer to the basic behavioral abilities required to carry out marketing work, mainly including market research, customer visit, product promotion, business negotiation, public relations, customer management, logistics distribution, Bill filling, market development, sales service, advertising practice, on-site management, etc. Third, marketing core skills refer to the ability to formulate marketing strategy and technology and plan marketing activities. According to the external environment and internal resource conditions of the enterprise, formulate the strategic technology of the enterprise and plan an effective scheme for the marketing activities of the enterprise [6]. Fourth, professional development skills refer to determining the employment direction and broadening the employment path according to the market demand and students’ personality, interests and specialties on the basis of mastering corresponding knowledge and skills. Professional development involves retail, service, insurance, international trade, franchise and other industries, enterprises and posts. The above four practical teaching skill modules are contained in in class training, special training, comprehensive training and social practice. On the other hand, when arranging the practical skills teaching in the above four aspects, we should pay attention to the measurement and time sequence of class hour arrangement. The class hour distribution of professional basic skills, professional basic skills, professional core skills and professional development skills can be determined according to a certain proportion. According to the total practice class hours and the percentage of each module, the specific class hours of each module can

Construction of Online Education Model

203

be quantified. When the total class hours of each module are known, the teaching hours of each specific practical project can be further decomposed and refined [7]. The first mock exam of four modules of the marketing practice skill is the ladder level. The first mock exam is the basis of the latter module, and the order relation should be considered when designing the teaching plan. 2.3 Choose the Online Education Content of Marketing Major The online education content of marketing specialty is selected on the basis of providing students with meaningful expressive activities. The education content is carried out on the basis of the subject knowledge structure of the course. The content is divided into three chapters, of which the third chapter is divided into five units, as shown in Table 1. Table 1. Online education content of marketing specialty Content section

Unit

Educational objectives

Introduction to marketing planning

–

Understand the concept of marketing planning, understand the discipline characteristics and research objects of marketing planning

Marketing planning process

–

Understand the principles and processes of marketing planning, and be familiar with the structure and content of marketing planning book

Marketing planning training

Marketing planning purpose

Complete a marketing plan through group cooperation

Market situation analysis Analysis of market opportunities and problems Determine specific marketing planning Implementation and control of marketing planning scheme

Take the content in Table 1 as the educational content, and take it as an important part of the online education model of marketing specialty. 2.4 Collect Online Education Resources of Marketing Specialty The automatic collection of educational resources in school for marketing majors uses topic crawling technology to collect topic related information resources from the web in an automatic way. Topic crawling refers to selectively collecting topic related information

204

L. Mao and Z. Xu

resources from the web for a specific topic field. Topic crawling needs to filter the content irrelevant to the topic, predict the links related to the topic, then add the relevant links to the URL queue to be collected, and determine the information collection order according to a certain priority algorithm [8]. The goal of general crawling technology is to collect as many web information resources as possible. When the collection capacity is limited, generally give priority to collecting influential site information, and the subject field of these resources is secondary. However, for topic crawling, its goal is to crawl topic information resources to the greatest extent in unit time, and bypass websites or web pages not related to the topic as much as possible, so as to save software and hardware resources, storage space and network bandwidth resources. In order to ensure the application value of educational resources, it is necessary to process the missing data in the resources. The processing process can be expressed as follows:  n   (X − X0 ) · · · (X − Xn ) × Yi βn (x) = (Xi − X0 ) · · · (Xi − Xn )

(1)

i=0

In formula (1), X and Xi represent the mean value and missing value of education resource data respectively, Yi is non missing education resource data, and the final calculation result βn (x) represents the interpolation filling result of missing education resource data. On this basis, the initial resource data is normalized, and the processing formula is as follows: y(x) =

βn−max (x) − βn (x) βn−max (x) − βn−min (x)

(2)

In formula (2), βn−max (x) and βn−min (x) are the maximum and minimum values in the resource data respectively. Finally, in order to realize the efficient management of educational resources, it is necessary to classify the collected educational resource data. The similarity between any two educational resources is calculated by formula (3). t 

wi,j × wi,β   dj · β(x) i=1   =  sim q, dj =  t  dj  × |β(x)| t   2 2 wi,j wi,β i=1

(3)

i=1

In formula (3), variable dj represents the category of educational resources, wi,j and wi,β respectively correspond to the weight of feature items in educational resources and their weight values in resource category dj . Set the classification threshold to ηthreshold .  If the calculated sim q, dj value is higher than the classification threshold, the corresponding educational resources need to be classified. Otherwise, calculate the similarity of the next group of resources until all the collected resources are classified. 2.5 Using Cloud Computing Technology to Share Educational Resources The collected and processed online education resources are stored in the modified cloud shared memory. Through the compression and transmission of resource data, the resource

Construction of Online Education Model

Service interface

Applicat ion code

Third party software

Identity authent ication

Account management

Usage bill ing

Task scheduling

Life cycl e management

task management

Fault recovery

Fault detect ion

Load balanci ng

Computi ng resource

205

Environment configuration

Acces s authori zation Com prehensi ve protection Security audit

Software resources St orage resources

Network res ource

Dat a resources

Fig. 5. Cloud computing technology block diagram

sharing is realized with the support of cloud computing technology. Figure 5 is a block diagram of cloud computing technology. The compression process of online education resources of marketing specialty can be expressed as follows: F=

(2x + 1)uπ 1 y(x) · cos 4 16

(4)

In formula (4), the variable u is the transformation scale. The teaching resources initially collected and processed are uploaded from the teacher side to the cloud server side. The upload process can be expressed as: H =ψ

μi Time (i) + (1 − ψ) μj Time (j)

(5)

In formula (5), ψ is the proportional coefficient, Time (i) and Time (j) respectively represent the upload transmission time of resources i and j, and μi and μj correspond to the channel selection coefficient [8]. Student users can apply to the cloud server for sharing. After the application is passed, a communication connection is formed between the server and the user, and finally a sharing task of educational resources is realized through reverse transmission. 2.6 Design of Online Education Activities for Marketing Major The implementation of the activity is mainly divided into clarifying the activity content and requirements, collecting and sorting learning activity resources, discussion, display

206

L. Mao and Z. Xu

and exchange of discussion results, classroom summary and other steps. Teachers make clear to students the teaching arrangement, the organization of learning activities, how to carry out group activities, specific tasks, relevant evaluation methods, etc. Students collect resources according to their division of labor. Collect and sort out the required data by consulting books, the Internet, questionnaires, interviews and other means, and complete the tasks of information screening, classification and analysis. Teachers should give appropriate guidance to students’ online learning activities and guide students to learn to collect and process information. At the same time, students can feed back information to teachers and exchange opinions with other groups at any time, so that teachers can solve the problems of students in the learning process in time. In the process of discussion, teachers should pay attention to appropriate guidance, stimulate students’ positive thinking ability, and cultivate students’ questioning ability, debate ability and expression ability. Teachers can prompt, ask and ask questions, stimulate students’ interest and self-confidence, and enable students to participate in activities efficiently. 2.7 Marketing Professional Online Interaction Online interaction of marketing major is to watch and learn courseware in the network environment. After logging in the model, the user clicks the courseware on demand function to first display the list information of the courses selected by the user. The user can enter the learning forum for learning exchange, download simulation questions or directly click to play the courseware. When ordering a lecture courseware, you first need to obtain the information related to the courseware, such as the course, knowledge point and playback path of the courseware. Go to the background to verify whether the courseware is valid according to the obtained information. If it is valid, it is allowed to order, otherwise it is rejected [9]. When the course on demand request is sent out, the server side processes the user request and returns it to the user with JSP page to realize the jump from the business function interface to the course on demand interface. At the same time, the corresponding entity records the click event: the current time and the number of on-demand courseware are accumulated once, and the learning time length of these courseware is tracked. The learning record data obtained from the page is parsed into entities that can be recognized by the model, and stored in the database through background processing. This data is used to track and feed back students’ progress, and can also evaluate students’ learning quality.

3 Model Test and Experimental Analysis In order to test the educational function and operation performance of the online education model of marketing specialty based on cloud computing, a model test experiment is designed. Cloud computing is built by the Open Stack framework and provides a Web-based visual online education interface with the help of Horizon components provided by Open Stack. Through the access interface provided by Open Stack, log in as an administrator. After successful login, enter the cloud management platform to realize the functions of user creation, resource sharing, online teaching and so on. To enhance the validity of the results, ensure that the control and experimental groups of the same environment.

Construction of Online Education Model

207

Before starting the experiment, it is necessary to prepare the course information, user information and test task information of marketing specialty. The setting of some test task information is shown in Table 2. Table 2. Model test task information setting table Task number

Test content

Anticipate result

1

Educational curriculum selection and participation

Successfully play educational courseware

2

Educational resources query

Output educational resources search results

3

Educational resources download

Educational resources successfully downloaded to local

4

Upload educational resources

The ECS and cloud shared memory store corresponding educational resources

5

Put forward the learning problems of marketing major

The learning questions are successfully displayed in the interactive interface

According to the model test task information table in Table 2, the total number of test task information was set to 800 and divided into 8 groups on average to avoid the influence of accidental events on experimental results. In addition, in the model test experiment, the number of generated users is 400, and through the control of the number of users online, the constraint of the model test environment conditions is realized. The quantitative test indicators set in the model test experiment include the success rate of educational function operation, response time and memory consumption. The numerical results of the success rate of educational function operation are as follows: λ=

nsucc × 100% nall

(6)

In formula (6), nsucc and nall respectively represent the number of successful samples and the total number of samples set in the experiment. By comparing the model output results with the expected effect of the setting, determine whether the current task is running successfully, and obtain the specific value of nsucc through the statistics of the number of tasks. The numerical results of response time index are: T = tsr − tout

(7)

In formula (7), tsr and tout are the start time of educational task and the output time of result respectively. In addition, the memory consumption can be directly read from the background data of model operation to obtain the test result data. In order to ensure the application value of the designed online education model in the actual teaching of marketing specialty, it is required that the success rate of function operation shall

208

L. Mao and Z. Xu

not be lower than 95%, the corresponding time shall not be higher than 8000 ms, and the memory consumption shall not be higher than 50% of the total memory. Since the total memory in the model test experimental environment is 2.0 GB, the total memory consumption shall not be higher than 1.0 GB. In the built experimental environment, the design model is converted into program code, and the implementation results of the functional tasks of the educational model are obtained through the operation of the computer and the input of the design and test tasks. Figure 6 shows the output results of educational course selection and adding test tasks.

Online education model of Marketing Specialty Current location: Home > > course content

Educational content of Marketing Specialty

Chapter I Introduction to marketing planning Chapter II marketing planning process Chapter III marketing strategy training Unit 1 purpose of marketing planning Unit 2 market analysis Unit 3 Analysis of market opportunities and problems Unit 4 determine the specific marketing plan Unit 5 implementation and control of marketing planning scheme

Unit 4 determine the specific marketing plan

Teaching day standard: understand the contents and methods of product planning, new products, product portfolio, trademark and product portfolio planning, and understand the concepts and contents of trademark planning and packaging planning

Fig. 6. Schematic diagram of online education model of Marketing Specialty

According to the schematic diagram of online education model for marketing major in Table 6, the operation output results of other educational function tasks can be obtained similarly. The operation status data in each group are counted to obtain the test results reflecting the model function, as shown in Table 3. Table 3. Functional test results of professional online education model Experimental group

Total number of tasks/piece

Number of successfully run tasks/piece

1

100

100

2

100

100

3

100

98

4

100

99

5

100

100 (continued)

Construction of Online Education Model

209

Table 3. (continued) Experimental group

Total number of tasks/piece

Number of successfully run tasks/piece

6

100

7

100

97

8

100

100

95

By substituting the data in Table 3 into formula (6), it can be concluded that the average functional operation success rate of the design model is 98.6%, which is higher than the preset value, that is, it meets the functional design requirements of the model. In addition, through the calculation of formula 7 and the statistics of relevant data, the test results of model operation performance are obtained, as shown in Fig. 7. 1000 Educational function response time Model running memory usage

900 800

Response time / ms

8000

700 600

6000

500 400

4000

300

Maximum m emory usage / MB

10000

200 2000 100 0

50

100

150

200

250

300

350

400

Number of concurrent users / Piece

Fig. 7. Performance test results of online education model

According to Fig. 7, as the number of concurrent users increases, the response time and memory footprint of the model run increases. It can be seen intuitively from Fig. 7 that the maximum response time is 7400 ms and the maximum memory consumption is 830 MB, so it can be seen that the operation performance of the designed online education model meets the design and application requirements.

4 Conclusion Based on the basic theoretical knowledge of marketing planning, the marketing planning course guides students to understand the marketing environment and the methods, strategies and steps of marketing planning, so as to form a modern marketing planning concept and strengthen students’ professional skills and professional quality. The average functional success rate of the designed model is 98.6%, which is higher than the

210

L. Mao and Z. Xu

preset value. The maximum response time of the model is 7400 ms, and the maximum memory consumption is 830 MB, which meets the functional design requirements of the model. This paper analyzes the application of cloud computing technology and the construction of education model, in order to provide some reference value for academia and industry. The next step is to make a detailed and in-depth study on how to effectively use the online education model of marketing major based on cloud computing to play an important role in students’ learning, so as to better serve teachers and students. Fund Project. Exploration and practice of online and offline hybrid teaching of management courses under the Internet and background (YGL201842).

References 1. Zurawski, S.A., Pickett, K.A., Widmer, M.: Expanding OT’s role in the mental health treatment of Parkinson’s disease through high-quality online education. Am. J. Occupat. Therapy 75(2), 275–282 (2021) 2. Harrison, D.E., Ajjan, H.: Customer relationship management technology: bridging the gap between marketing education and practice. J. Market. Anal. 7(4), 205–219 (2019) 3. Weber, M.M., Larkin, D.J., Patrick, M.: Creating informed consumers of aquatic invasive species management programs through online education for nonprofessionals. Invasive Plant Sci. Manage. 15(1), 41–48 (2022) 4. Ag, A., Ma, B., Mpm, C., et al.: Four questions of entrepreneurial marketing education: perspectives of university educators - ScienceDirect. J. Bus. Res. 113, 189–197 (2020) 5. Jing, D.A.: Cloud computing database and remote system for real-time image acquisition application in english classroom teaching - ScienceDirect. Microprocess. Microsyst. 82(6), 1–10 (2021) 6. Wu, Z.Y.: A secure and efficient digital-data-sharing system for cloud environments. Sensors 19(12), 2817 (2019) 7. Wang, S., Mu, M.: Exploring online intelligent teaching method with machine learning and SVM algorithm. Neural Comput. Appl. 6, 1–14 (2021) 8. Dolighan, T., Owen, M.: Teacher efficacy for online teaching during the COVID-19 pandemic. Brock Educ. J. 30(1), 95–116 (2021) 9. Xu, N., Fan ,W.-H.: Research on interactive augmented reality teaching system for numerical optimization teaching. Comput. Simul. 37(11), 203–206, 298 (2020)

Design of Tibetan Vocabulary Online Learning System Based on Multi-terminal Integration Min Li1(B) and Nan Li2 1 School of Literature, Capital Normal University, Beijing 100048, China

[email protected] 2 Boda College of Jilin Normal University, Siping 136000, China

Abstract. Provide auxiliary tools for Tibetan vocabulary learning, and solve the problems of long response time and poor application performance of existing online learning systems. Using the multi-terminal fusion technology, the optimized design of the Tibetan vocabulary online learning system is realized from the three aspects of hardware, database and software functions. Adjust the connection mode of the server, and modify hardware devices such as embedded processors and resource collectors. Use the optimized system circuit to connect hardware devices to complete the optimization of the hardware system. Collect Tibetan vocabulary online learning resource data, and connect each database table according to the logical relationship between the data. With the support of hardware devices and databases, set permissions for different users and choose the online learning mode of Tibetan vocabulary. Use multi-terminal fusion technology to achieve resource integration. Realize system functions such as Tibetan vocabulary learning information retrieval, online interactive practice, and online testing. Through system testing, it is found that the operating success rate of the designed system reaches 99.2%, and the response time is less than 6000ms. And through the application of the design system, the students’ Tibetan language scores have been significantly improved. Keywords: Multi-end integration · Tibetan vocabulary · Tibetan learning · Online learning system

1 Introduction The Tibetanlanguage is the Tibetan branch of the Tibeto-Burman family of the SinoTibetan language family. It is distributed in 5 regions in China’s Tibet Autonomous Region and Qinghai Province, Ganzi Tibetan Autonomous Prefecture in Sichuan Province, Aba Tibetan and Qiang Autonomous Prefecture, Gannan Tibetan Autonomous Prefecture in Gansu Province and Diqing Tibetan Autonomous Prefecture in Yunnan Province. Tibetan is mainly divided into three major dialects: U-Tsang dialect, Kham dialect, and Amdo dialect [1]. Tibetan areas have a large proportion of the Tibetan population and a high proportion of farmers and herdsmen. The transportation here is inconvenient and education is backward. In contrast to ordinary Chinese-based distance © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 211–228, 2022. https://doi.org/10.1007/978-3-031-21161-4_17

212

M. Li and N. Li

education, minority education in Tibetan areas faces language barriers. Tibetan language belongs to the Tibeto-Burman language family of the Sino-Tibetan language family. In addition to the Tibetans in my country, some people in Nepal, Bhutan and India also speak Tibetan. Tibetan language is a hot topic of linguistic research at home and abroad. Vocabulary is the basis for analyzing and understanding sentence structure and meaning. Learning vocabulary is beneficial for learning the Tibetan language. Therefore, online learning of Tibetan vocabulary is also very active, and it is getting more and more attention. In order to solve the teaching problems of students in Tibetan areas, an online learning system for Tibetan vocabulary was designed and developed. In view of the current needs of education in Tibetan areas and the inevitability of social development, develop and build a Tibetan/Chinese online learning system to adapt to my country’s national conditions and social development. It plays an extremely important role in improving the educational situation in Tibetan areas. The accumulation of Tibetan vocabulary plays a key role in improving the level of Tibetan language ability. Vocabulary is one of the basic elements of language. Vocabulary can help people effectively carry out all communication activities, such as listening, speaking, reading, and writing. It can be said that the level of a person’s Tibetan language proficiency often depends to a certain extent on the amount of vocabulary the person has mastered. Lexical meaning is the basis for navigating the entire semantic system, so an online learning system and model for Tibetan vocabulary is designed. It is not only a foundational knowledge project, but also a very important basic theoretical research. With the development of the Internet, online education has become more and more popular. Online education refers to the method of content dissemination and rapid learning through the application of information technology and Internet technology. Compared with traditional education, online education has the characteristics of high efficiency, convenience, low threshold and rich teaching resources. Based on the above characteristics, coupled with the promotion of “Internet+”, online education platforms have emerged, and their scale has gradually opened up, and has won the favor of the capital market. Modern distance education is a new educational mode that uses multimedia technology and computer network to achieve autonomous and interactive learning. However, in the current distance education platform, due to the lack of technical means and unified standards, different platforms use different document formats and develop their own resource systems independently. System platforms, data structures, databases, type definitions, resource descriptions, and the final storage form of resources vary widely. This has resulted in the current situation that online resources are scattered, information islands and resources are not standardized. In this situation, although there are a large number of educational resources, it is difficult to reuse and share resources due to the opaque format of resource files and the inability to extract information. Moreover, it is difficult or even impossible to modify and maintain the resources after the resources are released, let alone to effectively retrieve the relevant resource information, all of which result in a great waste of resources. The fundamental reason is the lack of a unified standard and effective means to describe resource metadata and its content structure. In order to solve the operation problems of the current Tibetan vocabulary online learning system, multi-terminal fusion technology is introduced.

Design of Tibetan Vocabulary Online Learning System

213

Multi-terminal fusion is a technology that fuses resource information of multiple different ports. In order to realize the effective integration of Tibetan vocabulary information, reduce the coupling degree of system modules, and enhance the maintainability and flexibility of the system. In order to improve the teaching function and application performance of the Tibetan vocabulary online learning system.

2 Design of Hardware System for Online Learning of Tibetan Vocabulary 2.1 Server The optimally designed Tibetan vocabulary online learning system sets up three different types of servers, including registration server, proxy server and redirection server. The registration server can know the system users registered in its area in time, and the proxy server is responsible for receiving the request sent by the user agent. Send the request to the corresponding server according to the network policy, and respond to the user according to the received response. Proxy servers are mainly used to forward SIP messages. It initiates requests on behalf of UAC and returns responses to UAC, and is a SIP entity that acts as both a client and a server. Each domain of the Tibetan vocabulary online learning system has a corresponding proxy server. When the user agent called by the calling party is in the domain controlled by the proxy server, the proxy server queries the registration server, finds the address of the called party, and forwards the request to the called party. When the user agent called by the calling party does not belong to the domain controlled by the proxy server, the proxy server queries the proxy server address of the domain to which the called party belongs. And route the request to the next-hop proxy server address until it is forwarded to the proxy server of the domain to which the called party belongs [2]. The redirect server is used to return the user’s new location to the caller when needed. Redirect servers help locate SIP user agents by providing selectable locations that connect to the user’s location. It does not emit any behavior to locate the target agent. And only return the possible location information of the target agent, and the caller can call again according to the new location obtained. 2.2 Embedded Processor The optimally designed Tibetan vocabulary online learning system based on multiterminal fusion selects the C*CORE model processor to replace the traditional processor equipment. The processor executes instructions with a 4-stage pipeline: instruction fetch, instruction decode/register file read, execute, and register file write-back. These 4-stage overlapping operations allow most instructions to be executed in a single clock. The structure of the C*CORE processor is shown in Fig. 1.

214

M. Li and N. Li Address occurrence

System bus

General register file 32 bits * 16

Address selector Immediat e selector

Address bus

Branch jump adder

Alternate regist er file 32 bits * 16

Cont rol register file 32 bits * 16 X port

Y port Dat a adjustment

PC sel f adding

Barrel shift mult iplier and divider

Instructi on pipel ine

selector

Sym bol extensi on

selector

Instructi on decoder Adder / logic priority decoder / zero det ection resul t selector Writ e back bus Data bus

Universal port

Fig. 1. C*CORE model processor structure diagram

The C*CORE processor adopts a pipelined RISC structure and uses the AMBA bus as the carrier, which accommodates many of the same performance enhancement measures and specific implementation technologies as the desktop RISC processor [3]. The processor adopts a 32-bit load/store reduced instruction set computer architecture with a fixed 16-bit instruction length. The strict load/store structure eases the control complexity. Using a fixed 16-bit instruction encoding technique significantly reduces the memory bandwidth required to maintain high-speed instruction execution. The relatively short 16-bit instruction code achieves the purpose of reducing memory power consumption [4]. The data selector in the processor can select a specified combinational logic circuit from a set of input signals and send it to the output terminal according to the given input address code. Multiple data selectors can ensure efficient system processing efficiency when processing a large amount of Tibetan language information. The carefully selected instruction set makes the code density and overall memory efficiency of the C*CORE structure exceed that of the CISC structure. In addition, C*CORE also adopts measures such as fully static design, dynamic power management and low voltage operation to reduce power consumption. Dynamic power management includes two measures: dynamic clock and low power consumption. C*CORE uses dynamic clock management method to automatically power down the internal function modules that do not need to operate clock by clock. C*CORE has 3 low-power operation modes, namely Wait, Doze and Stop, which are entered by executing corresponding instructions.

Design of Tibetan Vocabulary Online Learning System

215

2.3 Tibetan Vocabulary Learning Resource Collector In the process of collecting and sorting out, the Tibetan language is divided from three aspects: the distribution of the number of Tibetan entries, the frequency and frequency of the vocabulary. The function of this division is mainly in two aspects: one is the construction of a Tibetan vocabulary corpus as a computer Tibetan information processing system to serve Tibetan information processing. The second is to serve as a reference book for Tibetan language learning and a basic resource for Tibetan language research, for Tibetan language learners and researchers. The structure of the optimally designed Tibetan vocabulary learning resource collector is shown in Fig. 2.

DDR II Microsecon d signal

Voltage type sens or

Ethernet module

FPGA

Transmi tting and recei ving module UATR modular

Current type sens or FLASH

Fig. 2. Structure of Tibetan vocabulary learning resource collector

As can be seen from Fig. 2, FPGA is the core element in the collector. The component consists of programmable input and output units, configurable logic blocks and other parts. The main task of the programmable input and output unit is to meet the driving and matching requirements of the input/output signal. For this purpose, the I/O in the FPGA is divided into different type groups. These type groups meet different electrical characteristics and can be flexibly configured in the software. The storage unit of the programmable input and output unit module introduces signals into the interior of the FPGA. At this time, the holding time of the external signal can be 0 by default. There are many groups of IOBs in FPGA, and the interface standards corresponding to banks of different interface voltages are also different. There is one and only one interface voltage for the same bank, but the VCCOs of different banks can be different [5]. To facilitate management and adapt to multiple electrical standards, only ports of the same electrical standard can be connected together. The configurable logic block is the basic logic unit in the FPGA. The number and characteristics of CLB vary with different chips, and the same place is that there is a configurable switch matrix consisting of 4–6 inputs, selection circuits and flip-flops. At the same time, the flexibility of the matrix is also

216

M. Li and N. Li

very high, and it can better handle combinational logic, shift registers or RAM through its configuration. In addition to the need for each CLB module to be implemented as combinational logic and sequential logic. 2.4 System Circuit Design Power Circuit The 5 V DC power supply is selected, and the 5 V input DC power supply generates a stable 5 V output voltage through a special voltage regulator. Then the obtained 5 V stable voltage is converted into the required 3.3 V through the corresponding voltage regulator. The optimized design result of the power supply circuit is shown in Fig. 3. D2

220V

+

SCR D1

R3

G N D

R1

U1 7805 Vout 3 Vin

C1

+ 2

RL

1

C2

R2

C3 R4

Fig. 3. System power circuit diagram

Through the optimization of the power supply circuit, stable power support can be provided for different hardware devices in the hardware system. Clock Design The whole high-speed acquisition system is completed under certain timing control. If the timing is wrong, the function of the system cannot be realized. The maximum sampling rate of the system is 32MSPS. In the optimally designed multiple interactive sharing system of electric power special teaching course resources, the highest sampling frequency of AD converter is 32MSPS. The clock frequency required by SDRAM is 200 MHz, so the system clock is selected as 50 MHz. Use the PLL in the FPGA to achieve the clock frequency required by each module by dividing, multiplying or shifting. The system clock is realized by a 50 MHz active crystal oscillator. Compared with the passive crystal oscillator, the clock signal output by the active crystal oscillator is more stable and of good quality.

3 Database Design of Tibetan Vocabulary Online Learning System Database design is an important part of Tibetan vocabulary online learning system design. While the design of the database table follows the database theory, it must be able to use development tools to realize the functional requirements put forward by users in

Design of Tibetan Vocabulary Online Learning System

217

various aspects. When designing the database, it is necessary to minimize the hardware overhead and maximize the running speed, efficiency and performance stability of the system. The background of the system adopts MySQL database, which is connected to the database through JDBC. There are 11 data tables in the more important tables in the database. They are curriculum table, vocabulary table, video table, course classification table, etc. According to the overall functional structure of the system, comprehensively analyze all the required data objects and their access plans and structures, and design the background database. Designing a database means creating an optimal database schema and establishing a database and its application system under a given application environment. It can effectively store data and meet the application needs of users to the greatest extent. Conceptual model is the core of database design. It uses entity-connection method to accurately complete the modeling of the information world. And because this model is closer to the way of human thinking, it is easy to be accepted by users. When solving practical application problems, an entity-relationship model should usually be designed first, and then converted into a data model. According to the logical relationship between the various entities of the learning system, the system database table is designed. Taking the Tibetan vocabulary data table as an example, the construction results are shown in Table 1. Table 1. Tibetan vocabulary data sheet Field name

Field description

Type of data

Is it nullable

Tibetan_id

Vocabulary ID

Integer(8)

No

Tibetan_spell

Vocabulary spelling

Text(20)

No

Tibetan_yinbiao

Vocabulary Phonetics

Text(6)

No

Tibetan_meaning

Vocabulary definition

Text(20)

No

Tibetan_picture

Vocabulary pictures

Blob(200)

Yes

Tibetan_example

Vocabulary example

Text(100)

No

Tibetan_voice

Vocabulary pronunciation

Blob(120)

No

Similarly, the construction results of other information tables in the database can be obtained. And according to the logical relationship between the information, complete the design of Tibetan vocabulary online learning system database. In addition, the Tibetan vocabulary learning system is running in real time, so the data in the database table needs to be updated in real time.

4 Software Function Design of Tibetan Vocabulary Online Learning System With the support of hardware equipment and database, through the design of multiple modules such as user classification and management, learning mode determination,

218

M. Li and N. Li

learning resource collection and processing, etc. Completed the design of the software functions of the Tibetan vocabulary online learning system. 4.1 System User Design The users of the Tibetan vocabulary online learning system can be divided into three categories: ordinary students, teachers, and system administrators. The management of system users is mainly used to ensure that users can access the system within a certain authority and perform legal operations. In this module, executable operations include user registration application, data review, and permission settings. After filling in the form and submitting the system, it will be reviewed by the system administrator, and the registration will be successful after passing. The flowchart design of the user management module is shown in Fig. 4. start User login Enter user name and password N

Is the login information correct? User identity judgment

Y

Is it a student user? Y Student us er interface

N

Is it a teacher user? Y Teacher user interface

N

Are you an administrator user? Y Adminis trator user interface

N

Login succeeded End Login failed

Fig. 4. User management module flow chart

Student users can register on this website through their mailboxes. After registration, users can enter the system learning platform according to the registered mailboxes and passwords to perform related operations. Student users can view the four modules of course, community, article, and personal center on the homepage. In the course module, users can view course information and comment information, and join courses of interest according to their interests. The community module includes questions asked by other users, and users can view and answer questions from other users. In the article module, users can view articles written by other users, and users can also publish articles. In

Design of Tibetan Vocabulary Online Learning System

219

the personal center module, users can modify personal information and passwords, and can view the user’s courses, articles, concerns, questions, etc. Teacher users can register on this website through email. After registration, teacher users can enter the system learning platform according to the registered email address and password to perform related operations. The functions of teacher users include personal center management, course management and other operations. In the personal center is the maintenance of personal information and the maintenance of my articles. Course management includes the maintenance of video information and comments, video transcoding and statistics on student learning. Administrator users include operations such as user management, feedback management, message push, course management, article management, and community management. 4.2 Choose Tibetan Vocabulary Online Learning Mode The learning mode of the optimally designed Tibetan vocabulary online learning system is a blended learning mode, which can be divided into three modules: front-end analysis, teaching activities and learning evaluation. According to the needs of blended learning, the teaching process can be reasonably designed by analyzing the characteristics of learners, learning objectives and learning content. Learner Analysis: Analyze the characteristics of learners, existing learning level, learning habits, learning ability and learning autonomy. Analysis of learning objectives: What kind of objectives does teaching need to make learners achieve, and the results are clear, specific, and quantifiable. This is the guidance of teaching work and an important indicator of teaching evaluation. Analysis of learning content: Closely combine the knowledge learned by students with the characteristics of students, stratify the teaching content, and pay attention to the order, structure and rationality of the teaching content. The design of teaching activities in the blended learning model is mainly divided into three steps. First, the design of pre-class teaching activities: lets students learn through the intelligent learning platform before class, and for the content that they do not know. Or the pronunciation of Tibetan characters can be learned through the speech synthesis system on the intelligent platform, and questions and feedback can be given. At the same time, teachers can monitor students’ learning by viewing the learning records. Then carry out the design of teaching activities in the class: on the basis of traditional teaching, teaching is carried out according to the learning situation of students before class. And answer the questions based on the questions of the students before the class. Finally, the design of after-school teaching activities is carried out: teachers assign homework on the intelligent platform, and students complete it after class [9]. Teachers can also provide targeted and personalized tutoring according to students’ learning situations. The evaluation of students’ learning should be examined from multiple aspects. The learning mode realizes evaluation diversification through online and offline learning evaluation, including teacher evaluation, self-evaluation, and intelligent platform evaluation. Evaluation is not the ultimate goal, but to reflect the students’ learning status to motivate students’ learning status and improve learning efficiency. Its advantage is to provide guidance and help with students’ learning, so as to improve the learning effect.

220

M. Li and N. Li

4.3 Collection and Processing of Tibetan Vocabulary Resources Use web crawler technology to collect Tibetan vocabulary resources in the network environment, when the web crawler program starts. During initialization, a pre-set initial URL seed will be obtained from the seed queue. Transfer the obtained URL seed into memory for DNS domain name resolution. Then download the parsed IP address web page, and perform HTML parsing on the downloaded web page through the parser. Analyze the HTML tags and organization information as much as possible to obtain the address, title, author information, abstract, date and other information of the web page. And extract the new URL in the web page. Deduplicate new URLs and remove already crawled URLs, then add them to the queue of pending URLs. Perform the above steps again, starting with removing the top URL from the list of URLs to be crawled. The output result is the resource collection result of Tibetan vocabulary learning. On this basis, the principal component analysis method is used to reduce the noise interference signal in the initial information. Suppose the time series of Tibetan vocabulary information collected in real time is: x(t) = xˆ (t) + n(t)

(1)

In formula (1), xˆ (t) and n(t) are effective learning information and noise information. There are the following relations defined:  yj (t) = si xi (t) (2) Ay = yy2j Among them, si is the orthogonal transformation matrix, Ay is the covariance matrix of the main element, and the main element yj (t) is the projection of the delay vector on the characteristic vector. The time average of the pivot variance is the pivot of the reconstructed signal covariance matrix, which represents the projection of the signal with unequal energy in each direction. Then the signal-to-noise ratio at different rotation coordinates can be expressed as:   yj2 (3) η =  2 n A subset of principal components with a large variance corresponds to a large signalto-noise ratio. A subset of principal components with a smaller variance corresponds to a smaller signal-to-noise ratio. Reconstructing the state space using the q subsets of pivot elements with maximum variance results in a reconstruction with a greatly improved signal-to-noise ratio. That is, the noise reduction processing of the initial information is completed. In order to define vocabulary difficulty more precisely, it is necessary to determine the vocabulary difficulty determination formula. Using the three factor parameters of word frequency, length, and degree of harmony of phonetic writing, the proposed difficulty determination formula is as follows: bi = Fi × W1 + Li × W2 + Hi × W3

(4)

Design of Tibetan Vocabulary Online Learning System

221

The variables Fi , Li and Hi are the word frequency parameter, length parameter and harmony degree parameter of the i-th Tibetan vocabulary, respectively. W1 , W2 and W3 are the weight values of vocabulary parameters. Tibetan vocabulary is classified and stored according to its difficulty level. The above-mentioned Tibetan vocabulary resource collection and processing flow is shown in Fig. 5. Start

End

Using crawler technology to collect Tibetan vocabulary resources in the network

Determine vocabulary difficulty and store vocabulary by category

HTML parsing of collected web pages

Calculate the signal-to-noise ratio of signals with different rotation coordinates

Computation of Tibetan vocabulary information time series

Computation of effective learning information and noise information

Fig. 5. Collection and processing process of Tibetan vocabulary resources

4.4 Utilize Multi-terminal Fusion Technology to Integrate Tibetan Vocabulary Resources Based on the collected and processed Tibetan vocabulary resources, the fusion processing of vocabulary resources of different ports is realized from three aspects: data layer, feature layer and decision layer. Data layer resource fusion refers to the fusion directly carried out on the collected original data layer, and the unprocessed data obtained by each information source is synthesized and analyzed. Usually, a fusion method for unified operation of information resources is adopted. The information fusion of the feature layer is an intermediate link compared with the fusion of the data layer. The fusion of the feature layer first extracts the key features from the massive data information that has been widely collected before. And according to these characteristics, the information is scientifically organized and divided [10]. This level of fusion compresses the total amount of information through the feature extraction process performed in advance, helping users to quickly retrieve the required information according to their own needs during retrieval. At the same time, when browsing the information, it is more conducive to the user’s understanding and mastery. Figure 6 shows the principle of feature layer information resource fusion.

Source n

Source 2

M. Li and N. Li

Source 1

222

Dat a Associat ion

Dat a level resource fus ion

Feature level resource fusi on

Attribute judgment

Decision l evel resource integrati on

Fig. 6. The principle of feature layer information resource fusion

The information fusion of ASEAN information resources at the decision-making level is to coordinate the data of each document information source from a macro perspective. Through the analysis, extraction and fusion of data extraction features from different sources, it directly provides support for decision-making. 4.5 Retrieve Tibetan Vocabulary Learning Information Tibetan vocabulary learning information and course retrieval is one of the important functions of the online learning system. Enter the keywords to be searched through the online learning front-end interface. Use formula (5) to calculate the similarity between the input keywords and the vocabulary words stored in the system memory.  (5) sim = (x − x0 )2 In formula (5), x and x0 are the input keywords and the Tibetan vocabulary to be matched, respectively. If the calculation result of formula 5 is higher than 0.9, the corresponding Tibetan vocabulary is directly output as learning information. Otherwise, it is necessary to match the next Tibetan vocabulary until the required learning vocabulary information is obtained. 4.6 Online Interactive Practice The practical links are also different, among which the practical links such as reading and writing mainly focus on textual information. Through the direct processing of text data,

Design of Tibetan Vocabulary Online Learning System

223

the interactive exercise function of the system is realized. Among them, listening and translation training is mainly based on students’ autonomy. To train students’ listening and translation skills, they can practice multi-theme, multi-knowledge and multimedia special skills. It has functions such as listening, translating and Chinese translation. In addition, in the process of oral pronunciation practice, the HMM model obtained by standard pronunciation training is used. The standard pronunciation here is selected from the matching pronunciation of the PEP textbook. At the same time, the corresponding language model is used for path search, so that the students’ pronunciation is forced to align on the basis of the text. If the alignment result cannot be recognized, it means that the student’s pronunciation content is too different from the target content, or the noise interference is too large. Then the grading or correction work at that time is meaningless, and the module will directly prompt the students to re-pronounce. By calculating the log-likelihood of each phoneme and weighting, the pronunciation score of each Tibetan word or phrase is obtained as the pronunciation score. 4.7 Online Study Test Students always want to test their knowledge by some means after completing a certain study or learning task. The traditional method of examination is the final examination of each semester or various entrance examinations, or even certificate examinations. But this method is impossible to pass in Tibetan vocabulary learning. In order to enable users to complete the self-testing task at any time, the relevant functional modules of the learning test are designed and implemented. This module can add various Tibetan vocabulary test questions in advance, and use these examples as a question bank. After learning, students can enter the self-test module, and the system randomly selects test questions to realize the online test of Tibetan vocabulary learning.

5 System Test The system testing process plays an important role in the guarantee of system function and performance, and can make the system more perfect. Therefore, after the system design and implementation are completed, it is necessary to conduct a comprehensive test on the function and performance of the system to determine whether the system meets the expected goals. This system test experiment will complete the test of the system from the aspects of function and performance, through the reasonable design of test cases, and carry out the detailed test process. In order to obtain accurate test results. First of all, it is necessary to configure the test environment of the system, mainly to connect the relevant hardware equipment and network environment. We set the operating value of the relevant parameters during the system operation, and complete the construction of the system development and test environment. On this basis, the specific operation of each functional module is designed by combining a variety of test methods. In-depth analysis of each test case is carried out to determine whether the test results of each test case meet the expected requirements. Some of the set system function test examples are shown in Table 2.

224

M. Li and N. Li Table 2. System functional test cases

Use case number

Test case content

Prediction effect

1

User login

The user logs in to the learning system with the correct user name and password, and the main interface of the system is displayed

2

Set user permissions

The administrator grants corresponding permissions to different users

3

Tibetan vocabulary learning content selection

Choose study courses according to the difficulty level of Tibetan vocabulary

4

Teacher users upload Tibetan vocabulary learning resources

The system database adds corresponding learning resources

5

Student users retrieve Tibetan vocabulary learning resources

The system outputs search results related to keywords

6

Join a study course

Display and play Tibetan vocabulary learning videos or courseware

7

Online interaction

Student users submit questions in the interactive module and get a response within 24 h

8

Online test

Obtain the test paper, submit the test paper results, and get the test result query results within one week

In order to ensure the credibility of the experimental results, a total of 500 test cases were set up in this experiment. And divide them into 5 groups equally. The optimally designed Tibetan vocabulary online learning system software based on multi-terminal fusion is written into program code and imported into the constructed system test environment. And then input the test cases in Table 2 to get the system function operation results. Figure 7 shows the test results of Tibetan vocabulary learning content selection and vocabulary query functions. Set the quantitative test index of the system function as the function operation success rate. The numerical results of this indicator are as follows: χ=

Nsuc × 100% Nall

(6)

Design of Tibetan Vocabulary Online Learning System

225

Fig. 7. Tibetan vocabulary online vocabulary system operation interface

In formula (6), the variables Nsuc and Nall are the number of use cases that the system function runs successfully and the total number of use cases set by the test experiment, Table 3. System function test results Experimental group

Nsuc /piece

Nall /piece

1

99

100

2

100

100

3

98

100

4

99

100

5

100

100

6

98

100

7

99

100

8

100

100

9

100

100

10

100

100

226

M. Li and N. Li

respectively. In order to meet the system design and requirements, the default value of the success rate of setting the function operation is 95%. And it is required that the calculation result of formula (6) shall not be lower than the preset value. Through the statistics of relevant data, the system function test results are obtained, as shown in Table 3.

Fig. 8. Statistical results of the success rate of the system operation

Substituting the data in Table 3 into formula (6), the result of the system operation success rate obtained by calculation is shown in Fig. 8. The average calculation result of the system function operation success rate is 99.3%, which is higher than 95%. That is to meet the system functional design requirements. Because this multimedia learning tool for English vocabulary focuses on the application of teaching, rather than the pursuit of commercial value in the enterprise. Therefore, the performance test is mainly tested from two aspects: running performance and application performance. The test index of the running performance is the response time, and the numerical results are: t = tstr − tend

(7)

In formula (7), tstr and tend are the start-up time and result output time of the system function, respectively. The functional response time of the designed online learning system is required to be no higher than 6000 ms. In addition, the application performance of the online learning system is mainly to compare the changes of Tibetan vocabulary learning performance of the students before and after the application of the design system. The test results can be obtained directly through the extraction and statistics of the students’ test data. After the calculation of formula 7 and the statistics of relevant data, the final test result of the system performance is obtained, as shown in Fig. 9. It can be seen intuitively from Fig. 9 that the maximum response time of the design learning system function is 5500 ms, which is lower than the preset value. And by designing the application of the online learning system, the average score of the research subjects improved by 8.6 points. It can be seen that the designed Tibetan vocabulary online learning system based on multi-terminal fusion has good performance.

Design of Tibetan Vocabulary Online Learning System 100

Student achievement without application system

90

Student achievement 6000 under application system 5000 Response time

85 4000

80 75

3000

70 65

2000

Response t ime/ms

Tibetan vocabulary test scores / points

95

227

60

1000

55 0

001

002

003

005 006 004 Student User ID

007

008

009

0

Fig. 9. System performance test results

6 Conclusion Tibetan is one of the minority languages in China. The quality and effect of language education is an important factor to maintain the diversity of Chinese nationalities. In order to facilitate the online learning of Tibetan, to solve the problems of low power and long response time. In this paper, an online vocabulary learning system based on multiterminal fusion is designed. Through the application of multi-terminal fusion technology, users can learn online at any time and any place, eliminating time and space barriers and reducing costs. With the support of hardware equipment and database, the system designed in this paper sets permissions for different users and selects online learning mode of Tibetan vocabulary. By virtue of its large amount of information storage, interactive and other characteristics, improve the efficiency of Tibetan vocabulary online education and learning. The success rate of the system is 99.2% and the response time is less than 6000 ms. The design of Tibetan vocabulary online learning system has made a basic arrangement for syntactic classification and tagging, automatic word segmentation, syntactic research, phrase research, machine translation, search engine and electronic dictionary compilation in the future Tibetan information processing field. It provides a new research method and means for the study of Tibetan literature in the future.

References 1. Li, M., Zhang, L.: Tibetan CSL learners’ L2 motivational self system and L2 achievement. System 97, 102436 (2021) 2. Sun, Y., Xia, T.: A hybrid network model for Tibetan question answering. IEEE Access 7, 52769–52777 (2019) 3. Khysru, Di, K., et al.: Morphological verb-aware tibetan language model. IEEE Access, 7, 72896–72904 (2019)

228

M. Li and N. Li

4. Sun, Y., Chen, C., Xia, T., et al.: QuGAN: quasi generative adversarial network for Tibetan question answering corpus generation. IEEE Access 7(99), 116247–116255 (2019) 5. Yang, H., Chen, J., Zhan, X.: Research on chinese education for preparatory students based on diversified development. In: Hu, Z., Petoukhov, S., He, M. (eds.) International Conference of Artificial Intelligence, Medical Engineering, Education. LNCS, vol. 107, pp. 503–515. Springer, Cham (2021). https://doi.org/10.1007/978-3-030-92537-6_46 6. Wang, S., Mu, M.: Exploring online intelligent teaching method with machine learning and SVM algorithm. Neural Comput. Appl. 6, 1–14 (2021) 7. Ho, W., Tai, K.: Doing expertise multilingually and multimodally in online English teaching videos. System 94(3), 102340 (2020) 8. Qin, S., Wang, L., Li, S., et al.: Improving low-resource Tibetan end-to-end ASR by multilingual and multilevel unit modeling. EURASIP J. Audio Speech Music Process 2022(1), 1–10 (2022) 9. Li, C.: Research on optimization and simulation of teaching resources equilibrium assignment in mobile network. Comput. Simul. 34(02), 238–241 (2017) 10. Mei Z.: The Recognition of tibetan handwritten numbers based on federated learning. J. Artific. Intell. Pract. 4(1), 1–12 (2021)

Design of Online Teaching System for Theory of Variable Order Fractional Gradient Descent Method Zhichao Xu1(B) , Chao Song1 , Li Li1 , and Lingli Mao2 1 Dalian University of Science and Technology, Dalian 116033, China

[email protected] 2 Guangzhou Huali Science and Technology Vocational College, Guangzhou 511325, China

Abstract. The functional modules of the online teaching system of healthy law theory are not perfect enough, and there are some limitations in practical application. Aiming at the problem of poor operation of online teaching system, this paper puts forward the theory of Variable Order Fractional step-down method, the design method of online teaching system, optimizes the system hardware structure, improves the operation performance of system hardware, and optimizes the function of system software. Combined with the theory of variable order fractional step-down method, the teaching management evaluation algorithm is realized. Finally, it is confirmed by experiments, The online teaching system of Variable Order Fractional gradient descent method theory has high practicability in the process of practical application, and fully meets the requirements of system design. Keywords: Variable order fractional step · Online teaching · Teaching system

1 Introduction At present, with the rapid development of modern educational technology, most colleges and universities begin to use network communication technology and multimedia technology to carry out teaching, and various teaching activities based on network environment are gradually carried out [1]. Compared with the traditional teaching mode, online teaching has great advantages in time, space and content. It not only provides rich teaching resources, but also provides a communication system between teachers and students. At the same time, it also improves students’ learning enthusiasm and ensures the learning effect. At present, the network has penetrated into all aspects of social life and played a great role. In college teaching, Network online teaching system is a supplement to the traditional “face-to-face teaching”, which expands the space and time of traditional teaching, prolongs its teaching process and improves the teaching effect [2]. The network online teaching system also makes full use of the teaching resources in the cyberspace, greatly meets the needs of students and effectively serves the teaching. At the same time, it can also use big data, big algorithms and other technologies to provide students with personalized learning content and learning forms, so that students © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 229–242, 2022. https://doi.org/10.1007/978-3-031-21161-4_18

230

Z. Xu et al.

can experience a new teaching mode and provide students with a more free way of autonomous learning. As teachers and students increasingly use the Internet for teaching, the traditional concept of office hours is gradually disappearing. Teachers in schools use online teaching systems to assign homework to students. Many students find it more convenient and free to keep in touch with their teachers in this way. By asking questions without face-to-face contact, students feel more at ease. Students now live in a world of computer networks. In fact, students don’t even need to set foot in university to complete the classroom knowledge, TV or distance learning has been used in continuing education and remote areas for more than 10 years. With the progress of technology and the development of Internet, online assisted teaching will become the mainstream of development. Therefore, the online teaching system based on the theory of step-down method of variable grade is designed. The system has better performance and can better serve teachers and students.

2 Online Teaching System of Variable Order Fractional Gradient Descent Method Theory 2.1 System Hardware Configuration The system design is mainly to realize online teaching. The system uses WebService under dotnet to provide system interface and provide application interface for system integration and data integration in Colleges and universities. The front-end implementation of the system requires generous and beautiful interface, friendly and convenient user operation. The main objectives of the system design are as follows: to meet the functional requirements of the online teaching system [3]. Students can easily enter the front end of the system and carry out information browsing, online examination, online communication, homework submission and other functions; The function of system design should be simple, easy to use and practical [4]. The excessively complex system will bring users into a misunderstanding. The goal of this system design is to adopt the current general online teaching process to realize the system operation, which is easy to use, simple and powerful; System design should pay attention to the principles of system scalability and reusability; Based on the functional requirements of the system design, the system network deployment structure is optimized, as shown in Fig. 1.

Design of Online Teaching System for Theory

231

Router

Web server 1

Producti on databas e server

Web server 2

Distribu ted cache databas e

Tes t databas e server

SQA interfac e server

firewall

Extranet

Fig. 1. System network deployment structure

In the process of building the system structure, the application mainly includes server-side and client-side. Therefore, the following types of development tools need to be applied. The development environment is usually based on Windows 7, and the

Data server

application server

Switch

Router

Pad Wireless routing PC IMac des ktop

firewall

internet

Extranet

Fig. 2. System hardware structure setting

Mobile Phone

232

Z. Xu et al.

version uses 64 bit DK16. The mobile client mainly uses the Android system [5]. The development server mainly applies the 12ef development version, and the database uses the sqlserver2010 version. In the specific development process, the specified server is not proposed, but the corresponding server system is built through the laptop. In this way, the running environment is simulated [6]. At the same time, the wireless routing is simulated with the help of the shared network and provided to the mobile terminal to realize the access to the network. With the help of this local area network form, the server is simulated by laptop, and the network teaching system is developed and debugged. The system adopts B/s network architecture, and the network architecture diagram is shown in Fig. 2. The online teaching system is developed with B/S structure (B/S structure, i.e. browser/server mode). It is a development mode of web application. Most clients use the browser as the main application software for browsing web pages. By using the existing server software, the client development is simplified, so that the core of the system function realization is concentrated on the server system. Users log in to the system through the Internet and use personal computers, mobile phones and other terminals to access the system interface for data interaction [7]. In the process of information exchange, the user uses the browser to connect with the server background, and the data is encapsulated into data packets of different protocols. The client browser and the server background interact with each other through different data packets. In the process of data exchange between the client and the server, if there are information security requirements, You need to use a network firewall to ensure data transmission and data security of the user’s computer. 2.2 System Software Function Structure Through the research on the key points and difficulties in the construction of online education information resource system based on campus network, the functional design of the system aims to meet the requirements of online teaching, and can fully interact, have knowledge base construction and feedback support. There are several modules such as course information management, audio and video management, online testing, interactive Q & A and resource center discussed in the above chapter, The composition is shown in Fig. 3. Teaching resources are an important part of the online teaching system. Students can learn according to the teaching resources released by teachers. The materials are rich, which makes students’ learning vision broader. The main purpose of the resource center is to break the restrictions of time and space, realize resource sharing, and enable teachers and students to broaden their horizons and learn more. Users can learn from each other and make progress. Teachers upload courseware resources of relevant courses. Student users browse courseware resources of courses through the main page of the system for downloading and online browsing. The resource center is divided into resource management, resource download and resource retrieval. Based on the above business processes, the requirements of the system are analyzed. Based on the theory of Variable Order Fractional gradient descent method, a teaching system is established, including teaching management and terminal [8]. The teaching management subsystem mainly corresponds to the service terminal of the system, while

Design of Online Teaching System for Theory

Cl assroom information management

233

course management Teacher management video on dem and

Audio and video management

Video teachi ng Teaching feedback

Online teaching system

Quest ion bank upl oad Online test

Random test Eval uati on feedback Real time interaction

Interacti ve Q & A Message discussion Resource upload Resource Center Resource download

Fig. 3. Functional structure model of system software

the application of the terminal corresponds to the system client. The interface of the teaching management subsystem to teachers and administrators is mainly displayed on the basis of the interactive interface, which fully reflects the main functions of the interactive layer, and also shows the Moodle interface. The entrance of the teaching management subsystem is mainly the login interface. Users enter different personal details, and then verify the information. When the verification is correct, the next operation can be carried out. In the theory of Variable Order Fractional gradient descent method, there are two main options for users in their specific application. Users can use it online or offline [9]. The difference between the two application methods is that when online, it is a direct connection with the server, and only enter the user name and password. After verifying the login information, the user starts to apply the corresponding functions. The overall demand analysis results of the system are shown in Table 1, which is divided into 9 functional modules.

234

Z. Xu et al. Table 1. Demand analysis

Serial number

Modular

Function

A

Account management

Distinguish user categories, etc.

B

Personal information management

User personal information maintenance

C

Programme Management

Teaching program management

D

Course management

According to different technical management courses

E

Training project management

Integrate into CD10 teaching mode

F

Teaching resource management

Including the overall management of teaching resources, video and other teaching resources

G

Teaching management

Teachers create classes and set directions for classes

H

Class management

Create classes and assign classes to students

I

System management

System data backup, etc.

In order to make teaching more meaningful, students evaluate and feed back the teaching of this course to teachers, score teachers’ comprehensive performance, and get the corresponding evaluation results. Students make comments and suggestions on the problems in learning, and the teacher will reply after seeing the comments. According to the evaluation feedback, teachers modify their own teaching courseware, improve teaching methods, let students improve learning efficiency, and achieve the purpose of modern network online teaching. 2.3 Realization of Online Teaching of Variable Order Fractional Gradient Descent Method Theory When designing a series of functions proposed by the online teaching system, the first is to select the development system and development framework selected by the system, and adopt a series of mature system components to speed up the development efficiency. After completing the above framework construction and component selection of the system, the next step is to code. Here, we need to design the common modules of the system first. The system adopts the general 3 + n design idea, that is, data access layer, business logic layer and user interface layer. Each layer is designed and divided into sub layers according to the complexity of specific functions. First, abstract the base class of data access operations commonly used in the system, and all business logic calls the data access layer to complete the required logical operations. The front-end page of the system adopts the theory of Variable Order Fractional gradient descent method for page design and layout. The system information display adopts the process of publishing from the background and displaying at the front-end, and the management and operation of

Design of Online Teaching System for Theory

235

information in the background. In the system, the information design of the system is shown in Fig. 4.

Browse news

System administrator

Release news and information

Login successful

Login system

News information base

Browse news

Y N

Student / teacher

Login successful

Display system

Y

N

Fig. 4. Online teaching information management process

Software security is to take measures from the perspective of code to maintain the security of online teaching system and prevent illegal users from entering online teaching system. Therefore, the system introduces identity verification function and session timeout function [10]. When the user is not in the user library of the online teaching system, the user’s login request needs to be rejected. In addition, when users do not operate for more than 30 min, they need to automatically jump out of the system to prevent other users from directly using the computer to enter the online teaching system to operate relevant functions, which will help to improve the security of the online teaching system. This encapsulates the data access base class. It should be noted that the system uses the NHibernate component for database access and persistence.Tomcat7 0 has built a basic system to form a unified access portal and a unified information display system. The software of this system is the operating system software and database system software installed on each server. The specific requirements are shown in Table 2. Table 2. System softwar requirements Software name

Accessory description

Quantity

Database

SQL SERVER2000

1

Middleware

Tomcat7.9

2

Report software

MicroStrategy

2

For the accelerated gradient descent method, such as momentum gradient method, the introduced variable y can accelerate the convergence speed of the algorithm, but λ When it is large, when it reaches near the extreme point, due to the large momentum along the original direction, it will cause serious overshoot and tremor and reduce the

236

Z. Xu et al.

convergence speed in the later stage of the algorithm. Resetting the controller in system theory can effectively weaken the overshoot and tremor in system control. An important function of reset control is to reduce the overshoot of the system and reset the integrator to zero when the system reaches the set value. In the gradient descent method, t is equivalent to the set value, so the reset momentum gradient method can be set as follows:   x˙ = x∇i f (θ ) − t (1) y˙ = −y − ρx∇f (θ ) The idea of reset control is used to weaken the overshoot and tremor phenomenon in the accelerated gradient method, accelerate the convergence speed and improve the stability of the algorithm. The gradient descent method is the simplest and effective algorithm to solve the convex optimization problem. It has important applications in engineering practice, such as parameter identification, machine learning and optimization control. Some important concepts in convex optimization problems and the basic framework and convergence characteristics of the general gradient descent method are given: consider the following unconstrained convex optimization problem minx f (θ ), which is a differentiable convex function with a unique global minimum point x and the corresponding minimum value is y. For this kind of problems, the gradient descent method can effectively search the minimum point ρ, As the name suggests, it is to iterate along the negative gradient direction ϕi to find the minimum value point of the function. The continuous gradient descent method can be expressed as:  z = −ρ ϕi − min f (θ )(˙x + y˙ ) (2) x

There are many different methods to prove the convergence of gradient descent method. Here, its convergence characteristics will be analyzed from the perspective of system. It is not difficult to see that the gradient descent method is a class of nonlinear feedback system, and its convergence is similar to the stability of nonlinear system. Therefore, the convergence analysis can be carried out by Lyapunov method. For the continuous gradient descent method, the Lyapunov function is taken as: V (x) =

1 x − y2 2z

(3)

The managers and relevant teachers in the education and teaching system of the theory of Variable Order Fractional gradient descent method should timely release the teaching resources related to the theory of Variable Order Fractional gradient descent method and the reference and registration materials of some large-scale competitions on the system. Therefore, it is necessary to better process and update the data of the teaching system. Timely unify the teaching of a series of courses, such as the theory of Variable Order Fractional step-down method, the theory of Variable Order Fractional step-down method, the related data processing program design of Variable Order Fractional step-down method theory, and the data processing program design of Android. In the theoretical education and teaching system of Variable Order Fractional gradient descent method, the administrator divides the course into multiple chapters for segmented teaching, and then fully protects the educational intellectual property rights while sharing

Design of Online Teaching System for Theory

237

resources. In the system, students can choose appropriate theoretical courses of Variable Order Fractional gradient descent method according to their own learning needs, and select appropriate resources for systematic learning. The resource processing module of Variable Order Fractional gradient descent theory course is shown in Fig. 5.

Information audit To examine Authority management Teacher

Student

Answering question

Online Q & A

Teacher student interaction

Online discussion

Consult

Put questions to

Interaction

Share

Fig. 5. Theory course resource processing module

The teaching resource processing module on the theoretical education and teaching system of Variable Order Fractional gradient descent method is shown in the figure. The system adopts dynamic course management mode, which is convenient to expand and adjust the course content at any time. In order to better integrate and classify teaching resources and facilitate the updating and storage of teaching resources, an online teaching system is designed as follows. The system builds an online learning and communication platform for students, which is divided into two subsystems: background management and foreground teaching. The main functional modules of the background management system are: administrator user management, authority management, role management, discipline management, class management, learning card management, chapter management and paragraph management; The main functional modules of the front desk teaching system are: user registration and login, viewing learning card, learning course, viewing questions, asking questions, answering questions and system message notification.

3 Analysis of Experimental Results System testing is the work to ensure the system quality. It is the test of the whole product after the function development of the software product is completed. The purpose of system test is to verify whether the system meets the needs of users, so as to ensure the overall quality of software. System test is a black box test. The problems found in the test process should be debugged to find out the cause and location of errors, and then corrected. After correction, verification test should be carried out. Functional

238

Z. Xu et al.

test is mainly demand verification, The purpose is to ensure whether the final system implementation is consistent with the initial requirements analysis, whether it can meet the needs of users, whether the requirements are correctly realized, and whether the performance test is completed through stress test. The purpose is to verify whether the system performance can meet the normal operation of the system when multiple users access the system at the same time. The test environment is shown in Table 3. Table 3. List of test environment Serial number

Name

Hardware environment

Software environment

A

The server

Lenovo x3850

Windows XP

B

Client

Intel Core

Windows7

C

Network environment

Wan 150 Mbps

–

Functional testing needs to cover all demand points of demand analysis. The use case design of functional testing is a very complicated work. Each demand requires a large number of test cases to verify. The system has designed a total of test cases. Due to the space limitation of the article, some use cases are used to illustrate the functional testing process, and the user login test cases are used to verify the correctness of the user login function, See Table 4 for details. Table 4. User login test cases Test case name

User login function test

Test Manual

Enter the website address in the address bar; Enter the user name in the login interface; Input password; Click the “login” button to observe the interface changes

Expected results Open the teacher user interface Actual results

Consistent with expected results

According to Table 4, after the training plan is formulated, the teacher user can continue to carry out course management and training project management, and set corresponding task course management test cases for courses and projects to verify the correctness of the teacher user’s course management function, as shown in Table 5.

Design of Online Teaching System for Theory

239

Table 5. Test cases of course management Test case name

Course management function test

Test Manual

Enter the web address in the address bar; User login; Click the “management center” tab; Select course management

Expected results Open the task list and display “duration” and other information Actual results

Consistent with expected results

According to Table 5, In order to better test the performance of the system under high load, the system uses the loader runner tool to create multiple groups of different numbers of users to operate at the same time and record the reaction time of the system. The test items include: login system, online communication and online teaching. The test results are shown in Table 6. Table 6. Login system test Login system test case Prerequisite

Normal login interface

Test target

Understand the performance of the system when multiple users log in at the same time

Method

Use tools to simulate multi-user online communication scenarios

Number of concurrent tests Average time (s)

Maximum time (s)

Average use of network packets

40

1.325

3.652

71

70

4.652

6.989

76

210

5.658

9.658

100

According to Table 6 and Table 7, performance is an important part of the system. Corresponding performance indicators have been set in the demand stage, which are the focus of the performance test stage. If the test results do not meet the set goals, the online teaching system cannot be applied to the school to deal with relevant business. This can not improve the efficiency of teaching management, but will reduce the efficiency. Therefore, the performance test must be carried out before deployment. The online teaching system can be deployed to the school only when the performance test results are consistent with the requirements and objectives. During the test, LoadRunner software is mainly used for simulation test. According to the performance requirements, it needs to meet the concurrent access of 500 users. The test results are shown in Table 8.

240

Z. Xu et al. Table 7. Online AC test

Online communication test case Prerequisite

Normal login system

Test target

Understand the performance of the system in multi-user online communication at the same time

Method

Use tools to simulate multi-user online communication scenarios

Number of concurrent tests Average time (s) 40

1.325

Maximum time (s) 2.652

Average use of network packets 74

70

3.655

6.325

75

210

5.652

12.325

81

Table 8. System performance test results Number of concurrent users

Response speed (seconds)

CPU utilization (%)

60

1.45

1

120

1.46

2

180

1.55

3

240

1.55

3

300

1.63

7

360

1.73

8

480

1.98

10

540

2.12

18

600

2.36

21

For the software performance test under specific operating environment and load conditions, the system mainly focuses on the impact of the number of people online on the response time, as shown in Table 9. According to Table 8 and Table 9, the performance test of the system is completed by checking the monitoring points in the performance test cases. It is very important to confirm whether the basic performance requirements of the system are met. In this online teaching system, the response time and throughput of the system are mainly tested. To sum up, the following conclusions can be drawn from the functional test results and performance test results. The online teaching system meets the application requirements of the school, has comprehensive functions, and the performance is consistent with the goals set by the school. The system has no major defects, and there are no functional defects that affect the online teaching business. The problems found in the initial stage have been solved.

Design of Online Teaching System for Theory

241

Table 9. Performance test cases Test item

Number of people online at the same time

Expected average response time

Actual average response time

A

100

Less than 18

0.526

B

200

Less than 18

0.765

C

300

Less than 28

1.065

D

400

Less than 28

1.365

E

1000

Less than 38

1.998

4 Conclusion The establishment of teaching system on mobile network mainly includes network teaching management system, which is mainly server terminal. Mobile terminal application subsystem, which is mainly the client. The specific application of these systems includes reusable development framework and data mining technology, which realizes the overall planning of mobile terminal, optimization design and Moodle to a certain extent. In this case, a relatively rich system module can be introduced into the system design. At the same time, it can also refer to the more robust background system to make the operation of different mobile terminals more convenient. The innovation of the method is to optimize the hardware structure of the system, improve the operation performance of the hardware and optimize the software function of the system. This paper analyzes and discusses the current research situation in the field of online education, and completes the design and implementation of the online teaching system based on the theory of step-down method of variable grade. However, as for the design of online teaching system itself, there are still many places to be improved and developed. The next steps are as follows: (1) Continue to study online teaching system, including online learning model, online teaching model, online education management model, virtual laboratory research in online education, the development trend of network education, campus network education application model research. (2) In-depth discussion of educational technology research field and its development trend, educational software development standards and standard coding technology. (3) Construct a cognitive student model to solve students’ questions about intelligent teaching software and play the role of intelligent agent teaching application.

Fund Project. Application research of fractional-order gradient descent method in neural network control (L2020010).

References 1. Dolighan, T., Owen, M.: Teacher efficacy for online teaching during the COVID-19 pandemic. Brock Educ. J. 30(1), 95–116 (2021)

242

Z. Xu et al.

2. Siegel, V., Moore, G., Siegel, L.: Improving nursing students’ knowledge and assessment skills regarding skin cancer using online teaching resources. J. Dermatol. Nurses’ Assoc. 13(6), 305–308 (2021) 3. Yang, Y.: On the normalization path of online teaching of ideological and political course in colleges and universities in the “Post Epidemic” era. J. Contemp. Educ. Res. 5(1), 15–19 (2021) 4. Lin, P.H., Chen, S.Y.: Design and evaluation of a deep learning recommendation based augmented reality system for teaching programming and computational thinking. IEEE Access 2(3), 45689–45699 (2020) 5. Wójcik, K., Piekarczyk, M.: Machine learning methodology in a system applying the adaptive strategy for teaching human motions. Sensors 20(1), 314 (2020). https://doi.org/10.3390/s20 010314 6. Sns, A., Ya, B., Ks, B., et al.: Development of A-txt system compatible introductory teaching materials for Electric Power Engineering using gaming simulation - ScienceDirect. Procedia Comput. Sci. 176(3), 1557–1566 (2020) 7. Wang, Y.: Comprehensive evaluation system of teaching quality based on big data architecture. Int. J. Contin. Eng. Educ. Life-Long Learn. 30(1), 1–10 (2020) 8. Kariapper, R., Samsudeen, S.N., Fathima, S.: Quantifying the impact of online educational system in teaching and learning environment among the teachers and students. Solid State Technol. 63(6), 12118–12132 (2020) 9. Jiang, W., Sun, L., Chen, Y., et al.: A Hardware-in-the-loop-on-chip development system for teaching and development of dynamic systems. Electronics 10(7), 801–810 (2021) 10. Lingxin, K., Yajun, M.: Big data adaptive migration and fusion simulation based on fuzzy matrix. Comput. Simul. 37(3), 389–392 (2020)

Research on Collaborative Learning Behavior Recognition of Online Education Platform Based on Data Mining Zhi Zhang(B) Technique Center of Modern Education, Jiangxi Science and Technology Normal University, Nanchang 330013, China [email protected]

Abstract. The time sequence information of traditional online education platform collaborative learning behavior recognition method is not comprehensive, resulting in low recognition rate. Therefore, a collaborative learning behavior recognition method of online education platform based on data mining is designed. Monitor the learning behavior through the progress management function, identify the educational characteristics of learners, obtain the timing information of the online education platform, and feed back the information to teachers and students in a visual way, optimize the interactive links of collaborative learning, meet the personalized requirements of learners, investigate the basic database of student scores by using rough set theory, and mine data from the database. In order to identify the collaborative learning behavior of online education platform. Experimental results: the average recognition rates of the online education platform collaborative learning behavior recognition method and the other two recognition methods are 68.142%, 56.310% and 57.988%, which proves that the online education platform collaborative learning behavior recognition method integrated with data mining technology is more reliable. Keywords: Online education platform · Collaborative learning · Learning motivation · Behavior identification · Learning activities · Interactive links

1 Introduction With the rapid development of my country’s economy and society and the continuous advancement of science and technology, mobile phones, tablet computers and various interactive information tools are gradually changing our production and life styles, and people continue to pursue more information and convenient lifestyles. In the traditional sense, teaching evaluation is an activity that analyzes and evaluates teaching results and teaching process and provides a scientific basis for teaching decision-making. It is a process of evaluating and judging the practical significance of teaching activities. The cumbersome manual teaching, limited teaching time and mechanical teaching methods that exist in traditional teaching are no longer suitable for current needs, and the demand © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 243–254, 2022. https://doi.org/10.1007/978-3-031-21161-4_19

244

Z. Zhang

for information and intelligence in the education field is particularly prominent [1, 2]. Classroom teaching is the main part of school education. Paying attention to students’ learning behaviors, communication speech, and learning emotions in classroom situations are issues that cannot be ignored in the process of educational development. With the continuous development of computer vision technology, many companies and teams are combining the “Internet+” way of thinking with big data, cloud computing and other information technologies to intelligently design the teaching methods of teachers and the learning methods of students, Improve the quality of the teaching environment and enhance the effect of classroom teaching. On the one hand, classroom learning behavior research can categorize students’ learning behaviors in detail, and through in-depth analysis of various influencing factors of student behaviors, the teaching evaluation can be implemented on objective data. To acquire and analyze student behavior, firstly, it is necessary to use image processing technology to process the video content shot by the classroom surveillance camera, and to detect and recognize the students in the video screen. On the other hand, by collecting the behavior data of students in the classroom, analyzing the learning behavior data of students, forming an evaluation of the students’ learning state, forming effective teaching effects are reflected in the data, which is conducive to improving the learning efficiency. The purpose of automatic teaching is realized through face recognition, so as to reduce the time teachers spend on classroom management, increase the teaching and learning time of teachers and students, and help the school to manage students. How to use the current computer vision technology and the related measurement methods of the teaching process to combine, how to measure the learning behavior of students in the classroom, etc., these are the difficulties faced by the current teaching detection work. The research on these issues evaluates the current teaching process The development of it is of great significance. Zhang et al. took CNKI and WOS as data sources and obtained 37 representative literatures highly related to research topics in recent 5 years at home and abroad through key word retrieval, retrospective retrieval and data cleaning. Integration of the information ecology theory and empirical analysis process form the research framework, from information, information, information technology and information environment such as dimension to explore the influence factors of online education platform collaborative learning behavior from aware of interest, interactive driving factors, content, learning orientation known variables identify online education platform collaborative learning behavior. Although the efficiency of this method is very high, the problem of incomplete timing information is not deeply studied, which leads to the decline of recognition rate. Wu et al. coded micro “click stream” and mesoscopic “activity stream” data based on 3D design software, defined verbs and objects suitable for 3D design platform, and obtained the wholeprocess behavior data of learners in 3D design platform. On this basis, according to the STEAM 3D education learner behavior analysis model, the data of the learning process of the 3D design course of small sample teaching practice are analyzed to complete the cooperative learning behavior. This method has the characteristics of better recognition efficiency, but it lacks the detailed description of time sequence information in the research process. The recognition rate of cooperative learning behavior decreased. Student behavior recognition is to analyze the head posture and facial expressions of students in the classroom based on the face recognition of the students, to judge

Research on Collaborative Learning Behavior Recognition

245

whether the students in the classroom are listening carefully, and to record the student’s head-up state, so as to minimize human intervention. Classroom is the core of school education, the evaluation of classroom teaching process is of great significance to the improvement of teaching quality, and the performance of students’ learning behavior is an important part of classroom teaching evaluation. In this way, it can help teachers adjust and change their teaching styles in time to achieve more efficient teaching. At the same time, it can also mobilize students’ learning enthusiasm and improve learning efficiency. By evaluating the learning behavior of students, forming effective feedback information and teaching orientation can effectively promote classroom teaching and student development. At present, the research materials on the combination of data mining and online education platform collaborative learning behavior recognition are not very rich, and they need to be continuously improved. The time sequence information of the traditional collaborative learning behavior recognition method of online education platform is not comprehensive, resulting in a low recognition rate. This paper aims to solve various problems existing in the traditional method, monitor the learning behavior through the progress management function, identify the educational characteristics of learners, and obtain the time sequence information of online education platform. The collaborative learning behavior of online education platform is identified based on the results of student score data mining.

2 Research on Collaborative Learning Behavior Recognition of Online Education Platform Based on Data Mining 2.1 Identify the Educational Characteristics of Learners In the online education platform, learners perceive and recognize the educational availability of the platform and realize what specific learning behaviors they can complete by using these technical tools, such as planning their own learning goals through the learning task management function. General characteristics mainly refer to the psychological, physiological and social characteristics of learners, which are inherent in their peers and easy to be understood by teachers. Obtain the courses you want to learn through the platform search engine, and monitor the learning behavior through the progress management function. The main purpose of understanding learners’ initial ability is to understand what knowledge and skills learners have before learning new subject content, students’ ability to master information technology and their learning attitude towards the learning content. Through the online test module, you can evaluate your learning achievements, summarize and reflect through the online log, and share learning resources with other learners and teachers. The characteristics of learners are analyzed mainly from four aspects: general characteristics, starting ability analysis, learning style and learning motivation, as shown in Fig. 1.

246

Z. Zhang

Basic features

Initial capability characteristics Learner characteristics Learning style

learning motivation

Fig. 1. Schematic diagram of learner characteristics

As can be seen from Fig. 1, learning style mainly refers to the learning style and learning tendency with personality characteristics shown by learners, such as field dependent and field independent style, concrete and abstract style, etc. When learners perceive that they are part of the group and that they are in a community, they will be more willing to actively participate in group and group activities. Mastering learners’ learning is helpful to teachers’ teaching. Learning motivation refers to the driving force of learners’ learning activities, that is, learning motivation. Motivation plays an extremely important role in people’s behavior and activities. Mastering learners’ learning motivation can make use of students’ learning. The existence of social availability makes learners realize that technology gives them the ability to communicate with other learners and teachers, which makes learners have a sense of community and belonging, so as to promote their communication with others and jointly complete curriculum learning and thinking. Learning task design is the focus and core of the whole collaborative learning, which can make learners clear their goals and solve problems. It is important that learners can actively participate in the process of collaborative learning, so as to cultivate learners’ ability to apply theoretical knowledge with practical application and solve problems [3]. Learners and other learners form a learning group to help each other, complete the task together, and share learning experience and experience with each other. The design of tasks should be suitable for the characteristics of learners, preferably within the scope of learners’ proximity to the development area, otherwise it is easy to attack learners’ learning enthusiasm. Learners communicate with teachers. For difficult problems, teachers answer questions for learners. Learners interact with the content, summarize and reflect on the learned content, so as to realize the re creation of knowledge. The completion of the task depends on the network and adopts a cooperative way. At the same time, it also has the opportunity of autonomous learning. The design of learning tasks should be close to learners’ life and have a certain real situation, so as to stimulate learners’ learning motivation. The results of learning tasks should be easy to generate and submit.

Research on Collaborative Learning Behavior Recognition

247

2.2 Obtain Timing Information of Online Education Platform The design of learning activities and resources is designed and formulated by teachers (or teaching designers) based on the comprehensive analysis of learning content, learning environment and learners’ characteristics. It is not only the most important part of collaborative learning, but also the guarantee for the smooth development of collaborative learning. The reasoning of time dimension information is an important means for artificial intelligence to complete the task of behavior recognition. It is a method to connect the target or object logically and sequentially according to the passage of time. In the collaborative learning environment, the original intention of learning activities and resources design is to help students maintain learning progress and enable students to actively participate in the learning process, so as to create a collaborative learning environment and atmosphere. Time series reasoning can enable intelligent species to infer what may happen in the future or strengthen the relationship between things before and after time according to the past and current situation. We solve this problem by using different minimum support and minimum confidence for different classes. We only need the user to specify a total minimum support, and then distribute it to each class according to the class distribution data as follows: l=

α(h)2 −1 |h|

(1)

In formula (1), α represents the number of instances in the training data, and h represents the minimum support. Therefore, the design of learning activities and resources must first have a clear purpose. The purpose of this part of the design is to make students’ learning closely related to their learning goals, improve students’ participation in curriculum learning, and closely link the entire process with learning evaluation to obtain timely feedback. In remote visual surveillance, people can easily distinguish the sequence of two sequence static images at a given time and infer the actions between the contents displayed by the two static images. The design of learning activities and resources in this research is divided into online learning part and classroom learning part in terms of presentation form. Time reasoning is extremely important in behavior recognition, and it provides a lot of help for inference of behavior categories. The online learning part needs to complete the learning of the core content of the course, so it is necessary to strictly plan the learning progress and the distribution of learning resources. Correctly distinguishing the time series is very helpful for behavior recognition. The behavior recognition in the video is the main point that we need to study and explore. Offline classroom learning mainly completes the work of answering questions, discussing, and evaluating. Each part is an important part of learning activities, and there is no distinction between importance and importance. The completion of each part requires the active cooperation of teachers and students. The learning behaviors of students in online courses will be recorded in detail in the learning log database, and visually fed back to teachers and students. It can also compare the learning progress of each student with the current learning progress and supervise students who are lower than the current learning progress, so as to effectively monitor the student’s learning.

248

Z. Zhang

2.3 Optimize the Interactive Links of Collaborative Learning Learning objectives are the goal orientation of collaborative learning. Designing appropriate learning objectives can ensure that learners are not “lively” on the surface, but substantive cooperation in learning. The behavioral level of collaborative learning culture mainly refers to students’ learning behavior, which is embodied in the operation of curriculum platform, the arrangement of e-learning time, the use of curriculum forum, online testing and assessment, the participation of offline learning activities, the discussion of classroom learning activities, learning partners, seeking help, etc. The main problems are the low utilization rate of curriculum forum, insufficient participation in classroom learning and discussion activities, lack of learning partners and inability to seek help in time. Therefore, in practice, the following three objectives can be achieved by integrating bloom and Gagne’s classification of learning objectives and combining the specific subject content. Knowledge and skill objectives: what knowledge points or theoretical frameworks learners can apply, what new knowledge they can discover or what new skills they can develop. In addition, there are differences between boys and girls in these aspects, that is, the main deficiency is the lack of human-computer interaction and face-to-face interaction. Method strategy goal: learners can develop strategies for problem solving, learning, cooperation and communication. Attitude experience goal: what kind of attitude and emotional experience learners get through collaborative learning. Some scholars have shown that there is a positive correlation between learners’ learning performance and the degree of social interaction in the context of online open courses. The main process of collaborative learning interaction is shown in Fig. 2:

Extracting learner features

Design collaboration tools

Clarify learning tasks

Design learning activities

Set up a study group

Design Interaction

Creating a learning environment

Design evaluation link

Fig. 2. The main flow chart of the interactive part of collaborative learning

It can be seen from Fig. 2 that learners perceive and recognize the social availability of the platform and realize what social activities they can accomplish with these

Research on Collaborative Learning Behavior Recognition

249

technical tools, such as communicating with other learners and teachers through synchronous and asynchronous social tools comminicate. Learning resources refer to all relevant elements that learners can use in the learning process, including people, money, materials, and information that can support learning. The socialized learning behavior of high-performing learners goes beyond simple knowledge sharing and interpersonal communication, and there is a deeper level of knowledge interaction and interpersonal communication. Therefore, the key to the construction of collaborative learning cultural behavior level is to strengthen the degree of human-computer interaction and face-toface interaction of student learning under collaborative learning conditions. The learning resources referred to in this article mainly refer to teachers, peers, teaching materials, network resources, etc. In collaborative learning, the design of these learning resources is not only restricted by the characteristics of learners, learning tasks, and platform conditions, but also learning resources are designed for collaborative learning activities, not for individual learners. On the one hand, it must be restricted from the system level; on the other hand, the human factor must be brought into play. Under the guidance of the concept of “teacher-led-student main body”, the strength of both teachers and students should be brought into play. Therefore, the resource design of learning resources should follow the following principles: it is closely related to collaborative learning activities, and learning resources provide support for collaborative learning. Therefore, the higher the correlation, the greater the help for collaborative learning. Secondly, the collaborative learning environment is complex and changeable, and requires the cooperation of a variety of learning behaviors to complete. Therefore, students should adapt to the open learning environment as soon as possible while also helping students develop good learning habits. The design of resources should be diversified, preferably with pictures, texts and sounds, which can support the development of collaborative learning activities from different angles, meet the individual requirements of learners, and facilitate the retrieval and search of resources. Emphasizes that learner-centered, whether the design and selection of resources are reasonable, the key depends on whether these resources can help learners discover, explore, and solve problems in collaborative learning, actively participate in collaborative activities, and can construct the meaning of knowledge. 2.4 Behavior Recognition Based on Data Mining Data mining is a process of extracting hidden, unknown but potentially useful information and knowledge from a large number of incomplete, noisy, fuzzy and random data, which is called data mining [4–6]. The performance of students’ learning behavior includes two aspects: one is the status of learning behavior, mainly including the overall atmosphere of competition, debate, cooperation, problem solving, partners, design and role play, discussion, the overall status of listening input and the overall atmosphere of the classroom; The other is the status of non learning behavior, including games, reading novels, arguing, etc. In recent years, the use of statistical technology to calculate students’ scores and data mining technology have been gradually changed to data mining technology [7]. Rough set theory is used to investigate the basic database of students’ scores, extract classification rules, and understand the factors affecting students’ scores [8]. The difficult problem is to solve the problem of weight, including negative weight and positive weight. After a large number of experiments, it is found that it is ideal to

250

Z. Zhang

combine the two. The positive weight formula is as follows: pη =

g ×T η

(2)

In formula (2), g represents the set of positive items covering the data instance, η represents the set of negative items covering the data instance, and T represents the confidence of the original positive rule. In traditional classroom teaching, teachers can form an effective evaluation of students to a certain extent by observing students’ learning behaviors and homework completion. However, it is difficult to form a comprehensive and systematic evaluation system of students’ learning behaviors by teachers alone. Classification rules usually include usual grades, classroom efficiency, and after-school study time, etc. These main factors affect the overall score. When there are multiple data with the same score, use the following priority formula to calculate:  (λ − G)2 (3) y= |Q| + |G| In formula (3), λ represents the discretization coefficient, G represents the overall score function, and Q represents the allocation index. In the collaborative learning environment under the teaching platform, learners’ learning and application of relevant knowledge and the cultivation of learners’ comprehensive ability are the core of collaborative learning, so more attention should be paid to problem solving and design learning task design. If a module that can automatically record students’ learning behavior can be built in the platform, the problem of incomplete and objective recording of student behavior can be solved. If it is the analysis mode of the classroom question and answer process, it is necessary to extract the event data of raising hands, sitting down, and standing up, and classify each piece of data according to the location field and the time field. By recording students’ learning behavior information, more meaningful results can be further dig out, which is of great significance to help teaching staff improve the teaching process. Among them, according to the classification of the time field, it is necessary to extract the time of each question and the time from standing to sitting down which is the closest to the time of the question. This time period is used as the period of each question [9, 10]. Form a study group with other learners in the online community, and help each other to complete a certain task together. The task of design should be open, and the standard of exploration results cannot be “correct” or “incorrect”. The problem-solving process is not just to expect learners to master basic theoretical knowledge. If it is the classroom activity analysis mode, it is necessary to extract the event data of bowing and lying on the table, and classify each piece of data according to the location field. Then the classified results are sent to the subroutine of each person’s activity analysis and the overall situation analysis of the classroom. Learners complete the in-depth understanding of knowledge, the mastery of learning methods and the improvement of thinking through interactive teaching.

Research on Collaborative Learning Behavior Recognition

251

3 Experimental Research 3.1 Set up the Experimental Environment This experiment is based on PC, Kinect2.0 somatosensory device, Hikvision PTZ camera, student behavior measurement in a small classroom scene. The database used in this article is My Sql, which is realized by the relevant functions integrated with Python. The database is mainly used to save the student’s identity information and head posture information for storage, so as to be able to view the heads of different students in the entire visual process. Frequency information of bowing. Since each Kinect can only identify 6 human skeleton information, if multiple Kinects can be added, it is also suitable for large classroom scenes, but the cost is relatively high. The original data collected in the experiment is processed and identified by the PC and then transmitted and stored in the server database. The parameters selected in this experiment are consistent, ensuring the authenticity and reliability of the experimental results. 3.2 Analysis of Experimental Results The experiment selects the PCA-based online education platform collaborative learning behavior recognition method and the SVM-based online education platform collaborative learning behavior recognition method, and compares the experiments with the online education platform collaborative learning behavior recognition method in the article. The recognition rates of the three online education platform collaborative learning behavior recognition methods under different minimum support conditions were tested respectively. The experimental results are shown in Table 1, 2, 3 and 4: Table 1. Minimum support 20 recognition rate (%) Number of experiments

PCA-based online education platform collaborative learning behavior recognition method

SVM-based online education platform collaborative learning behavior recognition method

Online education platform collaborative learning behavior recognition method in the article

10

85.647

86.487

96.118

20

88.261

83.614

98.515

30

84.632

82.520

97.662

40

85.799

86.144

98.667

50

84.155

85.219

97.335

60

86.332

83.117

94.582

70

85.166

82.505

93.707

80

85.616

82.119

96.825 (continued)

252

Z. Zhang Table 1. (continued)

Number of experiments

PCA-based online education platform collaborative learning behavior recognition method

SVM-based online education platform collaborative learning behavior recognition method

Online education platform collaborative learning behavior recognition method in the article

90

84.213

83.006

98.366

100

83.641

82.519

95.884

Table 2. Minimum support 30 recognition rate (%) Number of experiments

PCA-based online education platform collaborative learning behavior recognition method

SVM-based online education platform collaborative learning behavior recognition method

Online education platform collaborative learning behavior recognition method in the article

10

65.948

64.588

73.644

20

66.717

66.327

72.505

30

65.311

69.522

73.164

40

69.518

67.346

75.648

50

64.315

68.512

71.098

60

63.228

68.912

73.005

70

64.187

69.577

73.469

80

66.332

66.344

74.156

90

67.558

68.203

75.606

100

62.313

69.457

73.987

According to Table 1, the average recognition rate of the online education platform collaborative learning behavior recognition method and the other two recognition methods are: 96.766%, 85.346%, 83.725%; According to Table 2, the online education platform collaborative learning behavior in the article Recognition method, and the average recognition rate of the other two recognition methods: 73.628%, 65.543%, 67.879%; according to Table 3, we can see that the online education platform collaborative learning behavior recognition method in the article, and the other two recognition methods are the average recognition rate They are: 57.220%, 45.434%, 45.401%; according to Table 4, we can see that the online education platform collaborative learning behavior recognition method in the article has an average recognition rate of 44.953%, 28.092%, and 34.946% with the other two recognition methods. The performance of collaborative learning behavior recognition method of online education platform is better.

Research on Collaborative Learning Behavior Recognition

253

Table 3. Minimum support 40 recognition rate (%) Number of experiments

PCA-based online education platform collaborative learning behavior recognition method

SVM-based online education platform collaborative learning behavior recognition method

Online education platform collaborative learning behavior recognition method in the article

10

45.321

44.618

56.314

20

44.602

43.205

56.847

30

43.873

44.887

57.265

40

46.188

43.625

58.316

50

45.121

45.986

55.124

60

46.908

44.221

56.928

70

45.877

45.606

57.335

80

44.330

46.187

58.209

90

46.312

47.552

59.471

100

45.807

48.121

56.387

Table 4. Minimum support 50 recognition rate (%) Number of experiments

PCA-based online education platform collaborative learning behavior recognition method

SVM-based online education platform collaborative learning behavior recognition method

Online education platform collaborative learning behavior recognition method in the article

10

26.847

32.647

42.659

20

29.145

37.558

41.577

30

28.335

36.124

40.336

40

29.647

35.874

46.578

50

31.215

36.215

44.315

60

30.008

35.184

45.697

70

24.367

34.515

47.502

80

26.588

33.609

48.223

90

27.854

34.518

45.718

100

26.914

33.215

46.922

254

Z. Zhang

4 Conclusion The time sequence information of the traditional online education platform collaborative learning behavior recognition method is not comprehensive, resulting in a low recognition rate. This paper takes solving a variety of problems existing in the traditional methods as the research goal, monitors the learning behavior through the progress management function, identifies the educational characteristics of learners, obtains the time sequence information of the online education platform, and identifies the online education platform collaborative learning behavior in combination with the student score data mining results. In the future, under the guidance of the theories of collaborative learning and data mining, it is necessary to analyze the possibility and advantages of the combination of online education platform and collaborative learning, design collaborative learning teaching design, which consists of learner feature analysis, learning task design and other links, so as to design a targeted learning plan, ensure the diversification of students’ learning modes, and promote the further development of modern education modes.

References 1. Zhang, M., Zhu, A., Zhang, F.: Research on influencing factors of online education platform users’ continuous use behavior. Library Tribune 40(5), 82–91 (2020) 2. Yonghe, W., Yahui, T., Shouchao, G., et al.: Research on the model of learner behavior analysis based on online 3D education platform —taking GeekCAD online platform as an example. China Educ. Technol. 12, 61–67 (2019) 3. Ming-yao, L.I.: Personalized appeal and resources optimization of online education in the era of big data. Theory Pract. Educ. 40(4), 30–34 (2020) 4. Luo, Q, Long, J., Chen, H., et al.: Time series prediction of subway station energy consumption based on data mining algorithm. Urban Mass Transit. 23(6), 23–27 (2020) 5. Chen, D., Zhan, Y., Yang, B.: Analysis of applications of deep learning in educational big data mining. E-educ. Res. 40(2), 68–76 (2019) 6. Zhu, X., Shen, Y.: RPMA low-power wide-area network planning method based on data mining. J. Commun. 40(3), 28–35 (2019) 7. Liu, Y., Zhang, Y., Wang, C.-X., et al.: Behavior recognition algorithm based on RGB-D and deep learning. Comput. Eng. Des. 40(6), 1747–1750, 1762 (2019) 8. Mu, R.: Simulation on accurate Identification of Abnormal Access Behavior information of Network domain name users. Comput. Simul. 35(07), 339–342, 376 (2018) 9. Wang, C.X., Jiang, C.H.: Group behavior recognition method based on actionvlad pooling and hierarchical deep learning network. Data Acquis. Process. 34(4), 585–593 (2019) 10. Chen, C.H., Liu, Y.: Two person interaction behavior recognition based on improved sum product network. Comput. Technol. Develop. 29(10), 157–163 (2019)

Design of Online Education System for Ideological and Political Courses of Traditional Chinese Medicine Based on MOOC Model Na Zhao1(B) and Zhongwen Zheng2 1 Changchun Humanities and Sciences College, Changchun 130000, China

[email protected] 2 Fuyang Normal University, Fuyang 236037, China

Abstract. At present, the teaching mode of ideological and political courses in colleges and universities is single, resulting in poor teaching effect. In order to improve the teaching effect of ideological and political courses for Chinese medicine majors, MOOCs are used to reform and explore the teaching mode, and designed a MOOC-based curriculum thinking of traditional Chinese medicine. The government affairs online education system is based on ASP.NET development technology, with C# (pronounced as C Sharp) as the main development language, and is realized by SQL Svrer database technology. The system is developed on the basis of the C/S architecture of the system hardware, and modules such as user management, teaching management, educational resource recommendation, and question bank test papers are designed. The system software has designed the system deployment and migration, personnel training and other processes. The results show that: in the case of high concurrency, the system meets the basic needs of users, and the server CPU usage and memory usage reach the expected goals. The system aims to improve the teaching quality of ideological and political courses, enhance the teaching effectiveness of ideological and political courses, and at the same time expand the teaching orientation of ideological and political courses, and break through the difficulties of ideological and political teaching. Keywords: MOOC model · Ideological and political · Functional modules · Online education system

1 Introduction With the continuous update and development of network technology and information technology, more and more online education attracts the attention of educated and educators. Online education can also be understood as distance education or online learning, which generally refers to a kind of network-based platform. Teaching mode [1]. In the information age with the rapid development of science and technology network, as a

© ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 255–268, 2022. https://doi.org/10.1007/978-3-031-21161-4_20

256

N. Zhao and Z. Zheng

supplement to traditional education, online education occupies a higher and higher proportion in people’s lives. MOOC (English abbreviation for Large Open Online Course) has become an important part of teaching and learning in today’s world. One of the hotspots. The original intention of the MOOC teaching model is to realize an online classroom aimed at the general public. People learn online through the online classroom, replacing some of the traditional face-to-face teaching links in the past. MOOC is also the latest development of distance education. People do not have to conduct learning activities in centralized classrooms at the same time and place as before. It is realized and developed through the form of open educational resources, which has large-scale and open It has become a research hotspot at home and abroad. The ideological and political theory course in colleges and universities is a course set up to ensure the political direction of college students’ growth and to ensure that the mainstream consciousness of socialism can be effectively implemented in higher education. Ideological and political courses in colleges and universities have distinct characteristics that are different from other courses: First, other courses contain morality and cultivate people through professional teaching, while ideological and political courses in colleges and universities are dedicated to educating young college students on ideology and morality, political awareness, and social behavior. The second is that the nature of the course determines the teaching content, teaching process and teaching effect of ideological and political courses in colleges and universities, and the economic activities, political activities, and moral construction in the development of Chinese society. They are closely connected and are greatly influenced by various social events and information. At present, relevant scholars have made research on the design of the ideological and political online education system for Chinese medicine professional courses. Reference [2] proposed the exploration of the ideological and political teaching reform of the “Biochemistry” course for Chinese medicine majors. Taking “enzymology” as an example, it actively explored the specific measures of the ideological and political implementation of the course in biochemistry, and deeply explored the educating people contained in the course. With the elements of moral education, ideological and political education is naturally integrated into the biochemistry classroom of Chinese medicine majors through case teaching. While imparting professional knowledge, it guides students to establish a correct world outlook, outlook on life and values, and truly realizes the fundamental task of teaching and educating people. Noble duty. In view of the above, the reform and improvement of ideological and political education and teaching in colleges and universities is related to the overall development of higher education in my country. From the overall point of view of the education and teaching of ideological and political courses in colleges and universities, the classroom is only a platform for the implementation of courses, and it is only a way for teachers and students to communicate. The quantity is very limited. Therefore, in the information age, this study designs an online education system for ideological and political courses in traditional Chinese medicine based on the MOOC model. Using ASP.NET, SQL Server and other technologies to develop and realize the online education system of government affairs, through a variety of development languages, it can better realize the integrated teaching of traditional Chinese medicine and ideological and political courses.

Design of Online Education System for Ideological and Political Courses

257

And designed the system framework, functional modules, hardware, database, etc., and realized the design of the online education system of Chinese medicine ideological and political courses.

2 Design of Online Education System for Ideological and Political Courses of Chinese Medicine Majors Based on MOOC Traditional Chinese medicine culture is an important part of China’s excellent traditional culture. The core values of traditional Chinese medicine culture with a long history and profound heritage are “benevolence, harmony, refinement and sincerity”. It is an important theoretical resource for the ideological and political education of Chinese medicine majors in colleges and universities in the new era. Combining ideological and political education with traditional Chinese medicine culture and running through the education of professional courses will help the Chinese medicine major to cultivate Chinese medicine professionals with solid professional quality and noble personality, a strong sense of social responsibility and dedication. An emerging online course development model, which has the characteristics of large-scale, open, online, etc., enables learners to learn without time and region restrictions, and pays more attention to the subject status of learners, which can stimulate learners’ interest in learning. Improve the quality of learning. This brand-new educational technology has optimized the traditional classroom, caused a huge shock to traditional education, and also brought impact and challenges to ideological and political education. Therefore, for ideological and political educators, it is necessary to seize the opportunity to launch MOOC-based MOOCs. The ideological and political network teaching system, with the powerful functions of MOOC, can improve the teaching quality of ideological and political courses, enhance the effectiveness of teaching, and at the same time expand the position of ideological and political education and break through the teaching difficulties of ideological and political courses. 2.1 Key Technologies for System Development The system is mainly based on ASP.NET development technology, with C# (pronounced as C Sharp) as the main development language, and realized by SQL Svrer database technology. Taking into account the compatibility between platforms, the MOOC learning platform also adopts the same development technology. At the same time, ASP.NET ASP.NET is one of the mainstream network development technologies, especially the popular dynamic Web development technology in the field of Web development. ASP.NET Technology ASP.NET, part of the Microsoft.NET Framework, is an environment that simplifies application development in a highly distributed Internet environment. The.NET Framework includes the Common Language Runtime, which provides various core services such as

258

N. Zhao and Z. Zheng

memory management, thread management, and code safety, as well as the.NET Framework class library. The.NET Framework is a comprehensive, object-oriented collection of types that developers use to create applications. MySQL Database Technology MySQL was first developed and released by the Swedish MySQL AB Company, and later the company was acquired by Sun Company and became a product of its own. In January 2010, Sun was acquired by the database giant Oracle, and MySQL became a small open source relational database management system under Oracle, which is very suitable for the background data management of websites, so MySQL has always been based on PHP. The best partner is known [3]. The MySQL database server has the following fairly obvious advantages: Cross-platform support, support for various development languages, handle large databases with high processing speed and high security. 2.2 System Framework Design The application system is developed on the basis of the C/S architecture, and the tasks are assigned to the client and the server, so that the two are connected through the Internet. After the server request is found, the server will respond according to the capabilities previously set by the user. The B/S structure is actually an upgrade or improvement of the C/S model, while the browser is a simplified version of the client. The user interacts with it through the browser, inputs information into the browser, and the server starts to process the task and pass The browser displays the result processed by the server to the user [4]. Comparing the characteristics of the C/S structure and the B/S structure, we can understand that both have their advantages and disadvantages. The application system developed based on the C/S structure operates through the client, which reduces the operating pressure of the server, improves the response speed of the server, and obtains better fluency. The system developed based on the B/S structure can use the system only through the browser and the network, without installing the client, which is very convenient and fast; at the same time, the maintainer of such a system only needs to upgrade and maintain the functions on the server, without the need for User real-time operation, upgrade and maintenance are easier; after the system completes the request, it will cut off the connection by itself, no longer occupying network resources, thereby reducing network load [5]. By comparing the advantages and disadvantages of the two structures, we decided to use the combination of B/S + C/S structure, and load the plug-in in the browser instead of the client, so that the operation is smooth, the response speed is improved, it is easy to maintain and upgrade, and the The steps of installing the client are comprehensively considered, which is better than the independent structure of the two. The system is designed with a multi-layer architecture: The bottom layer is the network operating system; The second layer is the database system layer; The third layer is the database interface layer, which collects data information into each interface system through the application server;

Design of Online Education System for Ideological and Political Courses

259

The fourth layer is the service layer, which exchanges information with the application server through JSP, and exchanges information with the Browser through HTTP; The fifth layer is the user layer, which is oriented to the end user and exchanges information with the system through the user page; 2.3 System Hardware Design CS89712 is a 1632-bit microprocessor based on ARM7TDMI processor core produced by Cirrus Logic. It adopts high-performance 32-bit RSC structure and is a SOC chip designed for ultra-low power mobile communication and industrial control. The core logic function of the chip is built around the ARM72T processor, with an 8K-byte 4-way joint setting unified cache and write buffer, and an MMU with 64 entry TLBs. Figure 1 is a system hardware structure diagram. Real time clock

Serial interface module

storage

CS89712

ADC conversio n part

Serial Transce iver

Displ ay part

Ethernet interface Power Supply

Fig. 1. System hardware structure diagram

2.4 System Function Module Design Functional requirements refer to what services the system can provide users, what problems can be solved, and correct and fast feedback on various operations of users. Through the analysis of the needs of each functional module of the auxiliary system, according to the relevant theories of software engineering structural design, the MOOC-oriented course-assisted learning system is divided into six modules. User Management Submodule The user management module includes: realizing user registration function through email or mobile phone number, user login account and password verification, user realname authentication data submission, user name, nickname, password, gender, age and other basic registration information modification, maintenance, and renewal. And other functions [6]. Teaching Management Module The teaching management module is the basis for the news English audio-visual system

260

N. Zhao and Z. Zheng

to carry out news English teaching, and is one of the core parts of the system. According to the functional characteristics, it can be divided into: creating courses, course setting, publishing courses, class management, etc. [7]. Teacher users can manage courses, including adding courses, editing course information, and deleting courses. When a teacher user submits course information, the system will automatically detect whether the format of the submitted information meets the requirements. The class management module enables teachers to manage virtual teaching classes as if they were in reality. Including new classes, teachers fill in class information, and create virtual classes for students to apply to join. Teachers can also view, edit, and delete classes they create. Teachers can import students in batches through templates, and can also review students’ applications to join the class. After the course teaching is completed, it is necessary to test the students’ level to test the students’ mastery of English courses. Teachers can upload tests, etc. in this function module. Educational Resource Recommendation Module The educational resource recommendation module refers to recommending suitable materials, sample essays, statistical analysis, and displaying or exporting statistical analysis results for class students. The recommendation method adopts the mode of combining personalized recommendation and teacher’s supplementary recommendation. The personalized recommendation service recommends learning materials that match the students according to the students’ grades and the characteristics of the class. The specific recommendation process is as follows: Since the similarity between students can not only be calculated by the students’ scores on the recommended items, but also can be analyzed by the users’ tagging of the recommended items. When two students have similar labels on the target item, it can indicate that the two students have similar interests and preferences to a certain extent, that is, the two students have a high degree of similarity, and recommend items based on one of the students Should also be a good fit for another student’s needs. If student u is interested in the label properties of an item, then each student in the set of similar neighbor students to student u should have some commonalities with u’s interest preference. It is assumed that all Tag tag attributes can be represented by a set {Y1 , Y2 , Y3 , ..., Yk }, where Y represents Tag tag attributes; k represents the number of Tag tag attributes. Calculate the preference matrix W of the Tag attribute and the student’s rating data matrix F. ⎡ ⎤ w11 w12 w1k ⎢ ⎥ ⎢ w21 w22 w2k ⎥ ⎥ (1) W =⎢ ⎢ ⎥ ⎣ ... ⎦ wm1 wm2

wmk

Design of Online Education System for Ideological and Political Courses

⎡

f11 f12

⎢ ⎢ f21 f22 F =⎢ ⎢ ⎣ ...

f1n

⎤

⎥ f2n ⎥ ⎥ ⎥ ⎦

fm1 fm2

261

(2)

fmn

In the formula, mk represents the dimension of preference matrix W ; m represents the number of users; wij represents the total weight of the j th label feature of all items evaluated by student i. mn represents the dimension of the user rating data matrix F; fij represents the student i’s rating for item j, and has scored item j in the past period of time, with a rating value of fij = 0, no rating, and fij = 0. Calculate the similarity between the student’s label preference vector and the label preference vector. The calculation formula is as follows: Suv =

k



k

u v i=1 Qi Qi

 u 2 k  v 2 + i=1 Qi i=1 Qi

(3)

In, ⎧ wui u ⎪ ⎨ Qi = wu ⎪ Qv = wvi ⎩ i wv

(4)

In the formula, wui and wvi represent the total weight of students u and v to i Tag attribute of the item; wu and wv represent the total weight of all tag attributes included in all related items commented by students u and v; Qiu , Qiv represents the tag attribute preference vector of students u and v; Suv represents the similarity of the tag preference vector between students u and v. Calculate the final similarity. The final similarity, that is, the comprehensive similarity measure of students u and v, is processed by mixing based on the similarity of the Pearson correlation coefficient and the similarity of the label familiarity preference. Traverse all students to get the similarity matrix between the target user and other users. Generate recommendation results. When recommending a target user, it is mainly to select k adjacent users with a relatively high degree of matching in the generated similarity matrix as its neighbor users, which is customarily called K-neighbors, that is, the user’s neighbor set, which is composed of This process predicts the unknown score of the target user, and generates recommendation results according to the score. Question Bank Test Paper Sub-module In order to let students have a comprehensive understanding of their own learning situation, the system provides an online test platform. After learning a chapter, students can enter the online test module for self-testing. During the test, the system randomly selects a certain number of questions from the question bank to form the test paper. After the students complete the test, the system will automatically grade the paper, and the students can view the relevant answers to the test questions according to their own answers.

262

N. Zhao and Z. Zheng

After logging in to the system, teachers can add test questions to the test question bank of the corresponding course, or modify and edit existing test questions. Online Q&A Module In order to meet the students’ answering needs at different learning stages, the online answering module supports two answering methods at the same time: online answering and offline answering. Online Q&A is implemented using instant messaging technology, and offline Q&A is implemented in the form of message boards. When the teacher is online, students can communicate with the teacher through online Q&A, and when the teacher is offline, students can ask questions through offline Q&A and wait for the teacher to answer, and the teacher’s account will receive a message notification. The function of the Q&A module directly determines the teaching effect. In the system designed in this paper, students can ask questions and feedback on the teaching content through the Q&A module. Teachers can improve the teaching method according to the questions fed back by the students, and evaluate the cognitive level of each student through the quality of the questions, so as to assist the system to provide students with feedback. Recommend teaching resources close to their cognitive level to improve students’ learning efficiency and teachers’ teaching quality. Evaluation Feedback Module In traditional classroom learning, evaluation plays a very important role, such as motivating, guiding and other functions. In this system, the evaluation feedback module is divided into five parts: online survey, homework, test, voting and learning analysis. Online survey means that the evaluation of pre-school is mainly obtained through questionnaire survey, to understand the basics of students and their preparation for learning, etc. The homework module is mainly for students to complete the homework assigned by the teacher after learning the vocabulary, grammar, sentence structure and other knowledge of the course after learning the course. After completing the homework, you can submit it online. The objective questions are corrected by the system according to the answers set by the teacher. There are two forms of correction for subjective questions such as composition and translation. Correction. The course test module is after the unit or course is over, students take the test of the unit or course to test the learning effect of the course, the test module provides the test paper set by the teacher, and the students take the test online within the specified time. Student users can enter the course interface and select a course to test. The system will jump to the corresponding test interface. Students can choose to submit the test after the test is over. The learning analysis module is to evaluate the performance in the learning process. The content of formative assessment includes the assessment of learning attitude, participation, homework and test situation. Through data collection and analysis, the process can be evaluated as accurately as possible, and the learning problems can be summarized and fed back to teachers and students in a timely manner.

Design of Online Education System for Ideological and Political Courses

263

2.5 System Database Design Data occupies a dominant position in business and transaction processing. In recent years, its status and role have become very important in the field of statistics, engineering, in multimedia fields such as graphics, images, and sound, and in the field of intelligence. However, with the change of the status of data in software, the traditional data structure and data-centric programs have relegated to a secondary position. Therefore, the need to centrally and uniformly manage data and take data sharing as the goal has arisen., so there is a modern database system (database system). The database mainly includes a series of course-assisted learning system data such as teachers, students, test questions, test papers, news, and resources. Since the modules of this system are independent and separated, there are not many relationships between the tables, reducing the amount of data. Redundancy. Database plays a decisive role in software design, so we should pay attention to the following principles in database design: The database system should have a hierarchical structure, and the data is summarized and summarized layer by layer from bottom to top. The database structure should be standardized to ensure the accuracy and integrity of the data. Data should be secure, and should have the ability to prevent, detect, and recover from sudden attacks. Design database conceptual model based on database design principles, as shown in Fig. 2.

Student ID Student number

student

user manageme nt

administrat ors

browse

release

Login name password

Student name Post / browse

BBSID

Informatio n ID

author BBS

information

title

title content content

Fig. 2. Database conceptual model diagram

The system database consists of multiple tables, as shown in Table 1 below.

264

N. Zhao and Z. Zheng Table 1. Database table

Table Name

Effect

User Information Form

Basic information of storage system users

Course Information Sheet

Store information about the courses offered by teachers on the platform

Chapter Information Sheet

Stores chapter information for each course

Study table

Record the data generated by the user during the learning process on the platform

Course Note Sheet

Record the user’s experience in the learning process

Test paper content table

Store the content of the test paper

Exam Paper Additional Information Sheet Store additional information on the test paper, such as test paper questions, etc Score table

Store students’ online test scores

Exam paper basic information table

Store the content of the test paper, such as the test paper code, the teacher who issued the test paper, etc

Student response form

Store students’ online test answers

Topic table

Save test questions

Question answer sheet

Store alternative answers to objective questions

3 System Implementation and Testing In the process of software system development, in accordance with the theory, method, principle and steps of software engineering, the development and testing of most of the predetermined functions were completed in about one year, and the system was successfully deployed and put into operation. The ideological and political courses running on the original prototype have been transplanted to this system to run. 3.1 System Testing and Debugging In the software development process, the testing activities are accompanied by the whole process of requirement analysis, system design, and coding implementation. After the main functions of the system are determined by the requirement analysis, the test plan and test cases are prepared in time; in the design process, the design scheme is continuously verified and confirmed. Whether it can effectively support the realization of the use case function, write test cases for different functional modules and different interfaces; in the coding process, the programmer will continuously perform unit testing in the coding process to ensure that the code unit correctly implements the required functions of the unit module; In the process of incremental development of functional modules, integration testing and unit testing are performed alternately, so that the system can be

Design of Online Education System for Ideological and Political Courses

265

effectively integrated according to requirements; after the completion of system functions, a thorough confirmation test of the system is carried out, which is mainly based on the description of the overall functional requirements of the system; finally, After deploying the system to the production environment, several tests for sub-functional requirements were carried out, including security tests, stress tests, etc. 3.2 System Deployment and Migration With the cooperation of the Academic Affairs Office and the Information Center of the college, the system was successfully deployed and implemented on the campus intranet, running in parallel with the original prototype system. After the system is deployed, the basic information of the system is initialized. According to the staffing and process needs, several accounts are established and corresponding permissions are assigned. Import the data, teaching materials, account information, etc. of the ideological and political courses running under the original prototype system into the new system for testing and running, accompanied by minor debugging and modification. 3.3 Personnel Training After the system is deployed and migrated, the relevant teachers, teaching managers and students who are implementing flipped courses and those who plan to implement flipped classroom courses will be trained on the use of the system. The training method adopts a combination of online and offline: offline centralized demonstrations are used to mobilize, and online videos are publicly released to guide relevant personnel to operate the system. 3.4 Test Method System testing is an indispensable step in the development process of all software systems. Although the system has strict technical review procedures during the development process, errors in the development process are still difficult to avoid. If the system is not tested, it will start Putting it into operation, once the system has a problem, it will cost a lot to recover. Therefore, the system testing must have data to show that the development of large-scale high-quality systems, the workload of post-testing is in the overall work. The proportion of the software system is more than 50%, and for some particularly important software systems, this proportion is even higher. Software system testing can be divided into white box testing and black box testing, and only the former is tested in this test. White-box testing refers to whether the system can be executed in accordance with the specified procedures during operation, whether the functions between different software modules can be fully utilized, and whether the internal logic of the software is correct. This type of testing is also called structural testing, logical testing, or program-based testing. It mainly tests the internal structure of the software system, and finds problems and repairs deficiencies.

266

N. Zhao and Z. Zheng

3.5 System Test Results The software simulates the user’s access to the system. During the test, virtual access is performed through the software, and the concurrent access is simulated to increase the pressure on the server, so as to test whether the system can meet the user’s needs. Whether the server can meet user requirements at high concurrency is an important indicator of system performance. Taking the method of Reference [2] as the experimental comparison method, the test results are shown in Table 2 and Fig. 3. Table 2 simulates 500 users accessing the system from different regions at the same time to see if they can access the system normally, and test the average time and bandwidth required for access. Figure 3 simulates high concurrent access to the system, and tests the server’s CPU occupancy rate and memory occupancy rate during access. Table 2. Concurrent test table User number

Average response time/s

Denial of service rate/%

The method of this paper

Reference [2] method

The method of this paper

Reference [2] method

100

0.25

0.68

0

0.23

200

0.56

0.89

0

0.46

300

0.88

1.03

0

0.94

400

1.23

1.74

0.85

1.37

500

1.67

2.21

1.63

2.58

Average network utilization Memory usage Average CPU usage 70 60

The method of this paper

Reference [4] method

Expected results

50 ratio%

40 30 20

10

Fig. 3. System performance test table

Through testing this system, it can be seen that although the number of users gradually increases, the response time also becomes longer, but it can still meet the basic needs

Design of Online Education System for Ideological and Political Courses

267

of users under high concurrency conditions. Due to the application performance of the method in [2]. It can be seen that the system can meet the basic performance requirements.

4 Conclusion To sum up, with the progress and development of Internet technology, MOOC has become an indispensable part of the field of open education. From the beginning of its birth to the increasingly mature development process of MOOC, what has changed is not only the way people live and study, but also the educational concept and teaching mode, and conducts a comprehensive approach to the curriculum form, teaching organization, assessment and other aspects. Change and innovation. With the help of the MOOC model, the online education system for ideological and political courses of traditional Chinese medicine is designed. After the system design is completed, after systematic testing, the system runs stably and reliably, meeting the needs of various teaching services. With the gradual increase in the number of users, the system responds smoothly, and various functions run stably, basically meeting the expected demand for the platform. Due to the limited level of time, experience and professional knowledge, in the whole design and development process, the system has some deficiencies more or less, which needs to be further studied and improved in the future use: the MOOC system on the PC side is stable and mature. After that, the next step will be to consider developing a MOOC platform for mobile clients, which is bound to make MOOC learning more convenient and realize the sustainable development of the MOOC platform. Fund Project. 1. Research and practice of entrepreneurship education for college students in independent colleges. Social science research project of the 13th five year plan of Jilin Provincial Department of education, Ji Jiao Wen Ke He Zi [2016] No. 508 2. The integration of production and education of traditional Chinese medicine and the construction of innovation project of traditional Chinese medicine industry. Institute local cooperation project of Chinese Academy of Engineering (subject 2), subject No.: JL 2020–005

References 1. Wang, W., Wu, H., Hou, M.: Personalized dispatch simulation of distributed online learning resources under the Internet of Things. Computer Simul. 36(1), 417–420, 479 (2019) 2. Degang, D., Shufang, B., Mei, S., et al.: Exploration on curriculum politics teaching reform of the biochemistry course for traditional chinese medicine majors:taking “Enzymology” as an example. Chin. J. Biochem. Molecular Biol. 37(7), 983–988 (2021) 3. Liang, J.M., Su, W.C., Chen, Y.L., et al.: Smart interactive education system based on wearable devices. Sensors 19(15), 3260 (2019) 4. Saito, T., Watanobe, Y.: Learning path recommendation system for programming education based on neural networks. Int. J. Distance Educ. Technol. 18(1), 36–64 (2020) 5. Sheng, D., Yuan, J., Xie, Q., et al.: ACMF: an attention collaborative extended matrix factorization based model For MOOC course service via a heterogeneous view. Futur. Gener. Comput. Syst. 126(12), 211–224 (2021)

268

N. Zhao and Z. Zheng

6. Hermanns, J., Schmidt, B., Glowinski, I., Keller, D.: Online teaching in the course “organic chemistry” for nonmajor chemistry students: from necessity to opportunity. J. Chem. Educ. 97(9), 3140–3146 (2020). https://doi.org/10.1021/acs.jchemed.0c00658 7. Danaher, M., Schoepp, K., Kranov, A.A.: Teaching and measuring the professional skills of information technology students using a learning oriented assessment task. Int. J. Eng. Educ. 35(3), 795–805 (2019)

Assistant Teaching System of Human Resource Management Course Based on Data Mining Ying Ye(B) College of Labor Relations and Human Resources, China Institute of Labor Relations, Beijing 10048, China [email protected]

Abstract. The traditional assistant teaching system of human resource management course has some problems, such as small data transmission capacity, long system response time and so on. For this reason, this paper proposes an assistant teaching system of human resource management course based on data mining. Establish the snort plug-in mechanism, improve the connection behavior of the course detection engine according to the data mining principle, and then store the teaching information in the database host with the help of the transmission channel organization, so as to build the software execution environment of the teaching system, and complete the design of the auxiliary teaching system of human resource management course based on data mining in combination with the structure of relevant hardware equipment. The experimental results show that, compared with the traditional teaching system, under the action of the data mining assisted teaching system, the response speed of the teacher host and the student host has been effectively improved, which can better solve the problem of obvious accumulation of human resource management course data, and meet the actual application needs. Keywords: Data mining · Human resource management · Course assistant teaching · Student users · Snort plug-in · Course detection engine

1 Introduction Data mining is a hot issue in the field of artificial intelligence and database. The so-called data mining refers to a non trivial process of revealing implicit, previously unknown and potentially valuable information from a large amount of data in the database. Data mining is a decision support process. It is mainly based on artificial intelligence, machine learning, pattern recognition, statistics, database, visualization technology, etc. it highly automatically analyzes enterprise data, makes inductive reasoning, excavates potential patterns, and helps decision makers adjust market strategies, reduce risks and make correct decisions [1]. The process of knowledge discovery consists of the following three stages: ➀ data preparation; ➁ Data mining; ➂ Expression and interpretation of results. Data mining can interact with users or knowledge bases. Data mining is a technology to find the law from a large amount of data by analyzing each data. It mainly includes three © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 269–280, 2022. https://doi.org/10.1007/978-3-031-21161-4_21

270

Y. Ye

steps: data preparation, law search and law representation. Data preparation is to select the required data from relevant data sources and integrate them into a data set for data mining; Law search is to find out the laws contained in the data set by some method; The tasks of data mining include association analysis, cluster analysis, classification analysis, anomaly analysis, special group analysis and evolution analysis. With the development of computer technology, computer has become a part of people’s life, and the Internet has connected the networks all over the world as a whole. Computer technology is changing people’s study, life and work. The organic combination of network and education has a great impact on people’s traditional educational mode, thinking, content Methods and talent training programs have a significant impact. The development of teaching means from tradition to computer technology has made a new development and leap in educational technology. For human resource management courses, the establishment of computer-aided teaching system and the use of the system for teaching services can not only save teachers’ after-school counseling time, but also help students solve difficult problems in the learning process of computer courses. The test of basic theoretical knowledge in the auxiliary teaching system can help students consolidate the learning effect of the classroom [2, 3]. The use of interactive computer-aided instruction system can provide an interactive communication platform for teachers and students, facilitate teachers’ curriculum guidance and improve the efficiency of solving problems. The design and implementation of CAI system can change the existing traditional teacher centered teaching mode and form a new student-centered action oriented teaching mode. Literature [4] puts forward an auxiliary teaching system of human resource management course based on fuzzy logic [4], extracts the auxiliary teaching system data of human resource management course through data mining method, realizes the classification of teaching system through analytic hierarchy process, and realizes the auxiliary teaching system design of human resource management course according to fuzzy logic. This method can improve the data transmission capacity, but the system response time is too long. Literature [5] proposes an augmented reality based assisted instruction system for human resource management courses [5]. The teaching resources of human resource management courses are obtained through big data analysis method, and the augmented reality method is used to realize assisted instruction of human resource management courses. This method can shorten the response time, but the data transmission capacity is small. In view of the above problems, this paper proposes an assistant teaching system of human resource management course based on data mining. According to the principle of data mining, improve the connection behavior of the course detection engine, and then store the teaching information in the database host with the help of the transmission channel organization, so as to build the software execution environment of the teaching system, and complete the design of the human resource management course auxiliary teaching system based on data mining. This method can better solve the problem of obvious accumulation of human resource management course data, and meet the needs of practical application.

Assistant Teaching System of Human Resource Management Course

271

2 Hardware Design of Auxiliary Teaching System for Human Resource Management Courses 2.1 Student User Registration Student user registration is an important part of the auxiliary teaching system of human resource management course. Student users can log in to the system only after user registration through student number. After the student user enters the login page, click the “registered account” button on the page to register the user. The student ID is the only identification that distinguishes students. After entering the registration page, student users need to use the student ID as the user name for registration. When submitting information, the system will detect whether the user name has been registered. If the user name is registered, it will prompt that the registration fails and need to be re registered [3]. If the input information is wrong, such as the email address is wrong, a prompt will be given and you can continue to register after returning; If the registration is successful, you will be prompted on the registration success page. Click the login system connection to return to the system login page for user login. In order to facilitate the use of the participants in the auxiliary teaching system, the data mining framework has established an object-oriented, hierarchical, unified model, and related collection of class libraries with good scalability, which is called API here. The function of this class library is the same as the MFC base class library provided in C++ and the Java class library in Java. It reduces the developer’s writing of student user registration code and improves work efficiency. These class libraries encapsulate a lot of powerful and practical program codes for the auxiliary teaching system. In the data mining framework, all the class libraries established by the human resource management course are integrated and applied to the framework. At the same time, the data mining class library itself has strong compatibility with the development language, has its own debugging function, and has a lot of Good error handling function. The specific module diagram is as follows (Fig. 1).

Api-like structures in data mining frameworks Human resource Management course information database Web forms, Web services, Windows forms

Student user registra-tion module

Curriculum auxiliary teaching system operation interface Human resource management course information inquiry

Fig. 1. Schematic diagram of student user registration module

272

Y. Ye

In the entire student user registration module, the top layer is various types of programming languages. The data mining framework supports the development of multiple languages, such as C#, C++, VB, etc. A language specification process is designed for the programming language in the framework, as long as you follow this The programming language can be used at will according to the relevant requirements of the common language specification. 2.2 Question Answering Module The question answering module provides a platform for student users to communicate with teachers and students. When student users have questions in the process of learning human resource management courses, they can ask questions through this module and consult the teacher’s reply. At the same time, they can consult the questions raised by other students and the teacher’s answers in this module. After the student user submits relevant question information through the question page, the information will be submitted to the teacher background question answering management module. The teacher user can conduct corresponding management operations on the questions and review the questions raised by the student user. After passing the review and replying, the students will be able to see the corresponding information in the question answering module. Different from the student user registration module, the question answering module is mainly the object of processing human resource management course data, and it is also the core of data mining technology. The database can obtain the data through the above-mentioned data-related providers, and then temporarily store the data in the memory through the DataSet. It is a collection of many tables, not only to obtain data, but also to associate and constrain the various tables of the database through metadata. The DataSet object mainly contains two objects, namely DataTable and DataRelation. These two objects can obtain the relevant properties and methods of the DataSet, and Data mining framework

Database host

The DataSet host

Q&a module

Data mining Framework registration form The DataTable object

DataRelation object

Fig. 2. Connection principle of the question answering module

Assistant Teaching System of Human Resource Management Course

273

then manipulate the data [4]. The description of the data is realized through XML. The DataSet completes the reading and writing of the data in the form of an XML document, and then transmits the data through HTTP and provides it to other applications for related operations. For some distributed applications, the DataSet object can enhance the interactivity of the program. The complete module connection structure is as follows (Fig. 2). The question answering module mainly provides students with the ability to look up question information. The assigned questions are mainly updated and released by the teacher in the background. The completed question information is the upload folder under the homework directory. The type of information submitted by the students is set to RAR or ZIP format, and the question information submitted by the students is established The data folder with the current date as the folder name. 2.3 Data Statistics Module The data statistics module belongs to a kind of hardware application server structure, which can publish some small human resource management course sites, and can also provide services for auxiliary teaching hosts. The operation of this module must rely on Microsoft Windows NT and Windows 2000 servers. On these application servers, an integrated search engine is implemented, which allows student users to create required search tables through multiple data mining techniques, such as ASP and SQL query statements. At the same time, the browser used by the remote node can also manage the existing server host, thereby establishing multiple different virtual hosts. With the support of data mining technology, the data statistics module can also use ASP technology to build dynamic web pages [5]. At present, the combined architecture of Windows 2000 and Windows NT can make the Web-side server closer to the actual application needs of students. The corresponding configuration of IIS is relatively simple and convenient. For the auxiliary teaching system of human resource management courses, the existence of this module can Realize the query and arrangement processing of related teaching information. This module mainly includes three main subsystems, one is the real-time virtual teacher system, the second is the learning management system, and the third is the subsystem of learning content management. All subsystems are also implemented in a three-tier architecture. The development technology is based on the component model, and the system’s development back-end database can support multiple database designs. In the module design process, it is necessary to define nodes such as.NET and.ASP. The specific node definitions and naming methods are shown in Tables 1 and 2. To sum up, the development of data statistics module is suitable for human resource management courses, which has certain practical significance and advantages. The independent research and development of teaching assistance system is convenient for teachers to make appropriate adjustments to the teaching plan at any time, and can save a lot of course adjustment time. At the same time, with the emergence of new requirements, teachers can carry out secondary development of the system, so as to avoid the disconnection of the courses taught.

274

Y. Ye Table 1. Node naming method of data statistics module

The name of the node Named form .NET

Translation of human resource management course data in the form of decoding

.ASP

Encode human resource management course data

.XML

Identify the ability to transmit human resource management course data

.VS

Human Resource Management course data transmission direction

.SQL

Human resource management course data link form

.CLR

Human Resource Management course data storage location

Table 2. Ideal connection time Course data transmission amount /(Gb)

Connection Duration /(ms)

1.0

0.15

2.0

0.34

3.0

0.47

4.0

0.56

5.0

0.62

6.0

0.70

7.0

0.71

8.0

0.71

3 Software Design of Auxiliary Teaching System for Human Resource Management Courses With the support of the data mining framework, in accordance with the processing flow of SNORT plug-in mechanism settings, course detection engine connection, teaching information storage database construction, the software execution environment of the auxiliary teaching system is completed, and the relevant hardware application structure is combined to realize data mining-based The smooth application of the teaching system assisted by human resource management courses. 3.1 SNORT Plug-In Mechanism As an open source application, SNORT is a prerequisite for designing a data mining framework, allowing users to implement specific application-specific detection modules on their platform in the form of plug-in modules. For the auxiliary teaching system of

Assistant Teaching System of Human Resource Management Course

275

human resource management courses, the plug-in mechanism of SNORT is implemented in C language, which mainly includes the following two parts: (1) Plug-in management program: Mainly responsible for the installation, registration, exit, restart and other operations of the teaching plug-in. (2) Various teaching plug-in function realization programs: mainly realize the function of course detection, and at the same time call auxiliary plug-ins to realize teaching logs or other execution actions. The plug-in management program corresponding to each SNORT node is Plugbase. C. It is also the basic file of the data mining framework. It completes all plug-in management and service functions, provides installation functions for all plug-ins and some common functions used by plug-ins [6]. It provides 4 identical plug-in management interfaces for detection plug-ins, output plug-ins, and pre-processing plug-ins. With the complexity and large-scale structure of the auxiliary teaching system of human resource management courses, network detection methods tend to be diversified, and data transmission behavior is no longer a single behavior, but a situation of mutual cooperation and influence, but based on data control behavior The teaching system cannot meet the real-time connection requirements of the client host and the teacher host. Under the effect of the SNORT plug-in mechanism, the data mining framework will continue to be more detailed, which will not only make the teacher host and the student host Presenting a stable connection state will also greatly reduce the pressure of course information storage faced by the database host [7]. Let χ0 represent the initial coding coefficient of the SNORT node, χn represent the actual coding coefficient of the SNORT node, n represent the actual number of SNORT nodes in the data mining framework, and E represent the average output of human resource management course data per unit time, rˆ represents the mining feature value of human resource management course data, and f represents the real-time mining coefficient. Combining the above physical quantities, the function capability of the SNORT plug-in mechanism based on data mining can be expressed as:    (χn − χ0 )rˆ · f  (1) W = 2 nE Since the components in the auxiliary teaching system should be composed of multiple modules that can work independently, the data mining framework must have strong adaptability. It needs to process relevant human resource management course data with the support of the SNORT plug-in mechanism. As a result, the information transmission capacity in the system environment gradually stabilizes. 3.2 Course Detection Engine The function of the course detection engine is to use the network normal behavior class to detect the human resource management course data packets and filter those network normal data packets, so as to improve the real-time connection speed of the auxiliary

276

Y. Ye

teaching system [8]. With the support of data mining framework, the setting of course detection engine must follow the following principles: (1) Standardize the application network of auxiliary teaching system. (2) Calculate the similarity level between the data package of human resource management course to be mined and each auxiliary teaching information. (3) If the similarity between the human resource management course data package and all auxiliary teaching information is greater than the mining radius, it indicates that it is an abnormal network data package, and it is fed back to the SNORT plug-in machine for further inspection. (4) If the similarity between the human resource management course packet and some kind of auxiliary teaching information is less than the mining radius, it indicates that it is a normal network packet and can be discarded. The course detection engine must be supported by a complete human resource management course database. While judging the accuracy level of teaching information, it determines the real-time connection relationship between the teacher host and the student host. On the one hand, it can meet the storage requirements of the database host for teaching information, on the other hand, it can also avoid excessive occupation of system application space by the information to be stored. Suppose imin represents the minimum execution authority value of data mining instructions, imax represents the maximum execution authority value of data mining instructions, g  represents the screening coefficient of human resource management course data, β represents the detection execution coefficient of auxiliary teaching instructions, and h represents auxiliary teaching instructions The average value of transmission in unit time, D represents the storage value of human resource management course data in unit time. With the support of the above physical quantities, the simultaneous formula (1) can define the application requirements of the course detection engine as:     imax    2 (g − βh) d h   δ=  W × |D|2 imin

(2)

Through the idea of data mining to divide the human resource management course data, we can analyze the specific location of class nodes in the course detection engine, which is the key application problem of clustering mining framework. 3.3 Teaching Information Storage Database The analysis of the detection engine based on data mining shows that the application of human resource management course assisted teaching system also needs to design the necessary storage database structure. The purpose of database design is to organize the data according to a certain model according to the existing application environment, realize the functions of storage, maintenance and retrieval, so as to construct an optimal database system, so that the information system can easily, timely and accurately obtain the required information from the database and meet the needs of users. Database is the

Assistant Teaching System of Human Resource Management Course

277

foundation and core of information resource management [9]. It generally includes the following design links: (1) Demand analysis stage: to design the database, we must first understand the customer’s demand. Even if the demand will change dynamically in the future, it can still ensure that the demand is within the scope of development purpose. The focus of demand analysis is to investigate and analyze the user’s information requirements in data management to ensure its integrity requirements. (2) Logical structure design stage: In this stage, it is necessary to find the data model that is most consistent with the corresponding conceptual structure, construct the DBMS, and make storage arrangements according to the requirements of the DBMS to form an internal database model [10, 11]. (3) Conceptual structure design stage: According to user needs, comprehensive and generalize, abstract the conceptual model of DBMS, and express it with E-R diagram. (4) Database implementation and maintenance stage: In this stage, the database is established, application programs are compiled and debugged, operation tests are performed, and the database is continuously adjusted and modified. ˜ repreSuppose j represents the real-time storage authority of teaching information, Q sents the characteristic value of the human resource management course data to be stored, φ represents the connection coefficient of the teaching information storage database, and h˙ represents the necessary auxiliary teaching application indication. With the support of the above physical quantities, the simultaneous formula (2) can express the connection capacity of the teaching information storage database as: ˜ δ 2 × jQ S=   2 ˙φ2 + 1

(3)

h

Teaching information database is responsible for storing information parameters related to human resource management courses, and can query the feasibility of existing stored information under the action of data mining instructions.

4 Case Analysis In order to highlight the applicability difference between the auxiliary teaching system of human resource management course based on data mining and the traditional teaching system, the following comparative experiment is designed. A campus host with relatively stable operation ability is selected as the experimental object. Firstly, the data mining program is input into the host element. After running for a period of time, the connection time between the teacher host and the student host is recorded, and the obtained information is taken as the variable of the experimental group; Then disconnect the experimental group, input the routine execution program into the host element, after running for a period of time, record the connection time between the teacher host and

278

Y. Ye

the student host again, and take the obtained information as the variable of the control group; Finally, the variable indexes of the experimental group and the control group were compared. The connection time between teacher host and student host can reflect the accumulation ability of human resource management course data in the process of operation. In general, the shorter the connection time between the teacher host and the student host, the weaker the accumulation capacity of human resource management course data in the operation process, and vice versa. The following table records the change of the ideal value of the connection time between the teacher host and the student host (Tables 2). Analyzing Table 1 shows that, under ideal conditions, the connection time between the teacher-side host and the student-side host presents a continuous increase and then a continuous and stable value change state. The global maximum value reaches 0.71 ms, which is compared with the initial value of 0.15 ms. A rise of 0.56 ms. The following figure reflects the experimental numerical record results of the response time between the teacher host and the student host. Connection Duration/(ms)

0.9 0.8

Ideal numerical 0.7 The experimental group

0.6 0.5

The control group 0.4 0.3 0.2 0.1 0

0

1.0 2.0 3.0 4.0 5.0 6.0 7.0 Course data transmission amount/(GB)

8.0

Fig. 3. Experimental results of connection duration (the first group)

It can be seen from the analysis of Fig. 3 that with the increase of the course data transmission volume, the connection duration of the experimental group and the control group showed an increasing numerical change trend, but it was obvious that the average level of the response duration of the experimental group was lower. During the whole experiment, the maximum connection duration of the experimental group was 0.52 MS, which decreased by 0.19 MS compared with the ideal maximum of 0.71 Ms; The maximum response time of the control group was 0.80 MS, which increased by 0.09 MS compared with the ideal maximum of 0.71 MS, much higher than that of the experimental group.

Assistant Teaching System of Human Resource Management Course

279

Connection Duration/(ms)

0.9 0.8

Ideal numerical 0.7 The experimental group

0.6 0.5

The control group 0.4 0.3 0.2 0.1 0

0

1.0 2.0 3.0 4.0 5.0 6.0 7.0 Course data transmission amount/(GB)

8.0

Fig. 4. Response time test results (the second group)

It can be seen from the analysis of Fig. 4 that with the increase of the course data transmission volume, the response time of the experimental group still keeps increasing. During the whole experiment, the maximum numerical result reaches 0.64 MS, which is 0.07 MS lower than the ideal maximum value of 0.71 Ms; The response duration of the control group kept the numerical trend of first rising, then falling, and finally rising again. Throughout the experiment, the maximum numerical result reached 0.78 MS, which increased by 0.07 MS compared with the ideal maximum value of 0.71 MS, much higher than the numerical level of the experimental group. To sum up, under the action of the auxiliary teaching system based on data mining, with the increase of the data transmission volume of human resource management course, the connection time between the teacher host and the student host has been well controlled, which can solve the problem of obvious accumulation of human resource management course data.

5 Conclusion Compared with the traditional teaching system, the new auxiliary teaching system, under the function of data mining technology, re plans the actual connection ability of student user registration module, question answering module and data statistics module, and combines the snort plug-in mechanism to restrict the application ability of the course detection engine, so as to stimulate the rapid storage of course data information in the teaching information database. The experimental results show that this new assistant teaching system can avoid the accumulation of human resource management course data, and has strong practical value. Although this method has achieved high data response efficiency, it costs a high cost. Therefore, how to reduce the system cost needs further research.

280

Y. Ye

References 1. Cuihong, L.: Research on the integration of online teaching platform and accounting information system course teaching practice. Finance Acc. Learn. 19(03), 208–209 (2020) 2. Gen, L., Jie, J., Yan, Z., Zexun, Z.: Private cloud based multi course sharing assisted instruction system. J. Electr. Electron. Educ. 42(01), 10–13+51 (2020) 3. Rui, X.: Design of computer aided instruction system for oncology course based on data mining. Microcomput. Appl. 38(04), 22–25 (2022) 4. Jafari, M., Malekjamshidi, Z., Lu, D.C., et al.: Development of a fuzzy-logic-based energy management system for a multiport multioperation mode residential smart microgrid. IEEE Trans. Power Electron. 34(4), 3283–3301 (2019) 5. Nan, X., Wenhui, F.: Research on interactive numerical optimization augmented reality teaching system. Comput. Simul.37(11), 203–206+298 (2020) 6. Nieto, R.M., Garcia-Martin, A., Hauptmann, A.G., et al.: Automatic vacant parking places management system using multicamera vehicle detection. IEEE Trans. Intell. Transp. Syst. 20(3), 1069–1080 (2019) 7. Wang, Y., Moura, S.J., Advani, S.G., et al.: Power management system for a fuel cell/battery hybrid vehicle incorporating fuel cell and battery degradation. Int. J. Hydrogen Energy 44(16), 8479–8492 (2019) 8. Dutta, S.K., Laing, A.M., Kumar, S., et al.: Improved water management practices improve cropping system profitability and smallholder farmers’ incomes. Agric. Water Manag. 242(03), 106411–106419 (2020) 9. Verdegay J.L., Rodriguez Z.: A new decision support system for knowledge management in archaeological activities. Knowl. Based Syst. 187(7), 104843.1–104843.10 (2020) 10. Ebrahimi, J., Niknam, T., Firouzi, B.B.: Electrical and thermal power management in an energy hub system considering hybrid renewables. Electr. Eng. 103(4), 1965–1976 (2021)

Design of Virtual Experiment Online Teaching System for Economics and Management in Colleges and Universities Based on Virtualization Technology Jianhua Zhang1 , Yingji Luo1(B) , and Qiuhui Yang2 1 Guangxi University of Finance and Economics, Nanning 530007, China

[email protected] 2 Guangxi Colleges and Universities Key Laboratory of Image Processing and Intelligent

Information System, Wuzhou University, Wuzhou 543002, China

Abstract. The traditional experimental teaching method is single, and students lack learning initiative and creativity, which leads to the problem of insufficient teaching quality. Therefore, this paper proposes a virtual experiment online teaching system of university economics and management based on virtualization technology. Firstly, optimize the system hardware structure and improve the allowable effect of system equipment configuration. Then, the three-tier structure of customer layer, application service layer and data layer is used to optimize the system software functions, and simplify the working principle and multi-layer architecture of the online virtual experiment system based on B/S mode, so as to achieve the online teaching goal of economic and management virtual experiment in colleges and universities. The experimental results show that the system has high practicability, can effectively improve the teaching quality, and fully meet the expected requirements of the research. Keywords: Virtualization technology · Economics and management · Virtual experiment · Online teaching

1 Introduction Experiment is an indispensable part of higher education, distance education and science popularization. It plays an irreplaceable role in cultivating students’ observation and experimental ability, realistic scientific attitude and arousing learning interest. The current experiment is basically carried out in the laboratory. The operation mode has some disadvantages, such as large investment, large loss, low efficiency, long cycle, difficult maintenance and so on [1]. In the virtual laboratory realized by computer technology, the experimenter can complete the predetermined experimental projects as in the real environment, and the learning effect is better than that in the real environment in some aspects. In recent years, © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 281–295, 2022. https://doi.org/10.1007/978-3-031-21161-4_22

282

J. Zhang et al.

due to the rapid development of virtual instrument and network technology, it is possible to build a virtual laboratory through the network. The so-called virtual laboratory refers to the effective combination of computer and instrument hardware by using the powerful software and hardware resources of computer and application program and bus technology. Users can arbitrarily combine a variety of instrument modules according to their own needs, design and build their own experimental instrument system, and realize a variety of instrument functions on one computer with the help of software panel, It has the characteristics of high performance, high reliability, low cost and convenient modification [2]. The network-based virtual laboratory realizes resource sharing through the network. Using virtual instrument technology and network technology, users can access the Internet or campus network to operate and use the instruments in the laboratory. At the same time, it can realize the integration of data collection, transmission, processing and analysis, realize multi-person collaborative work, complete course experiments, and provide a more open and Free and easy to manage mode and means. The online virtual experimental teaching system proposed in this paper essentially downloads the man-machine interface of the experimental system to the local through the Internet or campus network, so as to realize the operation of remote experimental instruments, such as modifying parameters, adding interference, analyzing experimental results, real-time observing the changes of curves, etc. [3]. Virtual laboratory can make the use of experimental resources more convenient, without the need to build a professional hardware and software simulation environment locally, and there is no need to download the experimental program source code locally, which facilitates the course experiment, especially the development of open experiment. Based on the above analysis, this study designs an online teaching system of virtual experiment of economics and management in Colleges and Universities Based on virtualization technology.

2 System Design 2.1 System Hardware Structure The virtual laboratory is combined with the campus network, so that students are not restricted by time, location and teaching resources; the experiment arrangement is flexible and convenient, as long as the campus network covers the place, you can do experiments independently, and teachers can also correct and guide without restrictions student experiments greatly facilitate the learning and teaching of teachers and students. The virtual experimental teaching system relieves the pressure of insufficient experimental equipment [4]. When the number of individual professionals increases significantly, virtual experiments can solve the pressure of the shortage of experimental instruments. In the case of combining virtual experiments and real experiments, the quality of teaching can be guaranteed and the teaching tasks can be completed [5]. Web based virtual experiments require almost no consumables, which greatly reduces the cost of experiments. There is no need for group experiments. Everyone has the opportunity to do it. Some high-risk experiments and high-cost experiments can also be repeated multiple times, which improves students’ interest and Quality. Virtual experiments can reduce

Design of Virtual Experiment Online Teaching System

283

the maintenance intensity of experimental equipment, reduce the work intensity of laboratory managers, and improve work efficiency. The virtual experiment teaching system adopts a three-layer structure based on B/S to realize, and its network structure is shown in Fig. 1. Management terminal

Web server

Application server

Database server

Office

Campus network

Computer Room

Internet

Student dormitory

Remote user

Fig. 1. System network structure diagram

This three-tier Web-based structure implements the main application logic on the application server in the middle tier. The first layer is the client program, that is, the user interface program, which realizes its functions by using the business services provided by the middle layer components. The client configuration is very simple, and it only needs to install the Windows operating system [6]. The customer uses the browser to complete the operation. The middle tier is the application service tier, which includes many independent middle tier components that provide various services. The access to the database adopts the method of object embedding engine, which avoids the inconvenience of using ODBC to set up. The third layer is the data service layer, and the system uses Microsoft SOL Server as the database server. The virtual experiment platform is similar to the real experiment platform, which is used for students to configure, connect, adjust and use experimental equipment for experiments. The virtual experiment platform allows the free construction of any reasonable typical experiment or experimental system case according to the equipment provided on the platform, which is an important feature of the virtual experiment platform different from the general experimental teaching courseware [7]. Typical experiments are usually designed by teachers who are proficient in relevant courses. Students are required to carry out experiments on this basis, which can meet the needs of teachers for experimental teaching at all levels. The virtual experimental platform also makes it possible for students to build experimental models freely. Students can not only operate through the virtual experiment platform, but also design experiments independently, which is conducive to cultivating design ability and innovation consciousness.

284

J. Zhang et al.

In the virtual experiment teaching system of B/S architecture, the virtual experiment platform is usually made into a Java applet, Activex or Flash plug-in, which provides functions such as interface simulation, model representation, model solving and operation control. In a three-tier Web-based structure, the system administrator only needs to maintain the application server, which improves work efficiency. Under the guidance of the teaching interaction level tower, operation interaction, information interaction, and concept interaction are integrated into the creation of each module of the virtual experiment teaching system, and the overall framework of the virtual experiment teaching interaction system is constructed as shown in Fig. 2.

Virtual experiment support module

Experimental application Experimental operation Real time release of experimental results

Human computer interaction module real time

Non real time

teacher

chat

real time

student

chat

real time

Operating system, experimental diagnosis, controllable object presentation, etc

Virtual experiment teaching system

Virtual experiment attribute module

Fig. 2. Virtual experimental teaching hardware equipment interactive system framework

The virtual experiment auxiliary support module provides learners with the necessary knowledge before the experiment operation, so as to realize the interaction between learners and learning resources. While learners interact with learning resources, they interact with their original concepts and concepts contained in new learning information to realize the transformation between old and new knowledge [8]. Virtual experiment operation module is the main part of virtual experiment teaching system. Learners realize operation interaction, information interaction and concept interaction through their own virtual operation and real-time feedback given by the system. Interpersonal interaction module is the decisive module for the success of virtual experiment teaching. This module provides real-time and non real-time interpersonal interaction between learners

Design of Virtual Experiment Online Teaching System

285

and learners, learners and teachers, so as to complete the interaction between learners’ new and old concepts and achieve the experimental goal. 2.2 System Software Function Optimization According to the demand analysis and the characteristics of network virtual experiment teaching system, we divide the system into two modules: experiment teaching and experiment management. The main structure is shown in the figure. The experimental teaching module includes virtual experimental platform, experimental teaching materials and experimental reports. The experimental management module has a series of auxiliary functions and provides a maintenance platform for experimental managers. The experimental management module provides a communication channel between the virtual experimental system and the personnel participating in the experiment. These auxiliary functions better help teachers complete experimental teaching, It helps students complete the experimental study and realizes the goal of managing virtual experiment informatization [9]. Based on this, the system function module structure is optimized, as shown in Fig. 3.

Experimental Preview Experimental teaching module

Experimental operation

Experimental report Network virtual experiment teaching system Teacher interface Test management module

Student interface Test administrator interface

Fig. 3. System function module structure

The virtual experiment auxiliary support module is an interactive module for learners to understand and learn the relevant information of the experiment in detail after logging in the virtual experiment teaching system. The auxiliary support module presents the experimental content introduction, experimental objectives, experimental principles, experimental steps and experimental applications to learners in the form of text knowledge box or demonstration video, so as to realize the interaction between learners and learning resources and the interaction between learners’ new and old concepts, so that learners can master the relevant detailed information of this experiment, it has sufficient theoretical support and goal orientation for the subsequent experimental operation [10].

286

J. Zhang et al.

Part of the experimental preparatory module is the electronic content of experimental guidance materials including experimental principles, experimental objectives, experimental requirements, experimental guidance, etc., which are unified and summarized in the experimental teaching document module for students to consult textbook knowledge at any time, using text presentation mode. After fully previewing the theoretical knowledge, enter the experimental instrument introduction module to become familiar with the experimental instrument, and it is required to master the main performance, operation method and use function of the instrument. This module allows students to intuitively familiarize themselves with the instrument scene in the experiment through the Flash two-dimensional animation demonstration of the experimental virtual venue and equipment. Use Flash technology or VRL technology to present the experiment principle process with multimedia technology, and consolidate the experiment knowledge such as experiment purpose, content, equipment, etc. at the same time. “Experimental instrument display” introduces the functions and usage methods of network equipment and transmission media used in the experiment through Flash animation: “Online simulation experiment” is the core part of the virtual experiment and is a place for virtual experiments with reference to experimental principles; “Experiment report submit” is to provide students with an experiment report template. Students can fill in the experiment report and submit it according to the experiment report template. The data of the experiment report is stored in a remote server for teachers to make unified corrections and unified management. The business process of the entire system is shown in Fig. 4.

student

Experimental Preview

Virtual experiment

teacher

Experimental effect test

Experimental report

Score query

field test

Fig. 4. Business process of the experimental teaching system

Students can communicate with teachers and classmates online to get help for the problems they encounter in the experiment. Teachers publish experimental notices and experimental requirements on the network platform, upload experimental teaching

Design of Virtual Experiment Online Teaching System

287

resources and answer students’ questions. In the experimental effect test, understand whether the completion of this virtual experiment has achieved the established purpose. Students may encounter various problems that cannot be solved independently during the experiment. They can discuss and ask questions through the course forum module. The work flow of the student module is shown in Fig. 5. Because “software is instrument” and “software is component” in virtual experiment, after using the virtual experiment teaching system, the problems such as incomplete components, rare shortage of high-grade instruments and difficult management can be solved. Before entering the real experimental environment, students must pass the virtual experimental test, master the relevant experimental knowledge, and be familiar with the working principle and operation steps of relevant experimental instruments. In this way, students can avoid misoperation of experimental instruments, effectively reduce the damage rate of experimental instruments, and correspondingly improve the utilization rate of experimental instruments. Student Internet

Yes Sign in No Enter the virtual experiment system

Experimental Preview

Conduct virtual experiments

Something the matter

Yes

Course discussion

No Fill in the experiment report

Next experiment?

Yes

No End

Fig. 5. Student module workflow

288

J. Zhang et al.

2.3 Realization of Online Teaching of Virtual Experiment In order to increase the realism of the scene, it is necessary to avoid the behavior of going through the wall or the learner’s character model passing through the experimental instrument during the virtual experiment operation. Collision detection is the most important item. Torque supports advanced collision detection. The containerraycas function is used in Torque for collision detection in the scene. The basic principle is: the engine has a global collision box container. When each sceneject’ is created, it will not only be added to the sceneGrap h for rendering, but also will be added to a collision box container. The collision box of the object is from Analyzed in the model. The algorithm for determining the collision box is to use the mean s and the quadratic covariance matrix statistic m to calculate its position w and direction e. Suppose the vertex vector of the n triangle is ai , bi , and ci , and the number of triangle faces enclosed by the collision box is z. Then the center position of the collision box is:   1  w/e − ai + bi + ci + zs 3n n

m=

(1)

i=1

The covariance matrix elements are:   1 Cjk = ai + bi + ci − we 3(n − m) n

(2)

i=1

The architecture of the large-scale online virtual experimental teaching platform is shown in the figure. The platform architecture mainly includes four aspects: support platform, basic resources, application services, and interactive expansion. The architecture design of the online virtual experiment teaching platform is shown in Fig. 6. The virtual experiment setting curriculum plan is also the permission of the administrator / teacher role. After creating the course, the teacher sets the course plan according to the class hours and teaching contents of the course, how many classes are divided in total, what will be taught in each class, and attach relevant courseware or other materials. If there are video materials, you can add the online on-demand function. In addition, the system will automatically attach the online live broadcast function. The flow chart of creating a course plan is shown in Fig. 7. The experimental operation sub module is the core content of the whole experimental teaching module, and the online virtual experiment module is the core of the core. The module has the characteristics of virtualization, interactivity, openness, practicality, sharing and reusability. It can “connect” with a variety of virtual experiment opening technologies and support multi-user real-time online. New virtual experiment courses can be added continuously, with good extensible function. Network virtual experiment can not separate teaching and experiment, and can practice while learning in the experiment. It not only improves the efficiency of the experiment and easily works as a group, but also stimulates the spirit of students’ autonomous learning. The teaching demonstration sub module has two demonstration schemes, one is the call demonstration inside the system, and the other is the teacher’s real-time online demonstration. The teacher can demonstrate the virtual experiment first, store the video

Design of Virtual Experiment Online Teaching System Interactive

Application service

Support platform

289

Basic resources

Experimental forecast evaluation

Computer virtual simulation experiment teaching support platform

Incubator

Online support PC end MOOC support Mobile terminal Offline support Virtual device

Experimental project library

Dynamic database Creative material library

Innovation Center

Fig. 6. Architecture design of online virtual experiment teaching platform

Start Select course Increase teaching plan Fill in basic information Related courseware

Completed?

no

yes End

Fig. 7. Creating an experimental course plan process

through the module and send it to the experimental students. The students can watch the demonstration and listen to the lecture anytime and anywhere just like in the classroom. The function of the instrument presentation module is similar to that of the instrument

290

J. Zhang et al.

introduction module in the experimental preview. Experimental help is a functional module that can prompt and explain in the process of students doing experiments. The process is also relatively simple. After selecting a course, teachers can click Add teaching plan, fill in basic information, and choose whether to associate courseware. Each teaching plan is equivalent to a class. After adding all classes, the process ends, so as to achieve the research and design objectives of landscape virtual experiment in school teaching.

3 Analysis of Experimental Results Based on the virtual experiment teaching system, four courses of virtual experiment including analysis, digital logic design, Linux operation management and computer network are developed. Each experimental course requires a virtual experimental platform. A virtual experiment platform can support more than two courses, as shown in Table 1. Typical experiments of experimental courses are continuously added by teachers in teaching practice. Table 1. Experimental courses and required virtual experimental platform Virtual experiment platform

Number of equipment

Experimental course

Circuit analysis

22 species

Enterprises run 12 simulation experiments

Digital circuit

26 species

Enterprise operation law analysis experiment

Computer network

40 species in four categories

Fast acquisition 25 experiment of enterprise business data Enterprise financial data integration analysis experiment

Number of typical experiments

9

12

Design of Virtual Experiment Online Teaching System

291

Online classrooms are mainly divided into online live broadcast and online ondemand. The main test points of online live broadcast are text chat, audio and video chat, document sharing, on-site testing and desktop sharing; the main test point for online on-demand is video on demand, and the test method adopts a role-based method. Test the corresponding function. The main test examples are shown in Table 2. Table 2. Online live broadcast module test case table Input

Expected results

Result

Upload handouts

Upload succeeded

Meet expectations

Teacher turns on mic sound

Mic sound is turned on

Meet expectations

Adjust mic volume

Varies with size

Meet expectations

Turn on the headset

The headset is turned on

Meet expectations

Adjust headphone sound size

Headphone sound changes with adjustment

Meet expectations

Select turn on camera

Teachers can collect camera information

Meet expectations

Turn off the camera

Show teacher’s default Avatar

Meet expectations

Click Share desktop

Teacher desktop is shared

Meet expectations

Use tools to write and draw

The operation is successful and can be displayed correctly

Meet expectations

A student raised his hand and the teacher clicked on it

Students can speak

Meet expectations

A student raised his hand and the teacher clicked reject

Students are not allowed to speak

Meet expectations

Enter text in the discussion area

Students can view the teacher’s speech

Meet expectations

Click to view online students

Show all online students

Meet expectations

Select upload document in the data area

Can be uploaded successfully

Meet expectations

The courseware management module is mainly a module that teachers and administrators can operate. The main test function points are to create and edit courses, create course plans, manage courseware, and assign and correct homework. Specific test examples are shown in Table 3. The test results are shown in Table 4. In order to better test the performance of the system under high load, this system uses the Loader runner tool to create multiple groups of different numbers of users to perform simultaneous operations and record the system’s response time. The items tested in this test are: login to the system, online communication, and online teaching. The test results are shown in Table 5.

292

J. Zhang et al. Table 3. Test case list of course management module

Input

Expected results

Result

Create a new course and enter the course related information

Successfully created new course

Meet expectations

Click publish after successful creation

Published successfully

Meet expectations

New courseware, input courseware information

Courseware creation succeeded

Meet expectations (continued)

Table 3. (continued) Input

Expected results

Result

Online classroom selection recording and broadcasting

Generate recording and broadcasting courseware

Meet expectations

Click course management

Get course list

Meet expectations

Click courseware management

Get courseware list

Meet expectations

Click Create course plan and submit

The course plan can be seen at the front desk

Meet expectations

Click to assign homework and set questions

Students can see the assignment information

Meet expectations

Click to correct the homework

See the homework handed in and input the correction comments

Meet expectations

Table 4. The performance of 200 users online at the same time Test items

Pre test data

Post test data

IO occupancy

0%

10%

Memory usage

100 M

400M

Packet increase

1265

Test average feedback time

3.5

Test maximum feedback time

4.7

After reading the relevant data and studying the relevant technologies of the system design and implementation, analyze the functional requirements and non functional requirements of the online teaching system, design the system based on the system requirements, and design the online teaching function, classroom management function, online discussion function, online examination function and system management function of the system. Then, through the key code design, the main functions of the

Design of Virtual Experiment Online Teaching System

293

Table 5. Online teaching effect test Online education test case Prerequisite

Normal login system

Test target

Understand the performance of the system under multi-user simultaneous online teaching

Method

Use the LoadRunner tool to simulate the multi-user online teaching scenario and execute the test

Number of concurrent tests

Average time for business completion (s)

Maximum time for business completion (s)

Average use of network packets

40

1.988

3.652

75

70

5.255

8.985

79

220

6.871

13.652

130

system are realized, and the main function points are tested to test whether the system can operate normally in daily teaching, optimize the progress of system performance, and further improve the problems existing in the system in the future. The online teaching system can edit the courses involved in teaching work Course sharing and student management are combined to realize the effective coordination and management of the teaching process. Through such a system, the distance of online teaching system can be solved. On this basis, the virtual experimental online teaching system of university economics and management based on virtualization technology designed in this paper is taken as the experimental group, and the traditional virtual experimental online teaching system based on cloud model is taken as the comparison group. The response time of different systems is verified with the instruction response time as the indicator. The results are shown in Fig. 8. Analysis of the results shown in Fig. 8 shows that with the increase of the number of experiments, the command response time of different systems also changes constantly. However, in the whole experiment cycle, the response time of the system in this paper is always lower than 600 ms, indicating that the response time of the system in this paper is higher than that of the traditional system.

4 Conclusion The remote experiment teaching system based on network is a new experiment teaching environment combining computer technology, network technology and virtual instrument technology. This paper designs a virtual experiment online teaching system based on virtualization technology. According to the experimental verification results, the system not only enhances the authenticity of experimental operation, realizes the online sharing of experimental hardware and test data, but also greatly alleviates the contradiction of experimental

294

J. Zhang et al.

Instruction response time/ms

1000 900 800 700 600 500 400 300 200 100 0

1

2

3

5 4 6 Number of experiment

System of this paper

7

8

Traditional system

Fig. 8. Comparison of command response time in different systems

room, instrument and equipment, experimental class arrangement, and has high response timeliness. It can be considered that the system provides a new idea and method for enriching experimental means, improving students’ experimental ability and stimulating students’ experimental interest, which has a wide application prospect. Fund Project. Youth Fund Project of National Natural Science Foundation of China: Research on Key Techniques of Microscope Image Depth of Field Expansion (62002268).

References 1. Ying, Z., Wenhui, H., Qiannan, W.: Design of virtual reliability experiment teaching system. Res. Explor. Lab. 41(07), 138–141 (2022) 2. Hongxv, P.: Research on the application of server virtualization technology in the economic management experimental center of private universities. Comput. Knowl. Technol. 18(15), 112–113+122 (2022) 3. Yanhu, B., Jianli, L., Yuanxiang, L., et al.: Knowledge mapping analysis of economics and management specialty experiments in colleges and universities. Res. Explor. Lab. 40(08), 263–269 (2021) 4. Lingxin, K., Yajun, M.: Big data adaptive migration and fusion simulation based on fuzzy matrix. Comput. Simul. 37(3), 389–392 (2020) 5. Scherer, R., Howard, S.K., Tondeur, J., et al.: Profiling teachers’ readiness for online teaching and learning in higher education: who’s ready? Comput. Hum. Behav. 118(3), 106–115 (2020) 6. Costa, L., Nascimento, T.P., Goncalves, L.: Online learning and teaching of emergent behaviors in multi-robot teams. IEEE Access 18(7), 53–61 (2019) 7. Cobos, R., Jurado, F., Blazquez-Herranz, A.: A content analysis system that supports sentiment analysis for subjectivity and polarity detection in online courses. Revista Iberoamericana de Tecnologias del Aprendizaje 56(22), 1–17 (2019) 8. Danaher, M., Schoepp, K., Kranov, A.A.: Teaching and measuring the professional skills of information technology students using a learning oriented assessment task. Int. J. Eng. Educ. 35(3), 795–805 (2019)

Design of Virtual Experiment Online Teaching System

295

9. Jing, L.A., Cq, B., Yz, A.: Online teaching in universities during the Covid-19 epidemic: a study of the situation, effectiveness and countermeasures. Procedia Comput. Sci. 187(12), 566–573 (2021) 10. Eberle, J., Hobrecht, J.: The lonely struggle with autonomy: a case study of first-year university students’ experiences during emergency online teaching. Comput. Hum. Behav. 121(3), 106– 114 (2021)

Design of Online Teaching System Based on Clustering Algorithm Xiaobo Xue(B) and Lan Zhang Shanghai Institute of Visual Arts (SIVA), Shanghai 201620, China [email protected]

Abstract. According to the poor online teaching effect, this paper proposes the online teaching system based on clustering algorithm, optimizes the structure configuration in the system hardware, improves the system software module function, screen and evaluate the teaching information combined with the clustering algorithm, and optimizes the teaching management steps. Finally, the design system has highly practical application and can improve the online teaching effect of garment rendering drawing course. Keywords: Clustering algorithm · Clothing drawing · Online teaching · Information screening

1 Introduction At present, with the extensive development of computer technology, it is playing an increasingly important role in the field of education. In this case, the continuous deepening and development of online teaching mode has given birth to many online teaching systems, which provide convenience for students’ online learning. For example, zhoujunping designed a computer-aided classroom teaching system based on data mining to solve the problems of poor information utilization, low intelligence and personalization in the existing computer-aided teaching mode. The system module design adopts threelayer b/s structure. Using information collection module to collect and store student information data; The user information preprocessing module preprocesses the information data as the data of the personalized data source in the personalized data analysis module. Through the personalized data analysis model, the data mining parallelization technology is adopted, the data clustering center is updated through the K-means clustering algorithm and the map reduce parallel computing framework, the clustering performance index is minimized, the data in the personalized data source is effectively mined and analyzed, and the analyzed conclusions are regularized to generate teaching rules, which are presented to students through human–computer interaction, so as to improve the intelligence and personalization of the teaching system [1]; Gaoweiwei designed a computer-based Chinese network assisted instruction system. Firstly, the system process, overall framework and network structure are comprehensively analyzed. Secondly, the system function modules are explored in detail based on the three users of © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 296–310, 2022. https://doi.org/10.1007/978-3-031-21161-4_23

Design of Online Teaching System Based on Clustering Algorithm

297

students, teachers and administrators, and the system design is completed [2]; Zhangliqun designed an online auxiliary teaching system for music courses based on ZigBee. The hardware of the system includes a transceiver, a sound quality sensor and a resource manager. A DIN rail type online auxiliary teaching area is added to the transceiver to improve the integration between the transceiver and other devices; The default output current of the sensor is 15 mA, and the working voltage is dc12–24 V ± 10%. The sensor is connected to other hardware through the two-wire mode. The memory of the resource manager is 256 µg, and the CC2430 chip is added to improve the sensitivity of the system. Through logical storage, information management, shared processing and online puzzle solving, the software flow is realized, so as to complete the design of the teaching system [3]. With the development of campus informatization and “Internet+”, universities have gradually transformed from digital campus to “smart campus”.Smart campus is a new stage of information development of colleges and universities, which reflects the deep integration of technology and business, and the high integration of resources and application [4]. With the help of a new generation of information technology, the class can fully perceive the work, study, life and other scene characteristics of campus groups, so as to establish an intelligent teaching environment and a comfortable living environment. At the same time, the online teaching system of the clothing renderings mapping course should unify the information portal and identity authentication, integrate the data scattered in various departments, and provide cross-departmental comprehensive information services such as personnel, teaching, students, scientific research, assets and equipment and finance. After the staff pass the identity certification, they will have the corresponding authority, and they can carry out teaching and scientific research management, administrative office and service information inquiry [5]. In addition, Smart Campus has realized the goal of integrating digital teaching resources, created an active, collaborative and exploratory digital learning environment, established a new teaching mode of teacher-student interaction, and can better serve the teaching of clothing renderings drawing courses. Based on this paper, this paper proposes an online teaching system design method based on the clustering algorithm. In order to better carry out online teaching of clothing rendering and drawing course, this paper introduces clustering algorithm into this field, and designs an online teaching system based on clustering algorithm. By optimizing the system hardware configuration, improving the system software module function, combining the clustering algorithm to screen and evaluate the teaching information, optimize the teaching management steps, and complete the design of the whole system.

2 Clothing Rendering Drawing Course Online Teaching System 2.1 Hardware Configuration of the Course Online Teaching System The system is mainly designed to realize online teaching, and the system uses DOT NET and Web Service to provide an application interface for system integration and data integration. The front-end implementation of the system requires a generous and beautiful interface, and a friendly and convenient user operation. The main objectives of the system design are the following: to meet the functional needs of the online teaching

298

X. Xue and L. Zhang

system [6]. Students can easily enter the front-end of the system and conduct information browsing, online examination, online communication, job submission and other functions; the system design functions should be simple, easy to use and practical. Excessive complex system will bring users into mistakes. The goal of the system design is to adopt the current more common online teaching process to achieve, so that the system is easy to use, simple and powerful [7]. The system design should pay attention to the principle of system scalability and reusability; the system design functions need simple operation, beautiful interface and good user experience. Design the system network deployment, as shown in Fig. 1.

Web server 1 Web server 2

Production Test database database server server

firewall

Distributed cache database

SQA interface server

Extranet

Fig. 1. System network deployment diagram

Based on B/S network online teaching system throughout the design process of independent learning ability training strategy, make full use of the existing network foundation to create network-based teaching platform, teachers do not need to consider the technical details of the system, will put more energy into the construction and integration of teaching content and resources, focus on cultivating students’ independent learning ability [8]. Students can access various teaching resources through the teaching platform, and can cluster through message boards, chat rooms, etc. Based on the above considerations, the B/S mode is adopted to design the network online teaching system, the background server adopts JSP components and SQL Server2005 database system, and the front desk client adopts the browser to realize it. The physical architecture of the B/S-based system is shown in Fig. 2:

Design of Online Teaching System Based on Clustering Algorithm Test database server

Production database server

299

Distributed cache database

client client

client

Switch

Router

The server

firewall

Internet

Fig. 2. Physical topology diagram of the network online teaching system based on B/S

The B/S-based network online teaching system uses the Windows 2008 Server development platform, developing tools and environments for the Microsoft ASP.NET3.5 and Visual Studio 2008. proscenium ASP. The NET network development language can establish a powerful WEB application service programming framework, using SQL Server2005Standard Edition at the background database end, and Standard edition is a data service management and analysis platform suitable for small and medium-sized enterprises. The system adopts the B/S three-layer service framework mode. 2.2 Functional Structure Optimization of the System Software The system is based on the Web software system of Internet network, using B/S development mode. The online teaching system includes four sub-systems: teaching management subsystem, multimedia teaching subsystem, student management subsystem, and teacher management subsystem. The functional structure of the system software is shown in Fig. 3:

300

X. Xue and L. Zhang Teaching management subsyst em Profess ional managem ent Teaching courseware Teaching statisti cs Teaching video

Teaching plan management

Ps eudoreal Laboratory

Degree Englis h

Classroom live broadcast Textbook management

Mult imedia t eaching system

Teaching evaluation

Online teaching system

Micro class on demand Teaching discussion

course management Net work broadcast ing

Experim ental teaching management

Cl assroom recording and broadcasting

Achi evement management

Student s tat us management

Exam inat ion management

Home School Interaction

Interacti ve Q & A

quality eval uati on St udent m anagem ent subsystem

Scholars hip Management psychological counseling

personal information

St udent aid

Scientific research management

Teacher management subsyst em

Eval uati on management

Traini ng managem ent

Charge management

St udent attendance management

Graduate m anagement

job management

Fig. 3. Functional structure optimization of the system software

As shown in Fig. 3, the system design part of this paper is mainly divided into four systems, including teaching management subsystem, teacher management subsystem, student management subsystem and multimedia teaching system. Each subsystem has different functions and constitutes a complete teaching system. The system adopts B/S mode, and the server is divided into three layers, namely Web service layer, middle layer and data service layer [9]. At the Web service layer, Is and ASP are generated in response Online learning module

Background management module Communication between teachers and students Leave a message and ask questions

Student user Homework examination

Student information management View learning progress

Online shopping system entrance

Register login

Video data browsing Teacher background management

Teaching resource management

Homework examination management Information release management

System settings

Fig. 4. Division of the main functional modules of the system

Design of Online Teaching System Based on Clustering Algorithm

301

to the request of users. The main functional modules include student online learning and teacher background management, and the division of the main functional modules of the system is shown in Fig. 4. Teaching resources are the basis of online learning. Teachers can realize the release and sharing of resources through the system. The richer the resources, the greater the choice of students in online learning. Based on this, students can further optimize their collaborative operation, and the steps are as shown in Fig. 5. monitor Student B

Student B

Example gate

Student A

Student C

Student D monitor

Fig. 5. System operation procedure optimization

Resource management includes resource upload, download, comment and resource playback functions, where teachers have resource upload function, and students have resource download, resource review and resource playback functions. To design a learning system that combines the CDIO and the Internet teaching platform, it is necessary to integrate the CDIO engineering education mode into the network teaching platform. The system business process combining CDIO is shown in Fig. 6. Teacher login

design scheme

Develop curriculum

Develop project

Video resources

Assign classes

Assign account number

Determine the scheme

Tailoring course

Set task

Corresponding courses

Develop transcripts

Course transcript

Identify project

Project transcript

Fig. 6. Teacher-end business processing process

Based on the above business process, the requirements of the system are analyzed. The overall requirement analysis results of this system are shown in the table and are divided into multiple functional modules. The specific system design requirements are shown in Table 1:

302

X. Xue and L. Zhang Table 1. The requirements analysis table

Serial number

Modular

Function

A

Account management

Distinguish between ordinary users and administrators. Administrators have more functions than ordinary users

B

Personal information management

Maintenance of user personal information

C

Programme Management

Teaching program management

D

Course management

Managing courses according to different technical directions is the basis for the division of teaching resources

E

Training project management

Integrate CDIO teaching mode, and extract technical points from practical training projects as key teaching contents

F

Teaching resource management

Including the overall management of teaching resources such as teaching materials, videos and lecturers

G

Teaching management

Teachers and users should first create classes

H

Class management

Administrators can create classes

I

System management

System status reading and setting, data backup, etc

When uploading the resources, you need to submit the information about the resource classification, resource name and detailed description of the resources to facilitate the students to query the resources. Once the resource is uploaded, students are able to browse and download it locally for subsequent learning. Teachers are responsible for the increase, modification and deletion of teaching resources to the courses managed, such as adding a new syllabus, new test questions, homework and examination subjects, modifying the existing test questions, homework, and must be removed from the database when the resources are outdated and inconsistent with the course or there are errors in the resources. 2.3 Implementation of Online Teaching of Garment Drawing Course In designing a series of functions proposed by the online teaching system of clothing drawing courses, it is to choose the development platform and development framework selected by the system, and to adopt a series of mature system components to accelerate the development efficiency. After completing the system framework construction and component selection mentioned above, the next step is to conduct the coding work, where the public module to the system needs to be designed first. The formula of the

Design of Online Teaching System Based on Clustering Algorithm

303

clustering space definition of the clustering algorithm optimization algorithm is: l = (a1 , a2 , ..., ai ) − d (x + y)/k − m

(1)

In this case, the parameters k and m in the algorithm will be fixed and will no longer be constrained by the parameter space of the problem. The sequence of parameter ai is transformed in the space of (x, y), and the range of sequence d is unlimited. The parameter space e of the practical problem is in the range of sequence space (nd , md ), and the sequence s is unchanged. The clustering space and parameter space can be obtained by transformation and inverse transformation (2). g=

(nd + md ) − dai /e − s 100l − x − y

(2)

After information conversion based on the above algorithm, teaching information management is realized in the database, and a basic platform is further formed to form a unified access entrance and a unified information display platform. The software of this system is the operating system software, database system software installed on each server, etc. The specific requirements are shown in Table 2: Table 2. System software setup requirements Software name

Configuration description

Quantity

Database

SQL SERVER2017

1

Middleware

Tomcat7.5

2

Report software

MicroStrategy

2

The online teaching system adopts the general design idea of 3 + N, that is, the data access layer business logic layer and user interface layer, on each layer according to the complexity of specific functions, the plan molecular layer. Account management is the main function of the administrator, and the administrator can add and delete accounts through this function. The specific use-case descriptions of this function are shown in Table 3. Table 3. Add a user account use case description table Use case description

Content

Business participants

Administrators

Describe

This use case verifies whether the administrator can add new users to the system and distinguish student and teacher accounts (continued)

304

X. Xue and L. Zhang Table 3. (continued)

Use case description

Content

Preconditions

Nothing

Trigger condition

After logging in the administrator account, click “user account management”

Basic process

Add a new account and set the initial password; Log in with a new account

Abnormal process

Add an existing account and promote “account already exists”

End

Login succeeded, jump to personal information page

The online teaching system of clothing drawing course is based on the Android operating system and is divided into blocks according to the functions. It is mainly divided into three modules: teaching application module, cluster control module and network communication module. The cluster control module and the network communication module both provide services for the teaching application module to support the user clustering and collaboration function in the teaching. The core of the design of the system is the choice of the knowledge base, a regular collection that can dig out different effective data and can use the data entropy-based data mining algorithm. In addition, the teaching application module is presented above the Mainactivity in the form of a custom Vew control, and directly clusters with the user. The system structure is shown in Fig. 7.

User interface mainactivity

Teaching application module

MST Player

Interactive selftaught

Interactive control module

Data transmission and synchronization

User state control

Collaboration group management

Network communication module

Communication control of TCP and XML

Message Dispatch

Message registration

Fig. 7. Optimization of the system teaching structure

The online teaching system uses training programmes as the basis for all course design. After the completion of the classes, students should first make a training plan, and then determine what courses the class will need to take and do according to the training program. The training scheme management includes the new scheme, modification scheme, deletion scheme and other functions. The specific program flow chart design of the training scheme management is shown in Fig. 8.

Design of Online Teaching System Based on Clustering Algorithm

305

N Y remove scheme

Delete scheme? N Y Replication scheme

Amplitude scheme? N Y N

Modification scheme

Modify scheme?

start

Select the course management tab in the management center page

Teacher login

Y

The setting scheme includes courses and projects

New scheme?

end

Fig. 8. Functional program flow chart of teaching training program

The course management function is developed for teachers and is one of the basic functions of teaching resource management. The premise of using this function is to complete the training program setting, curriculum management is the process of training program refinement, the purpose is to implement the knowledge points in the training program, and the engineering education mode in the specific curriculum. If the training plan is not set, the course list is not displayed in the course management list; if the training plan is set, then teacher users can continue to conduct course management functions such as syllabus management, teaching task management and teaching teacher management. The flow chart of the procedures for garment drawing course management is shown in Fig. 9:

New teaching task

New syllabus

N

start

Teacher login

Select the course management tab in the management center page

N

N

Y Is there a syllabus?

Are there teaching tasks?

Add instructor

Y

Y Are there teachers?

end

Fig. 9. Flow chart of the course management program

The main function of the teaching application module is to provide a teaching clustering environment for teachers and students to facilitate the development of teaching activities To achieve the above goals, the teaching application modules need to provide a variety of different types of teaching application tools to achieve multidimensional clustering with users. In order to make each teaching application independent and to be displayed in different regions, the view of the teaching application should be separated

306

X. Xue and L. Zhang

from its specific functions. Therefore, the teaching application tool framework is shown in Fig. 10: Operation event delivery

Teaching application tools

controller

Action event capture Event listener

View control view

View update

Fig. 10. The Teaching Application Tool Framework

The teaching application tools are all designed based on the view control view provided by the androk application framework. The view control is the basic control provided for the user interface by the Android system, which represents a rectangular area on the display screen and is responsible for image drawing and event processing in that area.

3 Analysis of the Experimental Results Functional test is mainly demand verification, which is to ensure that the final system implementation is consistent with the initial demand analysis, can meet the needs of the user, the correct performance test through pressure test, the purpose is to verify whether the operation of the system, whether the normal test environment is shown in Table 4. Table 4. List of test environments Serial number

Name

Hardware environment

Software environment

A

The server

12 GB memory

MySQL Server 5.8

B

Client

Six 160 gsata hard drives

Windows7

C

Network environment

Wan 100 Mbps

-

The installation operating system is Windows XP Professional, the background server is S QL Server, it uses the IS service, and the computer is minimally configured with 512 MB of CPU1.0G memory. For software performance tests in specific operating environment and load conditions, this system mainly focuses on the impact of the simultaneous number of online people on the response time, as shown in Table 5.

Design of Online Teaching System Based on Clustering Algorithm

307

Table 5. Performance test cases Test item

Number of people online at the same time

Expected average response time

Actual average response time

a

80

Less than 2S

0.465

b

150

Less than 2S

0.685

c

260

Less than 3S

1.325

d

350

Less than 3S

1.658

e

1200

Less than 3S

1.989

Performance indicators mainly include response time, throughput, and so on. Performance testing is to test whether the software system can meet the previously expected performance requirements. Performance is an important part of the system, and the corresponding performance indicators have been set in the demand stage, which are the focus of the performance test stage. Performance testing is designed to test the performance of clustering online teaching systems during mobile and teacher side clustering. The timely response to clustering data has a great impact on the clustering experience and requires focus. The test is divided into unoptimized situation and optimized, mainly to test the time delay of clustering data between teachers and students’ equipment. Considering the frequency of clustering, the test mainly adopts the cluster whiteboard tool as the test target. Compared with the courseware play synchronization, the number of cluster whiteboard synchronization data is larger, which can more explain the actual situation. First, first consider the cluster data transfer round time between the teacher and the student side without closing the Nagk algorithm, as shown in Fig. 11 and Fig. 12.

Optimized round trip delay (MS)

300

240

180

120

60

0 1

25

49

73

97

121

145

169

193

experiment

Fig. 11. Unoptimized system delay case

If the test results do not meet the set goals, the online teaching system can not be applied to the school to handle the relevant business. This can not only improve the

308

X. Xue and L. Zhang

Optimized round trip delay (MS)

100

80

60

40

20

0 1

25

49

73

97

121

145

169

193

experiment

Fig. 12. The optimized system delay situation

efficiency of teaching management, but will reduce the efficiency. Therefore, the performance test must be conducted before the deployment, and the results of the performance tests can be matched to deploy the online teaching system to the school. LoadRunner software is mainly used to conduct the simulation test, according to the performance requirements, the concurrent access to 500 users, and the test results are shown in Table 6: Table 6. System performance test results table Number of concurrent users

Response speed (seconds)

CPU utilization (%)

60

1.25

1

120

1.35

1

180

1.45

1

240

1.55

4

300

1.59

8

360

1.78

12

420

1.98

15

480

2.21

20

As described above, the system has completed the functional tests and performance tests. There are three rounds of system tests to verify the system stability. The specific test results are shown in Table 7. The test results from three rounds of system tests can show that the test defects showed an obvious convergence trend, and the V e r version 1.0 test passed 100%.Therefore, it can be proved that the design system can meet the functional and non-functional needs of the user, and can meet the system design principles. The reason for this good result is

Design of Online Teaching System Based on Clustering Algorithm

309

Table 7. Results of three rounds of system testing Test rounds

Software version

Number of test cases

Passing rate

First round

Ver0.6

1152

86%

Second round

Ver1.0

420

85%

Third round

Ver1.2

190

100%

that this design system optimizes the system hardware structure configuration, improves the system software module function, combines the clustering algorithm to screen and evaluate the teaching information, and optimizes the teaching management steps, which improves the functional effect of the system through all-round design.

4 Conclusion In order to improve the effect of online teaching, this paper introduces clustering algorithm into this field and designs an online teaching system based on clustering algorithm. By optimizing the system hardware configuration, improving the system software module function, combining the clustering algorithm to screen and evaluate the teaching information, optimize the teaching management steps, and complete the design of the whole system. The system realizes the main function of online teaching, and has good operation efficiency, but also easy to maintain and transplant, through the combination of big data technology, the system can collect data cannot obtain or high cost, such as course name, homework, interactive information, learning time, test performance analysis, student login system, students ‘learning preferences and other behavior data, and is easily used for teaching process analysis, in order to fully improve students’ learning experience, comprehensively improve the teaching effect. Although the system designed in this paper has the above advantages, it has not been compared with other systems in this study, so it is difficult to know the direction of improvement. In the next research process, it is necessary to strengthen the research in this regard to further improve the rationality of the design system.

References 1. Zhou, J.: Design of computer aided classroom teaching system based on data mining. Mod. Electron. Technol. 43(2), 84–86 (2020) 2. Gao, W.: Design of computer-based Chinese network assisted instruction system. Autom. Technol. Appl. 7, 166–169 (2020) 3. Zhang, L.: Design of Zigbee based music course online assistant teaching system. Microcomput. Appl. 38(4), 171–174 (2022) 4. Lin, P.H., Chen, S.Y.: Design and evaluation of a deep learning recommendation based augmented reality system for teaching programming and computational thinking. IEEE Access, (99), 1–1 (2020) 5. Jian, Q.: Multimedia teaching quality evaluation system in colleges based on genetic algorithm and social computing approach. IEEE Access 7(4), 1–1 (2019)

310

X. Xue and L. Zhang

6. Wójcik, K., Piekarczyk, M.: Machine learning methodology in a system applying the adaptive strategy for teaching human motions. Sensors 20(1), 314 (2020) 7. Yao, S., Li, D., Yohannes, A., et al.: Exploration for network distance teaching and resource sharing system for higher education in epidemic situation of COVID-19 - ScienceDirect. Procedia Comput. Sci. 183, 807–813 (2021). 8. Nancy, W., Parimala, A., Livingston, L.: Advanced teaching pedagogy as innovative approach in modern education system. Procedia Comput. Sci. 172(3), 382–388 (2020) 9. Li, X., Ying, Y., Zeng, Y.: Simulation of large data multi-resolution acquisition method based on Java3D network .Comput. Simul. 37(2), 416–420 (2020)

Design of Distance Teaching System for Textile Pattern Design Course Based on Online and Offline Integration Lan Zhang(B) and Xiaobo Xue Shanghai Institute of Visual Arts (SIVA), Shanghai 201620, China [email protected]

Abstract. In order to reduce the CPU utilization of the distance teaching system of textile pattern design course, this paper designs a distance teaching system of textile pattern design course integrating online and offline. The hardware design part includes power supply module, wireless communication module, logic control module and data reading module. The software design adopts B/S architecture to establish a three-tier structure model. Microsoft SQL server is adopted as the database management system; Use Hadoop framework to build cloud computing platform; Pick KX algorithm is used to realize dynamic load balancing and allocate server resources; Correct the difference of data name by entity alignment; Through ASP Net script file to design web interactive page. The test results show that when the number of concurrent users is 200 and 500, the maximum CPU utilization of database server is 53.7% and 65.5%, and the maximum CPU utilization of web server is 47.5% and 56.7%, which reduces the CPU utilization of the system as a whole. Keywords: Online and offline integration · Textiles · Pattern design · Distance learning system

1 Introduction Education is not only the basis of national competitiveness, but also an important embodiment of comprehensive national strength. In China, the development and reform in the field of education are under the heavy pressure of great challenges. The combination of education and big data technology has become an inevitable trend in the development of education. In the whole process of educational activities, all data related to educational activities and the collection of data used to create potential value for educational development are collectively referred to as educational big data. Under the current research situation in the field of education at home and abroad, from a strategic perspective, education data can be positioned as a scientific help to promote the transformation of new strategic assets in education, drive comprehensive reform in the field of education and the basis for the development of intelligent education. With Web 2 With the rise of 0, the scale of data on the Internet has increased sharply. For educational resources, thanks to © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 311–323, 2022. https://doi.org/10.1007/978-3-031-21161-4_24

312

L. Zhang and X. Xue

the vigorous development of educational informatization, a large number of educational resources with rich content and various forms have been accumulated on the network. With the increase in online teaching demand and users, the amount of curriculum resources for online distance teaching is increasing explosively [1]. Textile pattern design course is the main discipline of design major. In teaching, students express the things inspired in their own minds by relying on the existing basic pattern elements [2]. Combining the textile pattern design course with the online and offline integrated distance teaching system opens up a new idea for the teaching of art specialty, breaks the thinking imprisonment of traditional pattern design, improves the utilization of high-quality teaching resources and promotes the process of informatization of textile pattern design education. Therefore, the academic circles pay more and more attention to the distance teaching system of textile pattern design course. In the traditional method, Li Peiyun et al. [3] designed a distance multimedia teaching system based on virtual reality. In Hardware design, the system uses 3D scanner to model in the way of 3D stereo detection; In Software design, the small plane is used to simulate the physical surface, and the module size in the virtual teaching scene is adjusted with reference to the error function; Adjust the module pixels according to the pixel resolution formula to build a virtual scene; Simulate the equipment function through VRML language program; The distance teaching system completes the triggering of teaching program, data transmission and layout conversion according to the response function. Zhang Ling [4] proposed the design of distance teaching system based on artificial intelligence network. In Hardware design, the student learning module is composed of teaching cooperation agents and multiple other agents, which are responsible for the presentation of teaching materials and the solution of teaching problems. In Software design, Knowledge sharing is realized through cooperation mechanism to provide personalized teaching basis for the system. The teacher teaching module mainly provides students with corresponding teaching strategies according to the curriculum requirements and provides intelligent guidance for the problems encountered in the teaching process. The evaluation module uses the evaluation rules to comprehensively evaluate students’ learning behavior, learning attitude, learning effect and learning ability. Man machine interface is the communication medium between students, teachers and the system. Students and teachers complete personal login through the login interface. The research on the above distance learning system can innovate the teaching mode and provide power for students to change their learning methods, but it does not fully consider the characteristics of node resources in the system cluster, resulting in unreasonable differentiated allocation scheme of system node resources and high CPU utilization. In order to reduce the CPU utilization of the distance teaching system, taking the textile pattern design course as the research object, this paper designs a distance teaching system of textile pattern design course integrating online and offline. The system can provide teaching and learning services anytime and anywhere when the network environment allows, and improve the teaching efficiency of textile pattern design course to a certain extent. The hardware design of the system includes four modules: power supply module, wireless communication module, logic control module and data reading module. The software design part adopts Microsoft SQL server as the database management system; Build a data cloud computing platform using Hadoop framework; Innovatively

Design of Distance Teaching System for Textile Pattern

313

pick KX algorithm is used to realize dynamic load balancing and reasonably allocate system server resources; Using ASP Net scripting system. The experimental results show that the system designed in this paper can significantly reduce the CPU utilization and has high application value.

2 Hardware Design of Distance Teaching System for Textile Pattern Design Course with Online and Offline Integration The overall structure design of the software and hardware of the distance teaching system is shown in Fig. 1:

Using pick KX algorithm to realize dynamic load balancing and server resource allocation

Web interactive design

Power module

Wireless Network communication module server

Logic control module Network server

Integrated database Data reading module

Metabase

Fig. 1. Overall software and hardware design framework of the system

According to the Fig. 1 and design requirements of the remote teaching system of online and offline integrated textile pattern design courses, the hardware structure framework of the system is established as shown in Fig. 2.

314

L. Zhang and X. Xue

Power module

Wireless communication module

Hardware design of distance teaching system for textile pattern design course

Logic control module

Data reading module

Fig. 2. System hardware structure framework

As shown in Fig. 2, the hardware design section of the system includes four modules: a power supply membrane block, a wireless communication module, a logic control module, and a data reading module. The logic control module consists of FIFO queue and CSR (control and status logic) response queue. Before adding the msgdma IP core to the qsys system, the FIFO queue depth needs to be initialized, and its maximum depth can be set to 256. The PC side only initiates DMA read and write tasks for the cyclic buffer addresses with four consecutive physical addresses. Considering the alignment of pages (the page size is set by PCIe hard core parameters), the maximum number of memory descriptors used to complete a DMA read-write operation is 8. The logic control module needs to save the read-write address of each memory block to the read-write memory descriptor FIFO queue in the module. In the process of starting DMA transmission, read and write memory blocks in turn according to the address in FIFO queue until the whole queue is empty, and then complete this DMA operation process, so as to realize sgdma function. The wireless communication module adopts serial communication, which provides two groups of TTL interfaces. One group is a 2.85v-ttl interface with level matching, which can be directly connected with 3.3V single chip microcomputer; The other group uses pin VCC_ MCU is a 5v-ttl interface with level matching, and can be compatible with single chip microcomputer with 5 V, 3.3 V and other voltages. The power supply voltage of atmage1280 single chip microcomputer is 5 V, and its communication with sim900a module adopts 5v-ttl interface. The wiring method is shown in Fig. 3.

Design of Distance Teaching System for Textile Pattern

315

ATmage1280 TXD RXD

RXD TXD

SIM900A R

R

MC U

GND

VCC_GND

GND

VCC

C

Fig. 3. Schematic wiring diagram

Sim900a module contains two indicators, namely power indicator and network indicator. We can roughly judge the working state of the whole module by observing the working state of the two indicators. When the power indicator of sim900a module is on for a long time and the network indicator flashes slowly, the sim900a module is in normal working state, and the normal communication of system hardware can be carried out at this time. The data reading module is responsible for reading data from the register and converting the data into a high-speed Avalon data stream. The data reading module includes four ports: read command, read response, data master and data sink. The power module is the power source of the whole system. The system is powered by 5V voltage. The power supply circuit adopts the step-down switch type integrated voltage stabilizing chip LM2596 produced by Texas Instruments (TI)_ ADJ The chip has a receiver and a transmitter, and the highest transmission rate is 5mbps, which meets the design working environment and requirements. In order to achieve greater output current, several three terminal voltage stabilizing circuits need to be connected in parallel. When n voltage stabilizing circuits are connected in parallel, the output current increases n times. It is required that the chip model of each branch must be consistent with the manufacturer, otherwise the stability of the output current is poor, and even the circuit is burned in chain.

316

L. Zhang and X. Xue

3 Software Design of Remote Teaching System for Textile Pattern Design Course Integrated Online and Offline 3.1 System Structure Framework The software design of this system using the three-layer structure model of B/S architecture standard is shown in Fig. 4.

Presentation layer

Client browser user interface

Register login

Data management Business layer

Course management of textile pattern design Job management

Data layer

Database

Fig. 4. System software structure framework

As shown in Fig. 4, the user terminal puts forward an operation request to the system through the browser operation interface; The server side of the system runs based on the web server in the form of website, and the background links the MySQL database server to fetch data through the SSH framework based on J2EE; The server file system uses Hadoop framework to build a cloud computing platform. The platform supports the storage, viewing and downloading of textile pattern design teaching course resources and homework files, calculates and analyzes the data in the database and the files in the file system, and returns the results to the client browser. The module software code of distance teaching system of textile pattern design course mainly includes the following aspects: first, this kind of code file is mainly used to process distance learning information and make learning records, which can effectively help teachers master students’ learning situation in time, and then automatically save it, so as to provide reference for future teaching. Second, this kind of code can provide students with convenient operation services, simplify the previous cumbersome operation procedures, and meet the basic requirements of actual production. Third, this kind of code in the

Design of Distance Teaching System for Textile Pattern

317

document is mainly for users to help students formulate a scientific and reasonable textile pattern design teaching plan, provide high-quality services for students to find leaks and fill vacancies, and quickly solve difficult knowledge in learning. Fourth, in this kind of code file, it is mainly to conduct a comprehensive and reasonable evaluation of the course and provide targeted guidance for students’ self-learning in the future [5]. Each functional module of the system uses the cloud computing distributed file reading and writing method based on Hadoop framework to ensure the upload and download efficiency of data files and the stable operation of the system under the condition of high concurrency of big data. The distributed fliesystem object remotely obtains the datanode node path of the first few blocks of the file through RPC. The namenode will return the datanode node containing the block copy according to the actual network topology. If the requesting client node itself is a datanode and there is a block copy on its own node, it can be read locally directly. The current allocation method does not fully consider the characteristics of node resources in the cluster to allocate nodes differently. This system design adopts pick KX algorithm to realize dynamic load balancing. Pick KX algorithm is used to realize dynamic load balancing and reasonably allocate server resources. The process can be expressed as follows: ϑb =

Ra a  ha

(1)

a=1

In formula (1), ϑb represents load assignment probability; a, b represents node location serial number; Ra represents load of each node in the cluster; ha represents the current load of the server; The calculation formula of Ra is as follows: Ra =

H − ha H

(2)

In formula (2), H represents the total load. When the probability ϑb is larger, it indicates that the current load of this system is the minimum. At this point, the request submitted by the user can be dynamically allocated to the server for processing. Textile pattern design course management mainly includes a number of contents about learning and course, which realizes the effective storage of information. In the whole module, it mainly operates the corresponding table. In the course management of textile pattern design, students can get the corresponding information through the system, and then bind the course with data. Students can make repeated inquiries in combination with the current situation, so as to improve the effect of actual processing. 3.2 Design the Database Structure Database is the infrastructure for the normal operation of the whole system, which directly affects the success or failure of system design, especially the function of system operation. The system uses Microsoft SQL server as the database management system, and most of the information on the server side is stored as this. If you need to change

318

L. Zhang and X. Xue

the database system, you can achieve the goal by partially changing the configuration file of Spring or the persistence class of Hibernate. The database of the teaching system consists of multiple data tables, including: teacher and student information table, teacher courseware table, test paper management table, practice management table, student transcript, analysis table, question bank table, etc. The object entity model is the carrier of the system for data transmission and persistent storage process. The digital education resource management system includes many object entity models corresponding to users, resources, administrators and other entities [6]. The entity user attributes include user ID, account number, password, permission rule, creation time, etc. User ID is the ID number, which is unique. The account and password are used to log in to the system. The system judges the legitimacy of the user’s permission by reading the permission rules. Entity resource attributes include resource number, resource name, resource type, resource content, resource address, creation time and change time. The resource number is manually set and used for form records. The resource type sets the grouping category of the resource. The resource address saves the location information stored by the resource. The same entity name extracted from different educational resources may be different, resulting in inaccurate results. Therefore, this difference needs to be corrected through entity alignment [7]. Entity alignment can be judged by extracting the similarity between entities. Entities with high similarity may be different representations of the same entity. The formula for calculating the entity similarity can be expressed as: λ=

PQ |P||Q|

(3)

In formula (3), λ represents the entity similarity; PQ represents the inner product of two vectors; |P||Q| represents the product of the length of two vectors. The greater the cosine similarity, the smaller the angle between vectors, and the more similar the entities are, the more likely they are to be different representations of the same entity. The entities with high similarity are screened out by setting the threshold, and whether they belong to the same entity is judged manually. All forms in the system are created, edited and submitted by the user, and then approved by the administrator with high authority. However, each user may participate in a variety of businesses, so each user may produce many submitted forms, so there is a one to many relationship between user entities and form entities. When an administrator approves a saved form, an administrator may approve multiple forms. Therefore, the relationship between the administrator entity class and the form entity class is also a one to many relationship. 3.3 Function Realization of Textile Pattern Design Course System The login system module is mainly responsible for managing the user’s login, registration and deletion functions. Users of the system need to confirm their identity before use. Only the authorized user name and its corresponding correct password can use the functions of the software system and conduct the relevant operation. This section is designed and mainly includes three functions: registration, login and management. Each type of user (administrator, teacher, and student) has different permissions, and different types of users will enter different operating interfaces after logging in.

Design of Distance Teaching System for Textile Pattern

319

The data management function module consists of three sub-modules: data download, data upload and data maintenance. Student users download the data information through the data download interface, so the information download function depends on the data information entity; the teachers upload the data information, so the data information entity; the administrator user includes data maintenance through the data maintenance interface, so the data maintenance function also depends on the data information entity. Textile pattern design curriculum system can organize various types of multimedia teaching resources, can build multimedia database for text, pictures, video and other multimedia, and the text and gallery management separately, set up navigation system, realize the centralized teaching resources management and rapid retrieval and call of teaching resources function. Resources can be released in the form of courseware for teaching demonstration. Due to the special and important position of graphics in textile pattern design courses, the remote teaching system sets a variety of display methods for image resources, which fully meets the needs of teaching demonstration. In order to realize the display and interaction function of the system, through ASP Net script file for Web interactive page design. Each page design should correspond to the corresponding logical processing class structure. Class structure is mainly used to record relevant information and provide corresponding operation interfaces, so as to facilitate students’ learning, course learning, formulate learning plans and improve learning efficiency. The function module of homework interaction management consists of three sub modules: student homework, teacher homework and homework maintenance. The student user operates the student homework through the student homework interface, including homework information, so the student homework function depends on the homework information entity; The teacher user carries out the teacher operation through the teacher operation interface, which also includes the operation information, so the teacher operation function also depends on the operation information entity; The administrator user carries out job maintenance through the job maintenance interface, which also includes job information. Therefore, the job management function also depends on the job information entity.

4 System Test 4.1 System Function Test There will be errors and deficiencies in the process of system development. In order to ensure the complete and normal operation of the distance teaching system, the black box test is used to test the function of the system. Test the function of the system editing module to see whether it is perfect and whether it can run correctly. The test items mainly include teaching plan entry, editing drawings and notes, multimedia resource integration and teaching resource release. After testing, all editing items can be realized normally. Test the function of the system teaching module to see whether it is perfect and whether it can run correctly. The test items mainly include courseware import, courseware playback, view switching and related data. After testing, the teaching items of textile pattern design can be realized normally. The data set used in this paper system

320

L. Zhang and X. Xue

comes from the teaching course resources and homework file data of the textile pattern design major in a certain university. 4.2 System Performance Test This experiment mainly tests the CPU utilization rate of the design system. Test the maximum database server CPU usage and the maximum Web server CPU usage, with concurrent users of 200 and 500, respectively. The performance test results of the designed remote teaching system are compared with the remote teaching system based on virtual reality and AI network. The results of the database server CPU maximum utilization comparison are shown in Tables 1 and 2. Table 1. Maximum CPU usage of database servers with 200 concurrent users The number of experiments

Maximum CPU usage of database server (%) The remote teaching system of the textile pattern design course is designed

Distance teaching system based on virtual reality

Distance teaching system of textile design based on AI network

1

53.4

59.4

60.5

2

53.8

58.7

61.4

3

52.5

57.8

59.8

4

54.6

58.6

60.5

5

55.2

59.7

62.6

6

53.5

58.2

61.3

7

52.3

57.3

59.2

8

54.2

58.5

60.5

9

54.0

59.2

60.4

10

53.4

58.1

59.0

Design of Distance Teaching System for Textile Pattern

321

Table 2. Maximum CPU usage of database servers with 500 concurrent users The number of experiments

Maximum CPU usage of database server (%) The remote teaching system of the textile pattern design course is designed

Distance teaching system based on virtual reality

Distance teaching system of textile design based on AI network

1

64.4

68.4

70.8

2

65.8

69.7

70.4

3

66.6

70.5

70.7

4

64.5

70.6

69.8

5

66.9

70.9

68.5

6

64.5

66.8

69.6

7

66.2

69.5

69.5

8

65.6

70.4

70.8

9

65.3

70.1

70.9

10

65.2

70.2

69.0

According to the results of Table 1, when the number of concurrent users was 200, the maximum database server CPU utilization rate of the remote teaching system of the designed textile pattern design course was 53.7%, 4.9% and 6.8% lower than the remote teaching system based on virtual reality and based on artificial intelligence network. According to the results of Table 2, when the number of concurrent users was 500, the maximum database server CPU utilization rate of the designed textile pattern design course remote teaching system was 65.5%, 4.2% and 4.5% lower than the remote teaching system based on virtual reality and AI-based network. The Web server CPU maximum utilization comparison results are shown in Tables 3 and 4. Table 3. Maximum web server CPU usage with 200 concurrent users The number of experiments

Maximum CPU usage of Web server (%) The remote teaching system of the textile pattern design course is designed

Distance teaching system based on virtual reality

Distance teaching system of textile design based on AI network

1

48.4

55.0

56.2

2

47.7

55.5

58.4 (continued)

322

L. Zhang and X. Xue Table 3. (continued)

The number of experiments

Maximum CPU usage of Web server (%) The remote teaching system of the textile pattern design course is designed

Distance teaching system based on virtual reality

Distance teaching system of textile design based on AI network

3

47.8

56.2

56.7

4

48.5

55.0

58.8

5

46.6

57.4

57.9

6

47.2

57.8

55.6

7

46.5

56.6

58.3

8

47.9

57.5

57.5

9

47.6

55.9

55.2

10

46.5

56.5

56.5

Table 4. Maximum web server CPU usage with 500 concurrent users The number of experiment

Maximum CPU usage of web server (%) The remote teaching system of the textile pattern design course is designed

Distance teaching system based on virtual reality

Distance teaching system of textile design based on AI network

1

57.4

65.4

64.4

2

58.7

66.5

67.8

3

57.8

64.6

64.9

4

54.5

62.2

65.6

5

55.6

64.0

67.0

6

56.9

63.5

68.3

7

55.5

65.8

67.5

8

58.8

67.9

66.2

9

57.2

65.3

67.6

10

55.0

66.2

66.8

According to the results in Table 3, when the number of concurrent users is 200, the maximum utilization rate of web server CPU of the distance teaching system of textile pattern design course designed this time is 47.5%, which is 8.8% and 9.6% lower than the distance teaching system based on virtual reality and artificial intelligence network. According to the results in Table 4, when the number of concurrent users is 500, the

Design of Distance Teaching System for Textile Pattern

323

maximum utilization rate of web server CPU of the distance teaching system of textile pattern design course designed this time is 56.7%, which is 8.4% and 9.9% lower than the distance teaching system based on virtual reality and artificial intelligence network. Based on the above system test results, the maximum CPU utilization of the designed system does not exceed 70%, which meets the expectation of performance requirements, and the operation efficiency is higher than the two control systems when the number of concurrent users is 200 and 500, which has a good application prospect.

5 Conclusion After the test, the function of the distance teaching system of textile pattern design course integrated online and offline is basically normal, meets the design and application requirements, and the design and implementation are basically successful. However, due to the short time, the function of the system is not very perfect, and there is still room for further improvement. In terms of system functional design, limited by time and development experience, this paper only develops and realizes the basic functions of distance teaching. In the future, the functionality and practicability of the system could be optimized and expanded according to the actual application needs of the industry, so as to realize a more intelligent distance teaching system.

References 1. Mu, S., Wang, Y., Han, R.: Features, methods, and principles of hybrid instructional design. Educ. Res. 27(5), 63–72 (2021) 2. Cai, H., Dong, H., Wang, Q.: How teachers design graphical scaffolds to support STEM teaching effectively: a meta-analysis based on 30 experiments and quasi-experimental studies. E-educ. Res. 41(10), 73–81 (2020) 3. Li, P., Song, Z.: Design of English long-distance multimedia teaching system based on virtual reality. Mod. Electron. Tech. 43(14), 161–163,169 (2020) 4. Zhang, L.: Design of distance teaching system based on artificial intelligence network. Mod. Electron. Tech. 44(2), 131–134 (2021) 5. Yan, X., An, X., Dai, W., et al.: Image segmentation teaching system based on virtual scene fusion. Comput. Simul. 38(4), 331–337 (2021) 6. Xu, N., Fan, W.: Research on Interactive augmented reality teaching system for numerical optimization teaching. Comput. Simul. 37(11), 203–206, 298 (2020) 7. Deng, H., Zhan, C., Wang, Y., et al.: The online-offline blended teaching method and practice in discrete mathematics course. J. Southwest China Normal Univ. (Natural Science Edition) 46(9), 167–172 (2021)

Online Training System of Distribution Network Equipment Operation and Maintenance Security Based on Cloud Model Lizhen Zhang1(B) , Lu Liu2 , Yuexing Hu1 , Feng Gao2 , Wei Jin2 , and Chaojun Zhu3 1 ShanXi Electric Power Technical College, Taiyuan 030021, China

[email protected]

2 Beijing Kedong Electric Power Control System Company Limited, Beijing 100192, China 3 Department of Judicial Information Management, Sichuan Vocational College of Judicial

Police, Sichuan 618000, China

Abstract. Electricity demand is more and more big, the distribution of power distribution network equipment is becoming more and more complex, and the need for operational safety training, and the safety of the traditional functions of the system of online training modules are not comprehensive, unable to meet the demand of the current operational safety training, so based on cloud model to design new equipment operational safety online training system, the hardware part of the design of the ARM processor and FPGA chip, In the software part, the online training module of operation and maintenance safety is established, the online training database is designed based on the cloud model, and the online training of distribution network equipment operation and maintenance safety is realized, and the system test is carried out. The results show that the designed online training system of distribution network equipment operation and maintenance safety has good performance and certain application value. Keywords: Cloud model · Distribution network · Equipment · Operation and maintenance safety · Online training system

1 Introduction With the development of computer technology, computer has become a medium of auxiliary teaching, and computer-aided teaching has emerged. It can assist teachers to arrange teaching, manage or achieve individualized teaching [1], especially in imitating the actual situation and expanding the teaching level. In order to ensure the safe operation of power system, in addition to reliable equipment, reasonable power grid structure [2], scientific management mechanism, we must also strive to improve the operation skills of power employees. Due to the high-risk industry characteristics and special working environment of power grid, the traditional training methods can not meet the needs of power safety training. It has become a beneficial choice to apply virtual reality technology to build power virtual environment for operation skill training. © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 324–337, 2022. https://doi.org/10.1007/978-3-031-21161-4_25

Online Training System of Distribution Network Equipment Operation

325

For the power industry [3], due to the unpredictability of power accidents, there is a lack of experience in dealing with emergency accidents. Another way is a single book safety knowledge training, which is often one-way filling training and education. The active participation of employees is seriously insufficient, resulting in poor training effect. The third way is on-site training, which is a more effective training method. However, this method often requires a lot of funds and corresponding teacher investment. At the same time, due to the irrecoverability of power personal casualty accidents and personnel liability accidents [4], the on-site training can not fully simulate the wrong use of tools, possible dangerous points and possible safety accidents in actual production, and can not achieve a good training effect; In addition, in the process of technical competition assessment, it will also need to spend a lot of human, material and financial resources to prepare the work site and mechanical equipment. It is also restricted by weather and time, which is not convenient for regular development. Due to the high-risk industry characteristics and special working environment of power grid, the traditional training methods can not meet the needs of power safety training. It has become a beneficial choice to apply virtual reality technology to build power virtual environment for operation skill training. Therefore, this paper designs an online training system for distribution network equipment operation and maintenance safety based on cloud model, To solve the current training problems. In the hardware part, ARM processor and FPGA storage chip are designed, in the software part, the online training function module of operation and maintenance security is established, and the online training database is designed based on the cloud model to realize the online training of operation and maintenance security of distribution network equipment. The training and teaching of power safety regulations are assisted by computers to help learners learn, so as to replace the above-mentioned methods of conversational guidance, simulation exercises, situational learning, etc. [5–7]. The virtual simulation system of electric power safety regulations can provide learners with a more real environment, which is mainly to improve learners’ learning effectiveness, save teaching costs and improve teaching effectiveness. The simulation system learning of virtual reality technology not only assists the traditional vocational teaching, but also allows learners to contact simulation scenes anytime and anywhere. It also allows professional teachers and learners to have two-way communication opportunities and participate in discussions, so that learners have multiple choices of learning methods [8]. In addition, some security simulation systems further build virtual classrooms or laboratories, and use computer simulation to simulate the teaching materials and equipment in the laboratory [9], so that professional teachers and learners can learn, train and even test through an interactive virtual environment. Computer virtual vocational training has become the trend of power safety education in the future. In order to ensure the safe operation of power system, in addition to reliable equipment, reasonable power grid structure and scientific management mechanism, efforts must be made to improve the operation skills of power employees. The research shows that the designed system has good performance.

326

L. Zhang et al.

2 Hardware Design 2.1 ARM Processor In order to ensure the training effect of the operation and maintenance safety online training system and improve the processing speed of the system, the designed system uses ARM processor. ARM microprocessor is a 32-bit RISC processor with low power consumption and high performance. The core of ARM processor is unified and provided by arm company, while the on-chip components are diverse and designed by major semiconductor companies, which makes it possible to use different on-chip designs based on the same core when using arm to design embedded systems [10], which has great advantages. ARM processor supports seven operation modes, which can be roughly divided into user mode and privilege mode. It has 37 32-bit registers, of which 31 are general registers and 6 are special registers. Arm architecture supports storing word data in big end mode and small end mode. In big end mode, the high address of word data is stored in the low address, and in small end mode, the low address of word data is stored in the low address. At present, ARM processor cores can be divided into several series, such as ARM7, ARM9, arm9e, arm10e, ARM11, securcore, strong arm, X scale, etc. ARM7, ARM9, arm9e and arm10e are several mainstream series of ARM processor series. These four processor series provide specific solutions for specific embedded system requirements from different application performance. ARM7 series adopts three-stage pipeline. Arm7tmdi has low power consumption and relatively high performance. It is widely used in consumer electronic products that are sensitive to cost performance. ARM9 series adopts 5-level pipelining and Harvard structure, supports fullperformance MMU, and provides the best performance in terms of high performance and low power consumption. Arm9e series adopts 5-stage pipeline and supports DSP instruction set, which is suitable for occasions requiring high-speed digital signal processing. Securcore series is specially designed for applications with high security requirements. Strong arm/X scale is the arm core provided by Intel. ARM11 core focuses on the improvement of data processing ability, and implements a multiprocessor core in arm for the first time. Cortex is the latest ARM core. Based on armv7 architecture, it is a new product series, which is divided into three series: a (application field), R (real-time field) and m (control field). It improves the specific core for the high, medium and low-end needs. Therefore, this paper uses ARM9 as the microprocessor of the design system. 2.2 FPGA Storage Chip Compared with ASIC, FPGA has many excellent characteristics. However, the excellent characteristics of FPGA are at the expense of the performance of the chip. In order to realize the programmable characteristics, a large number of programmable switches are used in FPGA chip. These programmable switches have greater connection capacitance and resistance than wires, which reduces the speed, increases the circuit area, and has greater power consumption. For circuits of the same scale, the average area required by FPGA is about 40 times that of ASIC, the power consumption is about 12 times, and the average speed decreases by about 3.2 times L5. Therefore, in order to better meet

Online Training System of Distribution Network Equipment Operation

327

the needs of the market, FPGA is bound to develop in the direction of high density, high speed and low power consumption. In recent years, with the continuous progress of chip manufacturing technology, the gap between FPGA and ASIC in speed and area is narrowing. In 2010, Xilinx company launched artix-7 series FPGA products. The power consumption and performance of this series of products are greatly improved compared with the previous generation products. In addition to adopting 28 nm process, the unified architecture and scalable platform of this series of devices further simplify the system design. Compared with the previous generation spartan-6 series devices, its speed is increased by 30% and its size is reduced by 50%, And prices have fallen by 35% 6. With the further development of chip manufacturing technology, the gap between FPGA and ASIC in speed and area is still narrowing. However, with the increase of FPGA speed and smaller size, the power consumption of the chip is becoming more and more prominent. An FPGA device is actually a programmable logic unit array, and different logic units are connected through programmable wiring resources. The logic unit array is used to realize the logic functions of the circuit. Due to the limited functions that can be realized by each logic unit, wiring resources need to be used to connect the logic units to form a large system. The basic architecture of FPGA device is shown in Fig. 1.

I/O Pads Memory Blocks DSP Block

Special Feature Blocks

Configurable Logic Block

Routing Resources

Fig. 1. Basic architecture of FPGA device

As can be seen from Fig. 1, FPGA chip is mainly composed of programmable logic block, wiring resources and programmable input/output module. The programmable logic block is surrounded by prefabricated wiring resource channels, and the FPGA is surrounded by programmable input/output modules. In addition, modern FPGA devices also include embedded bottom functional units, embedded special hardware modules, block ram, clock management module and other resources. Programmable logic module, also known as configurable logic unit, is the basis for FPGA to realize various logic functions. Users can determine the function of each logic function unit and their interconnection relationship through programming, so as to realize the logic circuits with different functions. In the programmable logic block, the look-up table technology is used to realize the basic logic function.

328

L. Zhang et al.

The routing resources of FPGA devices are composed of programmable routing resources and global routing resources. Programmable wiring resources are one of the most important resources in FPGA. Programmable wiring resources connect each programmable logic block or input/output block to form a circuit with specific functions. Global wiring resources are used to realize global signals such as clock and reset of devices. Programmable I/O module is the interface between the chip and the outside world, which completes the interconnection between internal logic and external pins, and provides functions such as input buffer, output drive, interface level conversion, impedance matching and delay control. The programmable I/O module has the characteristics of low power consumption and high-speed connection. In order to make FPGA devices have the ability of joint design of software and hardware, realizing the same single-chip FPGA has become a system level design tool. Modern mainstream FPGA devices will provide many embedded bottom functional units, such as DSP, CPU, digital clock PLL, PLL and other software processing cores. With these embedded bottom functional units, FPGA devices can easily transition to SoC platform. In order to improve the performance of FPGA, FPGA manufacturers have integrated some special hard cores in the chip. These hard cores are equivalent to ASIC circuits and have strong processing power. For example, Xilinx’s high-end FPGA products not only integrate PowerPC Series CPU, but also embed DSP core module. In addition, in order to improve the multiplication speed of FPGA, special multipliers are integrated in mainstream FPGA. Modern FPGAs have a large number of configurable block ram. Block RAM can be configured into FIFO memory, single port RAM, real dual port RAM, pseudo dual port RAM (or simple dual port RAM) and content address memory (CAM), which greatly expands the application scope and design flexibility of FPGA. A large number of block ram provides very high memory access bandwidth for the logic in FPGA. Parallel memory access can greatly improve the performance of applications. Most FPGAs in the industry provide digital clock management. For example, Xilinx introduced the most advanced FPGA to provide digital clock management and phase loop locking. Phase loop locking can provide accurate clock synthesis, reduce jitter and realize filtering function.

3 Software Design 3.1 Establish Operation and Maintenance Safety Online Training Function Module In the final analysis, the security of power system is determined by the reliability of system equipment and the quality of production personnel. After the hardware facilities are determined, the technology, experience, proficiency and adaptability of production personnel are one of the important factors affecting the safe operation of power system. The virtual interactive training system of power safety regulations based on virtual reality technology is to develop a power virtual interactive training environment covering the combination of theoretical knowledge and practical safe operation skills of multiple disciplines such as power grid operation and maintenance according to the training needs of different trainees. Through these trainings, trainees can master the basic skills

Online Training System of Distribution Network Equipment Operation

329

of operation, maintenance and overhaul of power equipment faster and better, so as to avoid power safety accidents due to wrong operation to the greatest extent. The main content of theoretical knowledge training is the basic knowledge of electrical appliances and operation necessary for safety production personnel of power plant or power grid, that is, the (substation and line) of safety regulations. Multimedia methods such as pictures and words are used to explain, so as to make full use of this method to mobilize human senses and thinking imagination, deepen students’ understanding of the contents of power safety regulations, and improve students’ theoretical knowledge level, so as to avoid safety accidents in practical operation. It is mainly to realize the training on the working principle of power equipment (such as line, substation, transformer, etc.) and its related electrical connection relationship. In order to explain the power grid structure and the working principle of its equipment, the domestic leading three-dimensional virtual reality technology is used to simulate the power grid and geographical environment, which is presented in an interesting high realistic three-dimensional picture to give the trainees an immersive feeling. When the system is in the learning (training) mode and students practice themselves, if the students’ operation does not comply with the operation rules or the sequence of operation steps is wrong, the system interface will give students an alarm prompt. For example, the power must be checked before installing the grounding wire. If the student does not check the power, the system interface will pop up a prompt “you forgot to check the power!” At the same time, the student cannot proceed to the next step. The training management system collects and classifies relevant information such as trainees’ type of work and educational level, and formulates personalized training plans for different trainees. Training evaluation automatically tracks the completion of operation steps of each training content, records the training process of trainees, and automatically evaluates the operation results of trainees; The evaluation system forms an evaluation report according to the student information, training content, historical records, etc. This system fully draws lessons from the development mode and technology of 3D game. On the one hand, it creates an immersive feeling for the students in a realistic display mode, so that the students can get the effect similar to the on-site operation. For example, in the modeling of power construction process, in addition to simulating relevant power equipment, it is also necessary to simulate natural phenomena such as rain and snow and sparks splashing during electric shock, and provide sound to relevant places. At the same time, when simulating power operation, it is often designed according to the idea of three-dimensional dynamics, so as to well simulate the actions of power tools and equipment in the system, such as free fall, rolling, tilting and so on. It makes the whole training scene vivid and gives people a feeling very similar to the simulated objective world, just like in the real world. On the other hand, humancomputer interaction operation mode and integral system are adopted. In the virtual environment, the operator pushes the plot forward through independent selection. If the operation is wrong, the corresponding points will be deducted, so as to give full play to the students’ initiative. Through this game training method, improve the fun of power workers in the training process, change the traditional rigid training mode, and realize the transformation of students from “want me to train” to “I want to train”.

330

L. Zhang et al.

Cloud model is not only the specific implementation method of cloud, but also the basis of cloud based computing, reasoning and control. It can represent the process from qualitative concept to quantitative representation (forward cloud generator) or the process from quantitative representation to qualitative concept (reverse cloud generator). In essence, cloud model is an advanced computer user interface; In terms of application, it is the latest technology in the computer field, which is comprehensively developed by a variety of science and technology such as computer graphics technology, multimedia technology, human-computer interaction technology, network technology, stereo display technology and simulation technology. It is also the comprehensive application of various disciplines such as mechanics, mathematics, optics and mechanical kinematics. The system “moves” the actual power environment to the computer and provides users with various intuitive and natural real-time perceptual interactions such as vision, hearing, touch and so on. So that students can complete the training of various operation processes with the help of computer, keyboard and mouse without any danger, and carry out operation and theoretical assessment, multi-level learning and examination. By providing users with a virtual scene close to or better than the real environment, dangerous accidents are effectively avoided. However, in the process of distribution network equipment operation and maintenance, it is very easy for trainers to be tired, resulting in training error. The error expression is as follows:  1 π × (1) Ns = 2 s In formula (1), Ns represents online training error and s represents online training information. In this paper, the cloud model is used to eliminate the online training error, and the formula is as follows:  (2) Km = X 2 − Ns2 In formula (2), Km represents the online training situation after error elimination; X 2 represents distribution network operation and maintenance standard information.After eliminating the error, analyze the person in charge of online training. The training leader is the most important role in the whole training system except the system administrator. He is mainly responsible for training module management, training plan formulation, training result report generation and other affairs. Among them, the content management of training module not only needs to import relevant training content, but also needs to regularly upgrade, delete and add the expired content in the training system, so as to ensure that the training received by employees is up-to-date. For the training plan making part, because it is for the employees of the whole company, and the employees will be more or less different due to different foundations or positions. Therefore, in the training plan making part, different training plans need to be set according to different user types. In addition, for the overall training results after employee training, the training leader also needs to be able to regularly view the statistical results, so as to know the actual training situation, so as to make timely feedback and adjustment.

Online Training System of Distribution Network Equipment Operation

331

3.2 Design Online Training Database Based on Cloud Model Database is the core and foundation of information system. It can organize a large amount of data in information system according to certain rules; Provide the function of storing, maintaining and retrieving data; So that the information system can conveniently, timely and accurately obtain the required information from the database. Therefore, establishing a good data organization structure and database is the key to the design of information management system. The system database must store five types of information, namely model information, user information, training and assessment information, safety regulations and other regulatory information. Related model information table: stores the description information of related models and the drawing information in the virtual scene, such as the model category, name, number, three-dimensional coordinates, size and corresponding 3D model file name of the virtual power equipment. User information table: used to store personal information of system administrators, instructors and trainees. Training and assessment information table: used to store the assessment information of system students, such as theoretical examination, simulation operation, assessment system and scoring standard. Power safety regulation knowledge table: used to store safety regulations and other power regulation information. This paper integrates the relevant information to obtain a unified database table, as shown in Table 1. Table 1. Database table Field name

Explain

Type

Is it empty

Ma_ID

Name

Varchar

Yes

Ma_name

Full name

Varchar

No

Ma_Num

Number

Invarchar

No

Ma_Time

Time

Varchar

No

Ma_Type

Type

Char

No

Ma_Work

Type of work

Varchar

No

As shown in Table 1, the virtual interactive training system of power safety regulations is based on the internal LAN of the power system. Therefore, in the process of use, if there are accidents such as network disconnection, data loss will be caused. This not only brings inconvenience to users, but also leads to unfair assessment due to data loss in the process of training and assessment. On the other hand, in order to reduce the network burden. This topic adds Microsoft’s access database to the client, so that the client can save the data to the local encrypted database first, and transmit the data to the server at a breakpoint. When the service ends normally, the client will automatically empty the local database. The training and examination of safety regulations is a long-term and uninterrupted task. The safety regulation network examination module embedded in the system can realize the processing of batch or single knowledge points, automatic and manual volume forming, online monitoring, computer automatic marking and Score statistical analysis,

332

L. Zhang et al.

which greatly improves the work efficiency of training managers and can overcome the injustice of traditional training in the process of examination organization and implementation. Perfect evaluation system to achieve good evaluation effect. The application of virtual reality technology in power safety regulation virtual interactive training system has great advantages in operational knowledge evaluation and design compared with the traditional safety regulation training system. Because virtual reality technology can simulate the whole standardized operation process of power safety regulations and the corresponding hardware environment, not only questions (multiple-choice questions, blank filling questions, short answer questions, etc.), but also students can achieve good assessment results without the interference of time, region, environment or others. 3.3 Realize Online Training on Operation and Maintenance Safety of Distribution Network Equipment At present, cloud model has been widely valued, discussed and studied by the public, but it still focuses on academic units. Generally, the definition of virtual simulation is numerous. Views on cloud model definition: the first is that cloud model is the content that objects, attributes and relationships are adjusted into natural similarity or surface reality. Cloud model is a high-dimensional space generated by computers, including an environment that can enable participants to effectively recognize. The third view holds that the cloud model is a combination of sensing, feeling and thought. The computer changes the experience and emotion of the input through controlling the feeling. The cloud model has the following advantages. First, creativity: it can solve problems by using cloud model applications with different properties for different problems. Second, integration: the user’s senses produce the feeling of integrating into the virtual world and become a part of the virtual environment. Third, interaction: users can operate, change and interact with objects in the virtual world, just like the real world. Basic classification of computer virtual simulation technology. Due to the different requirements of virtual simulation technology for input and output equipment and the types of visual perception, it is generally divided into the following four categories, as shown in Fig. 2:

Virtual simulation technology type Desktop VR

Immersion VR

Project VR

Simulator VR

Fig. 2. Structure diagram of virtual simulation technology type

Online Training System of Distribution Network Equipment Operation

333

The specific contents of the virtual simulation technology type structure are as follows: First, desktop VR, desktop cloud model system, is a display system of personal computer architecture that users can afford. This system can allow individual users to integrate themselves into the virtual environment. Such a cost-effective system enables users to experience three-dimensional images through image, sound and real-time interaction. Users can place themselves in the desktop cloud model system through the guidance of the simulation program and the manipulation of the mouse or training rod. Its advantages are: the interaction of virtual objects can be controlled by the mouse, and the price of the system is low. The established virtual control platform system is the desktop cloud model system. Second, immersion VR, an integrated cloud model system, is a low-cost, professional and personal computer architecture cloud model system. This system allows independent users to completely immerse themselves in the virtual environment through threedimensional images containing video, audio and action. Users can roam in the virtual environment through the actions of hands and heads. Its advantages are: through the action of head and hand, it can fully interact with the virtual scene, so as to have a real feeling and experience. Third, in project VR, vision is generated by several projector devices around the user, and the whole virtual scene is projected around the user by polarizer to produce a three-dimensional feeling. This cloud model system is suitable for large-scale space and multi person viewing, so it is mostly used in large-scale conferences, exhibitions and entertainment. Fourth, simulator VR is the earliest developed cloud model system. It uses a special simulator to simulate special situations, which is mostly used in driving and flight training. Basically, the cloud model system is a new interface that integrates drawing, sound, image, animation and related peripherals to achieve communication and interaction between adults and machines. Cloud model provides a new three-dimensional visual, auditory and interactive man-machine interface. In recent years, due to the rapid development of computer software and hardware, the realistic situation of computer animation simulation generated by cloud model can be confused. The system consists of user, user group management and authority module, training content and training plan management module, employee training, examination module, virtual interactive display module, tutor system module and off-line rendering of power knowledge point scene. The following points are mainly considered in the system design, which are listed as follows: simulation training is the focus of the implementation of the system. Its implementation is completed by off-line rendering of the scene, and the display and interaction parts are completed by flash action script. Flash in the presentation layer interacts with the background in the way of web service. At the same time, the examination module also exchanges data with the background in the way of web service, so as to realize the SOA architecture in a complete sense. Flash is used as the foreground display layer and interaction layer. Since Adobe has compatible browser controls on major browsers, the final display style of each platform can be unified. In addition, because flash is a technology that has been tested by history and adopted by a large number of projects, it has achieved good results. Therefore, both interaction and

334

L. Zhang et al.

efficiency can well meet the requirements of this system design. The whole platform is built with SSH architecture, combined with the use of web service to build a loosely coupled system architecture. Combined with the display mode of JSP and flash, it realizes a series of comprehensive management of system user information, authority, simulation, assessment and so on. The off-line rendering of the scene used in the system adopts the 3D physical engine technology commonly used in 3D training and development, and fully considers the physical attributes such as gravity and collision. At the same time, the system strives to truly reflect the actual physical attributes of power equipment. With reference to the power facility dynamics physical engine technology, the authenticity of system simulation is greatly improved, It provides comprehensive support for the physical layer simulation of various models for the system. The use of artificial intelligence technology in the selection function of user question bank greatly improves the pertinence of user questionnaires and training projects, provides a complete intelligent learning system for each employee, and greatly improves the learning effect and training efficiency of employees.

4 System Test In order to test the training effect of the designed online training system, a test platform is built and the system is tested, as shown in the following contents. 4.1 Test Preparation In view of user simulation, the system will display the pre generated 3D videos and scenes in the form of Flash. At the same time, Action Script will be used to respond to the user’s mouse clicks. By judging the way of mouse placement, the mouse will be clicked into an object, and then the relevant events of the object will be invoked to feedback the user’s operation. The system test simulation diagram at this time is shown in Fig. 3.

Action script engine Pre generated FLASH scene event processing

Fig. 3. Test simulation diagram

It can be seen from Fig. 3 that in the system home page, click the “enter” button, and the system displays the following login input box. There are three function buttons

Online Training System of Distribution Network Equipment Operation

335

“login” and “exit” in the login part. The login button can be used by administrators and ordinary users, and the exit button can help users return to the system home page. Due to the limitation of software and hardware in the actual test environment, the test environment of the system continues to use the software and hardware in the development environment, and uses the development platform for relevant tests. The built test environment is shown in Table 2. Table 2. Test environment Hardware

Configuration

Processor

Intel I7

Memory

8G

Network card

10/100M adaptive Ethernet card

Hard disk

SATA 500G

Test tool 1

Apache JMeter

Test tool 2

Rational robot

Relevant test tools can be selected according to the test environment in Table 2. In order to make the whole system run normally in the production environment and ensure the error free and efficient operation of all realized functions, in addition, to ensure that all business requirements in the original requirements are realized, it is particularly necessary to test the system strictly, scientifically and comprehensively, For the design and implementation of the system, the objectives of the system test phase in the system implementation are defined as follows: ensure that 100% of the original business requirements are covered this time; All unit test cases pass 100%; Ensure that the code test coverage reaches more than 90%; No major defects in the system integration test stage; Ensure that the performance of the system program meets the original design standards. Because the system in this paper is relatively large, with many modules and a large amount of code, in order to ensure the normal operation of the system, and to ensure that the system can fully execute the whole logic according to the pre design, especially the normal operation of each function block, to ensure that the function block can produce the expected output results for the input, and to ensure the overall correctness of the whole system from the basic module level, it is necessary to unit test the overall code of the system, Ensure that the system can achieve the predetermined objectives in the best way. 4.2 Test Results and Discussion According to the above test environment, test the performance of the distribution network equipment operation and maintenance security online training system based on cloud model designed in this paper. The test results are shown in Table 3.

336

L. Zhang et al. Table 3. System test results

Module name

Number of use cases Number of executions Code coverage (%)

Permission system

12

12

95

User management

18

18

99

7

7

95

User test

13

13

94

Training content management

14

14

94

Content management of 12 examination question bank

12

93

Report form

10

92

User training

10

It can be seen from Table 3 that the code coverage of the design system is high, which proves that the performance of the design system is good, and it can realize efficient online training and has certain application value. The reason is that in the design process of the system in this paper, the establishment of operation and maintenance security online training function module, and then based on the cloud model design online training database, to a certain extent, is conducive to improving code coverage.

5 Conclusion The purpose of this study is to establish a platform system suitable for the training of electric power industry. For a long time due to the worker within the power company, the job is busy, less time to participate in safety training, training materials of old, makes the mass’s enthusiasm is not high, which leads to safety accidents and accidents frequently, only well-trained professionals, in order to better service for the development of electric power enterprises, this platform through relevant literature review, The theories put forward by various scholars for the establishment of safety discipline training simulation system are summarized, and the key links and practical requirements of domestic power safety training are discussed. By referring to some simulation system platforms that have been successfully run at home and abroad, the system that meets the needs of power company safety training is planned. The innovation of the research content is the establishment of operation and maintenance safety online training function module, and the design of online training database based on the cloud model. It is expected that the design and implementation results of this research can serve as a reference for the decision-making of upgrading the power system safety training in the future, so as to improve the quality and effectiveness of the power industry safety training. Fund Project. Science and Technology Project of State Grid Shanxi Electric Power Company Limited “Research and application of key technologies for intelligent analysis and evaluation of power distribution equipment operation and maintenance safety”.

Online Training System of Distribution Network Equipment Operation

337

References 1. Kuszewski, T., Braziewicz, J., Wysocka-Kunisz, M., et al.: Application of virtual environment for radiotherapy training system for educational purposes at Institute of Physics of Jan Kochanowski University in Kielce. Acta Phys. Polon. A, 139(3), 277–279 (2021) 2. Etinba, M., Butar, S., Akyüz, F., et al.: The effects of planting distance and training system on yield and fruit quality of peach. Mitteilungen Klosterneuburg 71(12), 74–89 (2021) 3. Sh, A., Pham, H.T.T., Dang, T.T.N., et al.: Nurses’ perception of individual and organizational changes caused by a novel clinical training system for new graduate nurses: A qualitative research using photovoice - ScienceDirect. Nurse Educ. Today 102(7),104901 (2021) 4. Iqbal, J., Sidhu, M.S.: Acceptance of dance training system based on augmented reality and technology acceptance model (TAM). Virtual Real. 5(1), 1–22 (2021) 5. Rashidov, N., Chowaniak, M., Niemiec, M., et al.: Assessment of the multiannual impact of the grape training system on GHG emissions in North Tajikistan. Energies 14(19), 1–13 (2021) 6. Espinoza, D.L., Carranza, V.G., Escamirosa, F.P., et al.: Integration of comprehensive metrics into the PsT1 neuroendoscopic training system. World Neurosurg. 151(7), 182–189 (2021) 7. Kumar, N.J., George, B., Sivaprakasam, M.: A sensor system to assess the ocular digital massage in an ophthalmic anaesthesia training system. IEEE Sens. J. 19(22), 10812–10820 (2019) 8. Wu, C., Gu, W., Zhou, S., et al.: Coordinated optimal power flow for integrated active distribution network and virtual power plants using decentralized algorithm. IEEE Trans. Power Syst. 36(4), 3541–3551 (2021) 9. Chong, G., Yuansheng, H., Haijie, M.: Simulation of spatial load density distribution in medium voltage distribution network. Comput. Simul. 37(3), 56–60 (2020) 10. Ingham, G., Plastow, K., Kippen, R., et al.: Closer supervision in Australian general practice training: planning major system change. Aust. J. Prim. Health 26(2), 184–190 (2020)

Design of Intelligent Education Management Information System in Colleges and Universities from the Perspective of Big Data Haiyan Zhao1(B) , Shiyuan Liu1 , Lin Xiao1 , Kun Liu1 , and Fuxia Liu2 1 Beijing Union University, Beijing 100101, China

[email protected] 2 Scientific Research Department, Hebei Open University, Shijiazhuang 050080, China

Abstract. There are many internal data types in the education management information system, which leads to the poor management effect of the intelligent education management information system in colleges and universities. Design an intelligent education management information system in Colleges and universities from the perspective of big data. The hardware part of the system focuses on the design of DSP module, FPGA module, analog-to-digital conversion module, A/D conversion module interface circuit and communication interface. In the system software part, the big data technology is used to divide and allocate data sets, so as to realize the design of intelligent education management information system in colleges and universities. Experimental results show that the proposed intelligent education management information system can effectively improve the accuracy and efficiency of resource mining, and has a good effect. Keywords: Big data horizon · Education management information · Excavate · Transformation · Clustering · Match

1 Introduction As one of the landmarks of the information age, big data is widely used in all aspects of society. Big data is a new thing gradually evolved with the rapid growth of massive data. It is different from traditional data. It has a huge volume, a wide variety of data, and the correlation between data is very complex, which has a great impact on business, education and culture [1]. The development of science and technology promotes the continuous progress of Internet technology. Big data is widely used in college teaching management information system. Big data is essentially produced with the in-depth development of the Internet, and it is still in a blowout development. This is the era of massive information storage, and its real significance lies in the analysis and use of data. Through continuous mining, data can play a real role. In fact, the proposal of big data also brings development opportunities and new challenges to colleges and universities in the forefront of science and technology, which requires college education to actively comply with the development of the trend © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 338–350, 2022. https://doi.org/10.1007/978-3-031-21161-4_26

Design of Intelligent Education Management Information System

339

of the times. Therefore, taking advantage of the advantages brought by big data, we should constantly innovate college teaching management information system and build a new college education model. The hardware part of the college intelligent education management information system under the vision of big data focuses on the design of DSP module, FPGA module, analog-to-digital conversion module, A/D conversion module, interface circuit and communication interface. In the system software part, the big data technology is used to allocate data sets to achieve the design of the college intelligent education management information system.

2 System Hardware Framework The hardware framework of the whole system is shown in Fig. 1:

Power module

Memory card

Communicatio n interface

DSP device

A/D conversion module interface circuit

FPGA module

Analog to digital conversion module

Fig. 1. System hardware framework

Among them, DSP device is a professional data processor, which can realize the highspeed processing of a large number of data and the operation of complex algorithms, and has won the favor of the majority of designers with its efficient performance. DSP is the core of data acquisition of the whole system. The data processing link of the system is mainly completed in DSP. DSP device has many optional IP core resources. The data output of the system is completed by the PCI bus interface of DSP device, and the data is sent to the back-end computer for further processing, storage and recording. FPGA is the heart and bridge of the whole system. Because the working frequency and data bit width between the data acquisition module and the data processing module cannot be directly matched, and the FPGA device has the characteristics of flexible design and compatibility with a variety of interface levels, the transition and connection between the acquisition module and the processing module are realized through FPGA. In this paper, a data buffer FIFO is designed in FPGA device to realize the cache of system data and the transformation of data bit width and transmission frequency [2]. PLL in FPGA device is used to double the frequency of clock signal to provide clock signal for the system.

340

H. Zhao et al.

2.1 DSP Module Design The system uses TigerSHARC Series High-Performance floating-point DSP: TS201 to complete baseband signal processing, so as to ensure high data transmission rate, large data exchange task and real-time transmission. TigerSHARC processor is the core of a software defined radio solution for baseband modem, and supports high-level language and operating system. Following ADSP-TS101, ADI released new members of TigerSHARC in 2003: ADSP-TS201, ADSP-TS202 and ADSP-TS203. The operating frequency of its core has reached 600 MHz, and the on-chip memory has been increased to 24 Mbit. ADSP-TS201 is a static superscalar processor with high performance, which is specially optimized for large signal processing tasks and communication structure. The main performances of ADSP-TS201 are as follows: First, up to 600 MHz operation speed, 1.6 ns command cycle; Second, 24 Mbit on-chip DRAM is divided into six 4 M bit blocks M0, M2, M4, M6, M8 and M10; Third, dual operation module, each calculation block includes 1 ALU, 1 multiplier, 1 shifter [3], 1 register group and 1 communication logic operation unit; Fourth, double integer ALU provides data addressing and pointer operation functions; Fifth, integrated I/O interface, including 14 channel DMA controller, external port, 4 chain junctions, SDRAM controller, programmable flag pin, 2 timers and timer output pins, etc. for system connection; Sixth, IEEE1149 1. Compatible JTAG port is used for online simulation; Seventh, four groups of fully bidirectional chain intersections, each containing 4-bit independent input and 4-bit independent output [4], and using LVDS technology, the link throughput reaches 4G bytes; Eighth, up to 8 TIGERSHARCDSPS can be seamlessly connected through the shared bus; Ninth, there are four boot modes: EPROM boot, host boot, link boot, and no boot. The main advantages of ADSP-TS201 are: (1) Provide high-performance static superscalar DSP operation, specially optimized for communication and applications requiring multiple DSP processors; (2) Excellent DSP algorithm and I/O performance; (3) The DMA controller supports 14 DMA channels and can complete the high-speed transmission of overhead between on-chip memory, off-chip memory, memory mapping peripherals, chain junctions, host processors and other multiprocessors; (4) Very flexible instruction set and DSP structure supporting high-level language are convenient for DSP programming; (5) A scalable multiprocessor system requires only low communication overhead. On this basis, the external memory interface is designed. Although ADSP-TS201 has 24 Mbit on-chip DRAM, it is not enough to use only on-chip memory because DSP undertakes heavy signal processing and data exchange tasks in the whole system. Therefore, in this hardware system, the external 128 Mbit SDRAM is used to expand

Design of Intelligent Education Management Information System

341

the storage space of DSP. In addition, a Flash is configured on the periphery of the DSP to guide the DSP loader after power on. The mode of SDRAM access needs to be set through the register sdrcon. Meanwhile, in order to meet the timing requirements of SDRAM, ADSP-TS20xs allows the addressing of SDRAM to adopt pipeline mode, pipeline depth (definition of pipelining depth: delay between address valid period and data valid period) is set in register sdrcon. In addition, SDRAM configuration requirements of different manufacturers can be met by properly configuring sdrcon register of ADSP-TS201 processor. Before selecting appropriate storage devices, first understand the characteristics of SDRAM controller of DSP and SDRAM controller characteristics As shown in Table 1: Table 1. SDRAM controller characteristics of TS201 Serial number

Controller parameters

Corresponding valid values

1

Operating voltage

3.3 V and 2.5 V

2

Maximum supported operating frequency

125 MHz

3

Maximum supported storage capacity

256 Mbytes

4

Number of SDRAM banks

2 or 4

5

CAS delay

Programmable: 1–3 SCLKS

6

Refresh rate

Programmable: 32–64 ms

7

BurstLength

Fullpageburst

8

Page size

Programmable: 256, 512 or 1024 words

9

Initialization sequence

MRS-> REF REF-> MRS

The above modules are used as the main controller of the system to improve the response speed of the system. 2.2 FPGA Module Design FPGA module acts as the heart, commander and Bridge in the whole system. The clock signal and control signal of the whole system are provided by FPGA. In order to solve the mismatch between the working frequency and data bit width between A/D module and DSP module, the system realizes the connection and data transmission between them through FPGA [5]. FPGA, namely field programmable gate array, is a high-performance programmable logic device developed on the basis of complex programmable logic devices. FPGA devices are generally based on SRAM process, and some devices are based on Flash process or anti fuse process. FPGA devices have a very high degree of integration, with

342

H. Zhao et al.

device density ranging from tens of thousands of system gates to tens of millions of system gates. They are grouped into programmable I/O units, embedded block RAM, underlying embedded units, rich wiring resources, basic programmable logic units, special hard cores for cores, etc. Compared with ASIC chip, FPGA chip has the advantages of short development cycle, low design cost, standard product test free, advanced development tools and realtime online detection. Therefore, FPGA devices are widely used in the original design of products. Traditional data acquisition systems are generally designed based on DSP structure. This system generally designs DSP peripheral interface circuit through smallscale digital devices. Such interface circuit occupies a large space and does not have scalability. FPGA device has the characteristics of programmability, good expansibility and high integration, which is very suitable for the application of DSP peripheral interface circuit. The system uses FPGA devices to realize the interface connection between DSP and A/D module, and undertakes the tasks of data preprocessing, caching and timing control in the system [6]. The CycloneII series chip Ep2C5 of Altera company is selected, which has 4608 logic units, 26 M4K random access memory modules, 13 embedded multipliers, 2 phaselocked loops and 158 user I/O interfaces. This series of chips are manufactured by lowK electrolyte process based on 90 nm. Therefore, the chip has good fast response and low power consumption performance. It is the lowest cost FPGA chip in the industry. CycloneII series chips have the following characteristics: (1) High density structure. With 4608-68416 logic units. M4K embedded memory module; 1.1M embedded RAM; working frequency up to 260 MHz. (2) Embedded multiplier. With 150 18 × 18 bit embedded multiplier (can be used as two separate 9 × 9); Operating frequency of 250 MHz; Optional input and output registers; (3) Advanced I/O interface technology. Support various types of high-speed differential I/O interfaces, including LVDS, RSDS, mini-LVDS, LVPECL, differential HSTL and differential SSTL; support single ended standard I/O interfaces of 2.5 V and 1.8 V; support PCI bus interface and PCI-E interface with 32-bit or 66 bit operating frequency of 33 MHz or 66 MHz; high-speed external memory interface, etc. [7]; (4) Flexible clock management circuit with 2–4 PLLs, 16 global clock networks; (5) Multiple chip configuration modes, including active configuration, passive configuration and JTAG configuration; (6) A variety of embedded IP cores. In order to optimize CycloneII devices, more than 40 IP cores are embedded, including NiosII embedded processor, PCI bus, fir compiler, DDRSDRAM controller, FFT/IFFT, etc. On this basis, the peripheral circuit of FPGA is designed: First, the configuration circuit. The CycloneII device is based on the SRAM structure. The configuration information of the chip is stored in the SRAM. Because the data in the SRAM is lost after power failure, the configuration information of the chip will be reloaded after each power on. FPGA chip has three configuration modes: active configuration mode, passive configuration mode and JTAG configuration mode. Active configuration mode: in as mode, the FPGA chip controls the configuration process, and

Design of Intelligent Education Management Information System

343

sends the control signal and synchronization signal to the configuration chip of EPCs series to complete the configuration operation. Passive configuration mode: in PS mode, the configuration process is controlled by an external processor or computer, and the configuration process is completed by enhanced configuration devices. Second, clock circuit. FPGA chip is the heart of the system and provides clock signal for the system. The system provides 50 MHz clock source for FPGA chip, which is provided by the clock generation circuit in the figure below. After entering the FPGA module, the clock signal is divided and multiplied by the PLL, and then sent to the A/D acquisition conversion module and DSP data processing module respectively. 2.3 Design of Analog-to-Digital Conversion Module The signal output by the sensor is generally analog signal. In the system with computer as the core, the analog signal output by the sensor must be converted into digital signal before it can be recognized and analyzed by digital electronic equipment. A/D chip is a special analog-to-digital converter and the primary link of data acquisition system. A/D converter is an extremely important key component of data acquisition system. The performance of a/D converter is directly related to the quality of the whole data acquisition system. The so-called A/D conversion is to convert the analog signal output by the sensor into digital signal for the identification and processing of digital circuit [8]. In the A/D converter, the amplitude of the input analog signal is continuous in time, but the output digital signal is discrete. Therefore, the sampling of the input signal can only be completed at some selected moments, and then the sampled value can be converted into digital electricity. The working process of A/D conversion is divided into four steps: sampling, holding, quantization and coding. Firstly, the analog signal is sampled. After sampling, the signal enters the hold state. Within the hold time, the sampled voltage is converted into digital quantity, and then encoded. Finally, the conversion results are output according to the fixed coding format, and then enter the next round of sampling. In this study, a high-performance A/D converter AD9254 launched by American analog devices company is used. The sampling rate of this A/D converter is millions of samples per second and has a resolution of 14 bits. This A/D converter works at 70 MHz and has a spurious free dynamic range of 83 dB. At the same time, the power consumption is only 50% lower than that of similar products. Because AD9254 has high SFDR, low power consumption and small package size (7 mm × 7 mm 48 pin LFCSP) has many perfect performances. The device is very suitable for the application of highspeed acquisition system. The Sha of AD9254 operates in two modes: sampling mode and hold mode. When SHA is switched to the sampling mode, the signal source must be able to complete the charging process of the sampling capacitor within half a cycle. The analog input of the AD9254 has no internal DC bias. When collecting AC signal, users need to set bias circuit externally. Generally, the bias voltage value should be set to 0.55 times of avdd. The common mode reference voltage of the circuit can be directly provided by the chip pin CML. At this time, the CML pin needs to be grounded through a 0.1 uF capacitor to realize decoupling. Power down mode: AD9254 has a low-power standby mode. When pdwn pin is connected to low level, AD9254 is in normal working mode; When pdwn pin is connected to high level, AD9254 is set to powerdown mode.

344

H. Zhao et al.

In this mode, the power consumed by the A/D converter is only 1.8 mW, and the output is in a high resistance state. 2.4 Interface Circuit Design of A/D Conversion Module The analog signal input terminal of AD9254 has three configuration modes: transformer input. When the signal-to-noise ratio is very demanding, the transformer is usually used as the input driving circuit. When the frequency of the input signal is in the second Nyquist region or above, the performance of many amplifiers can not meet the signal-tonoise ratio requirements of AD9254. When selecting the transformer, it is necessary to consider the characteristics of the signal. Many RF Transformers will reach saturation at a frequency of several MHz [9]. When the signal energy exceeds this frequency range, it will cause coresaturation and signal distortion. The transformer chip of the system adopts ADT1-1WT+RF transformer of Mini-Circuits company. The power consumption of the transformer is 0.5 W, the DC current is 30 mA, the signal bandwidth is 0.4800 MHz and the voltage ratio is 1:1. It has excellent performance in phase imbalance and amplitude imbalance transmission, and is very suitable for impedance matching and stable amplifier. The analog signal is input through the PRI pin of the transformer and then output in the form of differential signal from the two pins on the secondary side of the transformer. SEC_CT pin is used to provide A/D common mode voltage. A filter link is introduced between the pins of the secondary side of the transformer to obtain a stable output signal. AD8138 differential drive. When the input signal frequency is in the second Nyquist region, in addition to using the transformer as the driving circuit, a special driving chip can also be selected as the differential driving circuit. The system selects AD8138 chip of ADI company, which is a special low distortion differential A/D converter driving chip. It can convert single ended signal into differential signal, realize balanced signal transmission, and help the A/D converter achieve the most perfect performance. The collected analog signal is input by NI+pin of AD8138, and the differential signal is output by +OUT and −OUT pins after passing through AD8138 chip, The common mode signal is provided by the Vocm pin. The output pin also introduces a filtering link to filter out clutter and obtain a stable output signal. Differential double unbalanced transformer coupling drive. In those environments with very strict requirements for SFDR, using differential double unbalanced transformer coupling as input drive circuit is a good choice. In the data acquisition system studied in this paper, because the system does not have very strict requirements for SFDR, the first two driving modes are selected at the analog input of AD9254 chip, namely transformer driving mode and AD8138 driving mode. For the clock input circuit, the high-speed A/D converter has very strict requirements for the quality of the sampling clock signal. Any noise or jitter in the clock signal will cause the output signal distortion of the A/D converter. Therefore, the clock circuit design is an important link in the high-speed circuit design, which needs the designer to pay enough attention. The clock input structure of AD9254 is very flexible, including CMOS, LVDS, LVPECL and sinusoidal signal. However, no matter which clock signal is selected, jitter is still the most important indicator of clock source. A/D chip needs a low jitter clock

Design of Intelligent Education Management Information System

345

signal source to provide clock signal. Without a low jitter clock source, a low voltage differential signal can be AC coupled to provide a clock signal for the AD9254. A/D951x clock driver series chips have the performance of providing low jitter clock. The system selects ad9515 chip with more functions and better performance. The clock signal is provided by the FPGA module. After filtering by AD9515, it provides a stable clock signal for AD9254. 2.5 Communication Interface Design Considering that the final data source and data terminal of the whole system are PC, it is necessary to design the communication interface between the hardware platform and PC. In addition, in order to complete the data exchange between DSP and FPGA, it is also necessary to establish the corresponding communication interface between them. In the system, the network controller selects the 10/100M adaptive fast Ethernet control processor based on asynchronous ISA bus of DAVICOM company model DM9000E, and synthesizes MAC, PHY and MMU. DM9000E is a fully integrated, low-cost, single fast Ethernet control chip with 16 K high-capacity FIFO, 4-channel multi-functional GPIO, full duplex operation and other functions. The physical layer supports Ethernet interface protocol. Because data is sometimes received in burst form, DM9000E also integrates a receiving buffer, so that when data is received, it can put the data into the buffer, and then the data link layer can directly take the data from the buffer. The link layer usually includes the device driver in the operating system and the corresponding network interface card in the computer. They process the physical interface detail data with the cable together, and its buffer can be used to temporarily store the frames to be sent or received. The serial port chip used in this system is ST16C550, which is one of the most stable and reliable UART chips launched by EXAR company, and can provide serial/parallel and parallel/serial conversion functions of data. The synchronization function of serial data stream is realized by adding start and end bits to the transmission data to form data bytes. Ensure data integrity by attaching parity bits to data bytes. The receiver checks the parity bit to determine whether there is a transmission error. It will be very complex to realize these functions with ordinary circuits, especially when the circuit is integrated on a single chip. ST16C550 adopts 0.6 mm CMOS process to meet the requirements of low power consumption, high speed and high integration. ST16C550 is equipped with 16 byte transceiver FIFO memory to communicate with high-speed memory. Under the external clock of 24 MHz, the transmission rate of ST16C550 can reach 1.5 Mbps. Other features of ST16C550 can be obtained through internal registers. Optional receive FIFO trigger points, optional TX and RX baud rates are some features of ST16C550.

3 Software Design of Education Management Information System from the Perspective of Big Data In the system software part, big data technology is mainly used for education information management, and the specific contents are as follows.

346

H. Zhao et al.

3.1 Data Set Division Parallel association rule mining algorithm is to divide the data set into each data block, and then scan each data block for association rule mining. In parallel association rule mining algorithm, different partition methods will greatly affect the mining efficiency of the algorithm. The most effective data set partition method must make a frequent item set in the data set be a frequent item set in only one or few data blocks. Therefore, data clustering technology can be used to cluster the transactions in the whole data set, and each cluster can divide the whole data set into each data block. Common clustering methods generally need to scan the whole data set for many times. When the data set contains a large number of transactions and the computational cost of clustering is too high, the parallel association rule mining algorithm will no longer have advantages. Therefore, it is very important to find a better and faster data set partition method. There is often a classification hierarchy relationship between various items in the data set. When the average transaction length of the data set is quite different from the number of items, the classification hierarchy relationship is used to cluster the transactions in the data set. The calculation is as follows: |T − Ii | ≤α |T |

(1)

In formula (1), T represents the task to be divided, Ii represents the data block, and α represents the specified parameters. Based on the above calculation, it can be ensured that a transaction will not be divided into multiple data blocks. 3.2 Data Set Allocation In this part, for the distribution of education management information, the data distribution process is set in the program design stage. When executing the program, the pre arranged distribution method is used to distribute the data to each node for calculation. In the environment with light network load and machine load, static data distribution can effectively deal with the problem of load balance. Dynamic data allocation [10], the data allocation process is carried out dynamically during the calculation process. The master node does not allocate all data at one time. When a slave node completes the calculation and transmits the results to the master node, the master node redistributes the remaining data until all data are processed [11]. For cloud computing environment, static data distribution is prone to load imbalance, but it is easy to design coarse-grained parallel programs to reduce the traffic in the computing process; Dynamic data allocation is easier to achieve load balancing and facilitate the use of high-performance machines in the cloud environment for computing, but the corresponding parallel programs are difficult to design and usually require some additional overhead. For example, in order to understand the computing speed of each node, increase the traffic to determine the next data allocation. Based on the above analysis, this study uses the following methods to allocate data sets: αj =

Wj W

(2)

Design of Intelligent Education Management Information System

347

In formula (2), αj represents the proportion of data set allocation on node j, and Wj represents the weight of the overall processing capacity of the node executing the Map task in the cluster system. Based on the above calculation, the resource allocation is realized to realize the allocation of the education management information system.

4 Experimental Comparison In order to verify the application effect of the system, taking a school as an example, this paper analyzes the application effect of the previous system and the proposed system. The hardware tool used is an ordinary PC, which is basically configured as Intel(R) Core(TM)2 Duo CPU P8600 2.4 GHz and 32 G memory. The Quist synthetic data generator of IBM is used to generate the experimental data of association rule mining objects. The main experimental results are as follows. The comparison results between the correct mining quantity of the previous system and the correct mining quantity after the application of the system of this study are shown in Table 2: Table 2. Number of correct information mining Experiment serial number/piece

Item quantity/piece

Correct excavation quantity of the designed system/piece

Correct excavation quantity of previous system/piece

1

100

100

98

2

100

100

95

3

120

119

110

4

130

129

120

5

120

120

110

6

150

148

135

7

150

150

140

8

160

160

150

9

170

168

160

10

180

178

170

Based on Table 2, it can be found that the system studied can mine relevant information more accurately. However, in the previous system, there are less correct mining data and low accuracy. The comparison results of teaching resource mining time between the information management system of this study and the previous management information system are shown in Fig. 2:

348

H. Zhao et al.

Mining time of teaching resources / min

5

4

3 Previous system 2

Research sys tem

1

50

100 150 Dat a quantity / piece

200

250

Fig. 2. Comparison of teaching resource mining expenses under different data volumes

The comparison results of student information mining expenses of the two systems under different data volumes are shown in Fig. 3:

Student information mining time / min

5

4

Previous system

3

2 Research sys tem 1

50

150 100 Dat a quantity / piece

200

250

Fig. 3. Comparison of student information mining expenses under different data volumes

The comparison results of teacher information mining overhead time of the two systems under different data volumes are shown in Fig. 4:

Design of Intelligent Education Management Information System

349

Teacher information mining time / min

5

4 Previous system 3

2

1 Research sys tem 50

150 100 Dat a quantity / piece

200

250

Fig. 4. Comparison of teacher information mining expenses under different data volumes

Based on Fig. 4, the proposed intelligent education management information system is less affected by the amount of data and can ensure a faster information mining speed. In the past, the system used in the school spent more time on teaching resource mining, which was greatly affected by the amount of data. To sum up, the designed intelligent education management information system of colleges and universities from the perspective of big data can mine relevant information more accurately and is less affected by the amount of data, which can ensure faster information mining speed and better performance.

5 Conclusion Based on the above process, the research of University intelligent education management information system from the perspective of big data is completed, and the effectiveness of the system is verified by experiments. The innovation point of the system is to design DSP module, FPGA module, ANALOG-to-digital conversion module, A/D conversion module interface circuit and communication interface in the hardware part. In the software part of the system, the big data technology is used to divide and allocate data sets to achieve the design of intelligent education management information system in colleges and universities. The proposed university intelligent education management information system can effectively improve the accuracy and efficiency of resource mining, and has a good effect. However, the research time is limited, and the proposed system still has shortcomings. In the follow-up research, the system will be further optimized. Fund Project. 1. Funded by Beijing Higher Education Association (YB2021113). 2. Funded by research project of Beijing Union University (ZK90202105).

References 1. Chen, J., Wang, Z., Mao, T.: Resource management for hybrid RF/VLC V2I wireless communication system. IEEE Commun. Lett. 24(4), 868–871 (2020)

350

H. Zhao et al.

2. Momblanch, A., et al.: Untangling the water-food-energy-environment nexus for global change adaptation in a complex Himalayan water resource system. Sci. Total Environ. 655(10), 35–47 (2019) 3. Fu, S., Gao, J., Zhao, L.: Integrated resource management for terrestrial-satellite systems. IEEE Trans. Veh. Technol. 69(3), 3256–3266 (2020) 4. Schulze, C., Thiede, S., Thiede, B., Kurle, D., Blume, S., Herrmann, C.: Cooling tower management in manufacturing companies: a cyber-physical system approach. J. Clean. Prod. 211(20), 428–441 (2019) 5. Sinha, D., Roy, R.: Reviewing cyber-physical system as a part of smart factory in industry 4.0. IEEE Eng. Manag. Rev. 48(2), 103–117 (2020) 6. Hamidpour, H., Aghaei, J., Pirouzi, S., Dehghan, S., Niknam, T.: Flexible, reliable and renewable power system resource planning considering energy storage systems and demand response programs. IET Renew. Power Gener. 13(11), 1862–1872 (2019) 7. Tinoco, J., de Granrut, M., Dias, D., Miranda, T., Simon, A.-G.: Piezometric level prediction based on data mining techniques. Neural Comput. Appl. 32(8), 4009–4024 (2019). https:// doi.org/10.1007/s00521-019-04392-6 8. Sekiguchi, H., et al.: Computerized data mining analysis of keywords as indicators of the concepts in AHA-BLS guideline updates. Am. J. Emerg. Med. 38(7), 1436–1440 (2020) 9. Mohamed, A., Molendijk, J., Hill, M.M.: Lipidr: a software tool for data mining and analysis of lipidomics datasets. J. Proteome Res. 19(7), 2890–2897 (2020) 10. Campo-Vila, J.D., Takilalte, A., Bifet, A., Mora-López, L.: Binding data mining and expert knowledge for one-day-ahead prediction of hourly global solar radiation. Expert Syst. Appl. 167(8), 114147 (2020) 11. Pang, H., Zhao, W.: Big data efficient hybrid cloud storage model and algorithm simulation under 5 G network. Comput. Simul. 37(7), 350–353, 379 (2020)

Mobile Terminal-Based Remote Counseling Education System for Middle School Students’ Mental Health Fuyu Du1(B) and Jing Zhu2 1 Xinyu University, Xinyu 338004, China

[email protected] 2 Department of Building Engineering, Shi Jia Zhuang University of Applied Technology,

Shijiazhuang 050000, China

Abstract. Mental health education is conducive to the improvement of personality, so college students’ mental health education has gradually received attention, and it is necessary to comprehensively consider optimizing students’ psychological quality and improving students’ mental health level. The traditional mental health distance counseling education system has gradually been unable to meet the current teaching needs, so based on the mobile terminal, a new mental health distance counseling education system for middle school students is designed. The hardware part designs the time counter and storage, and the software part designs the mental health distance counseling education framework first. Secondly, based on the mobile terminal, a distance counseling education model for middle school students’ mental health is constructed. Finally, a functional module of mental health remote counseling is designed, and an auxiliary teaching database is designed. The system test results show that the designed mental health distance counseling education system has good performance, can achieve efficient distance education, and has certain application value. The system can be used as a reference for follow-up mental health distance counseling education. Keywords: Mobile terminal · Middle school students · Mental health · Distance · Counseling education system

1 Introduction Whether it is the needs of psychological development research, or the needs of educational practice. Guided by the system development concept and development concept, carry out research on the development of mental health system [1]. Under the unique social and cultural environment of our country, the research on the influence mechanism of different levels of social system on the psychological development of college students. It will greatly promote our systematic understanding of the psychological development of college students [2]. Individuals will inevitably experience various psychological conflicts and contradictions in the process of psychological growth. If they are not adjusted © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 351–364, 2022. https://doi.org/10.1007/978-3-031-21161-4_27

352

F. Du and J. Zhu

properly, they will also cause psychological disorders and even psychological diseases. The proposal of behavioral view has enlightening significance for our work on individual psychological counseling and psychotherapy. Only by organically combining the promotion of individual psychological growth with changing the external environment. The two can adjust to each other and change synchronously in order to receive long-term stable effects [3]. For example, only mental health education is provided to students, without changing the educational concepts and methods of teachers and parents. It is very difficult to work without changing the cultural environment of the community and of the society as a whole. The survival value of an individual is reflected in the realization of personal goals, and the realization of personal goals is closely related to the natural and social environment in which they live. The realization of personal goals depends on the behavior of the individual [4–6], and to understand the behavior of an individual, we must first understand the environment he is in. Individual behavior occurs in the interaction between individual and individual, individual and nature, and individual and society. Understanding the relationship between the individual and the environment can help to regulate and control the behavior of the individual. Make it move in a direction that benefits both the individual and the environment. In addition, people are individuals with subjective initiative. If people purposefully control the behavior of the environment [7], consciously promote the harmony between the natural environment and the social environment. Strengthening the correct guidance of individual behavior will be conducive to the realization of individual goals. The formation of a healthy psychological mechanism is the result of the joint action of multiple elements and multiple systems. Once it is formed, it will help college students to form a good personality and psychological quality and improve their psychological endurance to overcome learning difficulties. Stimulate the enthusiasm and creativity of learning, so as to improve learning efficiency and develop intelligence. In the learning process [8], if college students are full of vigor and happy, they will mobilize the enthusiasm of their intellectual activities and promote the development of intelligence. On the contrary, it will hinder the development of intelligence and is not conducive to the completion of studies. Poor academic performance in turn causes new psychological problems or exacerbates the original bad psychological state, leading to a vicious circle. It can be seen that overcome bad emotions and maintain a happy mental state. Maintaining a good and healthy state of mind is an important condition for creative thinking, improving learning efficiency, and mastering scientific and cultural knowledge. Healthy psychology is the foundation for college students to successfully complete their studies. It is the need of modern education to carry out mental health education for college students, which will play a positive role in promoting talent training. Therefore, this article proposed based on the mobile terminal middle school student mental health remote instruction education system. The hardware part designs the time counter and memory. Memory is the use of faster storage performance, cost consumption structure more reasonable pyramid structure. In the software part, on the basis of the framework of mental health remote counseling education, the related indexes of remote counseling model are calculated. And designed the remote tutoring function module,

Mobile Terminal-Based Remote Counseling Education System

353

auxiliary teaching database. To achieve rapid, effective and stable psychological health of secondary school students remote counseling.

2 Hardware Design 2.1 Time Counter Because distance tutoring education requires higher signal capture efficiency. Therefore, the designed system adopts the time counter to improve the communication quality of the system. After the signal is input to the counter, it needs to go through a shaping circuit. The main components of a general shaping circuit include an attenuator, an amplifier and a Schmitt trigger. The Schmitt trigger is required. It can change the output of the amplifier to be compatible with the internal format of the count register [9, 10]. The sensitivity of a counter is measured by the minimum value of a particular input pulse that the counter can recognize and count. It is determined by the gain of the amplifier and the hysteresis voltage difference of the Schmitt trigger. But the counter is not as sensitive to the input signal as possible. In general, conventional counters have a very sensitive front end and allow a wide range of frequency inputs. Then there is a possibility of spurious triggering due to noise. The optimum sensitivity of the counter is largely dependent on the impedance of the input. Because the higher the impedance, the more susceptible the counter is to noise, which can cause the counter to generate false counts. Usually, a reasonable sensitivity is set by a suitable input impedance, so as to avoid false triggering caused by noise as much as possible. In the system, the analog time is measured by converting it into a digital quantity. The quantization step size, ie the least significant bit, represents the smallest time interval that the system can measure. Therefore, LSB can be used to represent the measurement resolution of the TDC system. The concept of RMS resolution is often used when evaluating TDC systems. RMS resolution is calculated by taking multiple measurements for a fixed time. Calculate the mean square error between the measured value and the true value to get the RMS resolution of the system. The schematic diagram of the counter based on this design is shown in Fig. 1 below.

Clock

Start

DQ

Stop

Testing device

Register Fig. 1. Counter schematic

354

F. Du and J. Zhu

As can be seen from Fig. 1, the existing time-to-digital conversion technology is basically based on the “start-stop” model. The so-called “start-stop type” has a start signal as the measurement start signal, and a stop signal as the measurement stop signal. Using the timing device in the system, measure the time interval between the start signal and the stop signal. In contrast to “start-stop” measurement methods are pipelined and data-driven measurement techniques. The measurement resolution of the direct counting method directly depends on the frequency of the reference clock. To achieve higher resolution, it is necessary to increase the frequency of the reference clock. As a method of precise time measurement, it is necessary to achieve ns-level resolution. This requires the reference clock frequency to reach the order of GHz, which is difficult to achieve in the current electronics field. But this does not mean that the direct counting method has lost its use value. This method can achieve a large measurement range when the number of counters is sufficient. So this method is usually used in large-scale testing. Early time interpolation methods used tapped coaxial cables as delay elements, but the cables were bulkier. And the measurement consistency is poor, and then gradually eliminated. Due to developments in the semiconductor industry, tapped delay lines can be formed using basic CMOS gates as interpolating cells. Using CMOS gates as delay elements can reduce the complexity of the design and achieve a good level of integration. 2.2 Storage There is a lot of educational information within the Mental Health Distance Counseling Education System. Therefore, the system’s ability to access and call data determines the performance of the system. Therefore, the relevant memory is designed, and the architecture overview is shown in Fig. 2 below.

SRAM DARM

Storage Class Memory NAND

Fig. 2. Memory architecture

As can be seen from Fig. 2, in order to break this bottleneck in the von Neumann architecture, an integrated storage-computing architecture is proposed. This pyramid level memory structure. Each memory device in this structure deals only with its neighbor storage devices. A cache-based memory hierarchy works because slower storage devices are cheaper than faster ones, and because programs tend to show locality. The underlying

Mobile Terminal-Based Remote Counseling Education System

355

storage device has a block concept as the basic unit of reading. Usually a block contains multiple data, and later access to other objects in the block hits the cache due to spatial locality, making up for the first access block copy. Caching we use SRAM chips. The cached SRAM circuit is simple, so access is very fast. DRAM main memory chips are denser, have larger capacity and are cheaper than SRAM chips. SCM (Storage-ClassMemory) is literally defined as storage level memory, which has the advantages of both memory and storage. NAND flash memory is a non-volatile storage technology, that is, it can save data after power off. Its goal is to reduce the cost of storage per bit and increase storage capacity. This system architecture not only retains the storage and read/write functions of the storage circuit itself, but also supports different logic or multiplication and addition operations. Thus, frequent bus interactions between the central processing unit and the memory circuits are largely reduced. It also further reduces a large amount of data movement and improves the energy efficiency of the system. In the deep neural network processor based on the memory-computing integrated architecture. The weight data can be directly processed by MAC operation without reading, and the final multiplication and addition result can be directly obtained. So the throughput of the system will no longer be limited by the limited memory read interface. At present, there are two main types of in-memory computing design schemes: one is the in-memory computing design based on random access memory. The other is an in-memory computing design based on non-volatile memory. These two types of inmemory computing designs adopt the design scheme of analog-digital mixed circuit to reduce the energy and area consumption of the system. However, these two types of designs still have many design challenges, including the improvement of operation accuracy. The reduction of area consumption and the non-linearity and operation errors caused by changes in process parameters. The challenge facing in-memory computing is the limited bus bandwidth for the interaction between CPU and Memory. On the one hand, the in-memory computing design retains the storage and read-write functions of the original Memory circuit. On the other hand, in-memory computing designs can work in compute mode for Boolean logic or multiply-add operations. Early researches related to in-memory computing in a broad sense were mainly used with new non-volatile memory devices to realize simple gate logic operations. Also called non-volatile logic circuit. Several typical non-volatile memory devices include: memristor random access memory, spin torque magnetic random access memory, phase change random access memory and ferroelectric random access memory. In the related non-volatile logic circuit design, general logic operations, such as AND or non-basic gate circuits, etc., can be performed directly on the storage element. In traditional SoCs, complex interfaces are required to connect volatile random access memory and nonvolatile memory. Due to limited bus bandwidth constraints and the pyramid level of Memory. Data cannot be transferred in a massively parallel manner, which slows down the transfer. Moreover, the two groups of memory circuit units will occupy a larger chip area. therefore. When the system restarts from standby, data can be transferred and processed faster, resulting in higher energy efficiency. In the design of a deep neural network processor, whether it is a convolutional neural network layer or a fully connected neural network layer, a large number of multiplication and addition operations are required. At present, a lot of

356

F. Du and J. Zhu

research is still on optimization based on von Neumann architecture, such as multi-core and multi-thread operation, and high-bandwidth on-chip storage technology.

3 Software Design 3.1 Designing an Educational Framework for Distance Counseling for Mental Health The mental health system for middle school students is aimed at students in secondary vocational schools, aiming to build a comprehensive platform for students to learn mental health knowledge, understand their own mental health status, and consult psychological problems. Use the platform to prevent and solve mental health problems in middle school students. So as to help students develop good psychological quality, enhance psychological tolerance and self-solve psychological problems. To better provide strong support for the study and life of middle school students. According to the actual needs of middle school students’ mental health education work, the demand analysis of middle school students’ mental health system is carried out. The system is an important platform for students to learn mental health knowledge, and professional mental health learning knowledge is published online. To enable students to correctly deal with psychological problems in the learning process, to establish good habits and mentality. Parents of students can also browse the website to learn about students’ mental health. Help parents master the methods to guide their children to overcome psychological problems. Provide students with a database query function of mental health-related information. Teachers complete the related operations of publishing knowledge and modifying knowledge on the teacher page. Students can keep up to date with the latest knowledge on the browser. When students select psychological knowledge on the student interface, they can see the psychological knowledge sent by the teacher. Psychological knowledge includes title, content and release time, when the number of psychological knowledge records is automatically paginated. In this function, teachers can add, modify, and delete knowledge. The system adopts the B/S three-tier architecture mode, and the client browser accepts the user’s request and sends it to the application service. The application service obtains data from the database service. The App Service computes the data and submits the result to the client. The client’s browser displays the result in the form of a web page. The B/S three-tier architecture model is adopted, which can reduce the work recapture of the client, and does not have high requirements on the client. As long as the browser is installed, it is easy to implement, and the management and maintenance of the system are relatively simple. As long as it can be done on the server side, the architecture of the system is shown in Fig. 3 below. As can be seen from Fig. 3, the user layer: a user interface is provided to the user through the browser, and the user sends a request to the network server through the browser to perform related operations. Business logic layer: that is, the layer that implements functions, and is the core layer on the Web server side in the entire system. Its task is to accept user requests. If data access is required, the Web server sends SQL commands to the database server to apply for data processing. When the database server completes the data processing, it will return the result data to the Web server. The web

Mobile Terminal-Based Remote Counseling Education System

User layer

Login interface

Backstage classroom module Business logic layer

357

Front desk student module

Web server

Real time data access interface Data layer Database

Fig. 3. Mental health remote counseling system architecture

server performs logical processing on the data to generate the final web page information, and transmits it back to the client browser through the network. Data layer: This layer is mainly implemented by the database server, which performs logical processing services for system data. This layer mainly receives various data requests from the Web server. The user data request sent by the Web server is realized by querying, inserting, modifying, updating, deleting and other functions of the database. 3.2 Construction of a Remote Counseling Education Model for Middle School Students’ Mental Health Based on Mobile Terminals A mobile terminal or mobile communication terminal refers to a computer device that can be used on the move. Broadly speaking, it includes mobile phones, notebooks, tablet computers, POS machines and even car computers. But in most cases, it refers to mobile phones or smartphones and tablets with multiple application functions. With the development of network and technology in the direction of more and more broadband, the mobile communication industry will move towards the real mobile information age. On the other hand, with the rapid development of integrated circuit technology, the processing capability of the mobile terminal has already possessed a powerful processing capability. Mobile terminals are changing from a simple call tool to a comprehensive information processing platform. This also adds a broader development space to the mobile terminal. Therefore, the system designed in this paper is based on the mobile terminal, and constructs a remote counseling model for middle school students’ mental health. First, it is necessary to calculate the relevant indicators of the remote tutoring model, as shown in the following formula (1). √ 2 r (1) k= g In formula (1), r represents the mobile terminal coefficient, and g represents the error factor. In order to facilitate the use of the system, a unified system home page is provided.

358

F. Du and J. Zhu

The home page provides system descriptions, and users enter the corresponding permission interface through identity verification. In order to ensure the protection of students’ privacy and to record students’ psychological files, students are required to log in before conducting psychological assessments. The student account adopts the student number, which can be easily remembered and ensure the authenticity and uniqueness of the user. At the same time, it is also convenient to retrieve the password when the password is forgotten. Teacher accounts can be divided into two types: “Teacher” and “Administrator”. Users in the “Administrator” role can perform system configuration and add and modify the evaluation database, which is a role for system management. By teachers familiar with mental health education and system management. The role of “teacher” is responsible for psychological counseling, psychological question answering, designated psychological assessment and management of students’ psychological files, and is held by professional mental health teachers. The mental health remote counseling model constructed based on this is shown in formula (2). S=

√ r C g k2

(2)

In formula (2), C represents the teaching guidance coefficient. This system uses the more commonly used mental health scales to assess the health of students with related psychological symptoms. Scales vary according to the factors of concern and target groups, and there are many types of common scales. Considering the flexibility and scalability of the system, the target system supports adding and updating of scales. In order to reduce the workload, the management authority of the scale is assigned to the “administrator” role for unified management. Some of the scales in the system are aimed at all students, while others are only aimed at students with special symptoms. The use of the scale is assigned to professional psychology teachers to specify students’ use of the scale, and the system generates results after completing the assessment. 3.3 Design a Functional Module for Mental Health Remote Counseling The system is divided into two modules: student module and teacher module. The student module provides students with the main functions of psychological knowledge learning, psychological consultation, psychological evaluation and forum communication. The teacher module can use the roles of “teacher” and “administrator” for website management, mental health knowledge management, and psychological assessment management. Personal information management module: The student’s account information is uniformly assigned by the administrator role. When students need to modify personal information such as password and contact information, they can modify the settings through this function. Psychological knowledge learning module: The home page of the system provides graphic and video materials for mental health knowledge learning. At the same time, a “keyword” search function of learning materials is provided, so that students can learn relevant knowledge according to their own needs. Psychological consultation module: Students with psychological confusion can choose online mental health teachers for one-on-one psychological consultation. This ensures the privacy of the consultation. If the mental health teacher is not online, students

Mobile Terminal-Based Remote Counseling Education System

359

can leave a message to tell the teacher their confusion, and the teacher can solve it after the teacher goes online. Online psychological assessment: The role of “teacher” can periodically issue assessment questionnaires. Students choose to answer the questionnaire. When the evaluation is completed, the system can feedback the evaluation results and suggestions according to the factors and scoring standards of the questionnaire. At the same time, the evaluation result information will be formed into a file and stored in the database, and abnormal results will be sent to the “teacher” by a system message, so as to prevent psychological problems in time. Forum communication module: After logging in, students can post their own topics of interest or confusion, and other students can discuss topics of interest to them. Tell your own experience to fellow students who have the same experience to help each other as a team. At the same time, it can also be used as a common forum as a platform for students to communicate. System resource management module (administrator role): This module is used to: add, modify or delete student and teacher user information. Publish pictures, texts and videos of mental health knowledge learning, add and modify psychological assessment questionnaires, and manage system forums. Evaluation and result management module (teacher role): responsible for issuing the evaluation questionnaire and processing the evaluation results. The teacher role can select a questionnaire in the system and issue it to a class, grade or individual, and the results of the questionnaire will be provided to the teacher. There will be special prompts in the system for the evaluation results with problems, so that teachers can deal with them in time. Consultation problem management (teacher role): “Inquiry” is divided into two types: online consultation and message consultation. Teachers can communicate with students one-on-one online through this feature. You can also respond to student inquiries. Psychological file management (teacher role): Teachers can view students’ psychological files through this function to have a comprehensive and overall understanding of specific students. In order to fully and correctly solve the students’ problems. 3.4 Design Aided Teaching Database Database is the basis of information storage and processing of mental health system, and an important part of system development and construction. The design and installation of related databases are inseparable from the development of Web applications. The system database is responsible for organizing a large amount of entity information and connections in the system with a certain data model. In addition to the usual data table design, it also includes user-defined functions, views, and stored procedures. The correct installation of the database is the premise to ensure the smooth running of the web application. The system database is mainly used to realize the following functions: Record management user information of different roles is used for authentication when logging in; the system can store mental health learning materials. Record students’ consultation messages and teachers’ responses. Record forum messages and reply records. Store the psychological assessment questionnaire, and you can add, modify, and delete various information in the questionnaire. Record the question attribution assessment questionnaire and attribution factor information, and the question answer option score is used to calculate the assessment result. Stores a history of student assessments.

360

F. Du and J. Zhu

Among them, mental health assessment is the core of system data design. Mental health assessment mostly consists of a set of relatively independent questionnaires. The questionnaire may be designed to evaluate N aspects (factors) of psychological problems, and the questionnaires belong to these N aspects respectively. The respondents fill in and complete the questionnaire within the specified time, and the total score of the questionnaire belonging to N aspects is the mental health status of the corresponding aspect. The total score of each aspect has an upper and lower limit to judge the degree of mental health in this aspect. Such as: “Symptoms self-rating scale (SCL-90)”, the questionnaire includes depression, anxiety, hostility, somatic, obsessive-compulsive symptoms, interpersonal sensitivity, fear, paranoia, psychosis and other 9 symptom factors. The student user information table is used to store the basic information of students, including: student ID, name, password, gender, date of birth, enrollment time, class, major, place of origin, etc. It is used for system authentication, storage and query of student information, because the two roles of “teacher” and “administrator” in the system are both the responsibility of the teacher. The two roles have common features but different permissions, so the two roles “Teacher” and “Administrator” share this table. This table includes: serial number, password, user type, name, email, QQ number and other fields. Different roles are identified through the “User Type” field when logging in.

4 System Test In order to verify the educational effect of the designed remote counseling education system for middle school students’ mental health. In this paper, a test platform that meets the test requirements is built, and the system test is carried out, as follows. 4.1 Test Preparation Thorough software testing is necessary to ensure the correctness and stability of the system. Software testing is a complex and difficult process that requires a lot of manpower and time. Through testing, it is found that there are many related error processing information in the software, which improves the correctness and reliability of the system, thereby improving the product quality of the software. Software testing is playing an increasingly important role. Effective software testing under limited conditions is a critical issue for system success or failure. To carry out effective system testing, correct methods and adequate preparation should be adopted. The designed system testing environment is shown in Table 1 below. It can be seen from Table 1 that test whether the style of the user interface meets the design requirements, the aesthetics of the page, the correctness of the text, the friendliness of the operation interface and so on. To ensure that the user interface gets the actual results that are consistent with the expected results: user interface testing. Mainly by testing the functions of the user interface to ensure the correctness of user access or browsing functions, and the ease of operation. It specifically includes tests on menus, dialog boxes, and all buttons, text, prompt information, and help information on the form. For example, whether the buttons on the form are aligned, the font size of the characters, the location of the icons, and so on. In this process, in order to save manpower, time

Mobile Terminal-Based Remote Counseling Education System

361

Table 1. System test environment Type

Environment

Server hardware

CPU Intel Core 2.50 GHz

Server operating system

Microsoft Windows

Develop and debug software

Microsoft Visual Studio

Database software

SQL Server

Client software and hardware

Window XP

or hardware resources and improve test efficiency, the concept of automated testing was introduced. The login page of the middle school students’ mental health distance counseling education system is shown in Fig. 4.

Fig. 4. System user login page

Completing the programming according to the design plan and testing. The first is to conduct code testing to test whether the code of each part of the website is normal. Whether the website can be opened normally and whether the browsing speed has achieved the expected effect. The second is to carry out functional testing, and test the functions of each module one by one to see if all the functions are implemented as required. Again, backup and test the database to see if the system can ensure data security. This system adopts the test idea from small to large, and from inside to outside, step by step. The specific process steps of the test are as follows: Unit Testing: Unit testing is the smallest granularity, with the purpose of ensuring that each module or component works correctly. The white box testing method is mainly used, and the main test is whether the internal structure of the unit is correct. Integration testing: Integration testing connects these two tests after unit testing and before system testing. On the basis of unit testing, black-box testing is used to assemble the tested modules for integration testing. The purpose is to check the problems between the modules related to the interface and ensure the overall function between the units. The point of integration testing is to find out the invocation and parameter matching problems between unit

362

F. Du and J. Zhu

modules. Exclude the amplification of errors between modules caused by a unit error, and the mismatch of parameter types and quantities. Verify the role of public variables on each integration test module, etc. System test: System test is a comprehensive test of the system under the actual application of the system running in the required server, software and hardware environment of related software. System testing usually uses a black-box approach to verify that the system works in harmony with the actual environment. System testing can reflect whether the system as a whole is suitable for the design function, and whether it can achieve the system goals in the designed software and hardware environment. 4.2 Test Results and Discussion In the above test environment, design a login test case for system access rights. A test data sample is designed, and the role of “administrator” and “teacher” is used to release information, add questionnaires, and designate assessments to test the designed mental health distance counseling education system. Test the system feedback time when different numbers of users log in, and the test results are shown in Table 2 below. Table 2. System test results Number of logged in users 10

System feedback time

Standard feedback time

Test results

0.51 s

set point Y Get the final association rules

End

Fig. 2. Data mining algorithm flow of online intelligent education

Due to the complexity of iterative conditions of frequent itemsets, this paper uses the sequence model of cyclic neural network to model the characteristics of students’ behavior, and explores the internal relationship between students’ behavior characteristics and comprehensive quality. The sequence model exists not only between the adjacent input layer and the hidden layer, between the hidden layer and the output layer, but also between the hidden layer and the hidden layer in the time dimension. The update value formula of the hidden layer is: η = f (δy +  τ + υ)

(5)

In Eq. (5), η represents the hidden layer state; f represents the activation function; δ represents the hidden layer state weight matrix; y indicates neuronal status;  is the state weight matrix from the hidden layer to the input layer; τ represents the input data at the current time; υ represents the offset term. The weight parameters are updated by time-varying back-propagation algorithm. The weights of the three parameters of the network model are shared in the time dimension, that is, the model parameters at different times are completely consistent, which reduces the training parameters. The core of intelligent learning environment is to use intelligent information technology and virtual environment to assist students’ learning and life, provide different teaching situations, record and track students’ learning process in the whole process, for in-depth

634

Z. Zhang

exploration and analysis of teaching and learning modes, and explore learning paths suitable for different students. The student behavior data sequence is used as the input of the encoder, and then transformed into a series of high-dimensional implicit representations. The potential expression of hidden layer, namely behavior characteristics, is mined through deep RNN, so as to understand students’ behavior. By dividing the grade degree of the explicit learning behavior dimension of online intelligent education data, the corresponding grades of learners are converted, so as to obtain the results of the learning behavior of learners on the network teaching platform, so as to provide reference for improving the learning evaluation of learning users and making corresponding intervention measures for teachers’ teaching decision-making in the later stage. Based on the above process, the research on data mining technology of online intelligent education under collaborative learning is completed.

3 Experimental Analysis 3.1 Experimental Data Set This paper proposes a data mining technology for online intelligent education under collaborative learning. The effectiveness of this technology is verified by experiments. The practical research object selected in this experiment is the course a of China University MOOC network. Course a is one of the teacher education courses on the “China University MOOC” platform, which provides online e-learning services for learners. At present, the methods of learning behavior data collection of learning platform are mainly to track and record learning behavior through computer technology, such as log access, web server and so on. This paper makes use of the background database of the existing network teaching platform and the learners’ learning behavior data recorded in the database. Firstly, this paper divides students’ learning logs into different sessions. If the time interval between two consecutive questions is more than 24h, the record will be divided into two sessions. Therefore, his learning data can be divided into different session records. Obviously, if a conversation contains more records, it reflects that students spend more time studying. In order to collect the learning behavior data on the network teaching platform, when mining the learning behavior, the learning behavior data of learners are mainly counted from three functional modules: curriculum statistics, learning statistics and traffic statistics. Do some simple processing for the collected data, such as data format confusion, inconsistent form, partial data missing and so on. There is confusion in the data structure of students’ data, and there is also a lack of data in the collected data due to factors such as curriculum and students’ suspension. For example, some students’ course scores are failed, some are zero, and there are blank values. In order to ensure the integrity and availability of data, this paper processes the metadata by merging tables, unified types and peer-to-peer comparison of student information, deletes the redundant data and modifies and adds the missing data. The data logs downloaded through the background interface are all Excel forms. The data in the form are large, including useless information on this research, such as basic information such as course number, college, specialty, class, ID number, mobile phone number, etc. Considering that learning behavior data involves the privacy of learners, data desensitization has been anonymized.

Research on Data Mining Technology of Online Intelligent Education

635

3.2 Experimental Setup In order to improve the validity of experimental results, ensure that the control group and the experimental group of the same environment Settings. In this paper, python language and pycharm compilation platform are used as data mining tools to mine and analyze students’ behavior habits and academic performance data, and study the correlation between students’ behavior habits and academic performance. Six servers are used in the whole process of student behavior data analysis, which are virtual machines of VMware virtualization platform in cloud data center. These servers are in the same LAN, which constitutes the environmental foundation of Hadoop cluster. The specific configuration is shown in Table 1. Table 1. Experimental configuration Host name

Memory/G

Hard disk/G

Describe

Master

64

1000

Install Hadoop cluster deployment

Slave A

32

800

Rsyslog and other services

Slave B

32

800

Zookeeper and other services

Mysql

16

900

Database

Front end

32

500

Install NginxRedis

Back-end

32

500

Install redis

As the main operating system, Centos system has a 4-core Intel(R) Xeon(R) E5–2660 v2 2.20 GHz CPU. Three servers are used for big data analysis, one is used as the master node, and the other two are slave nodes. The other one is used as a MySQL database server to store the results of data analysis, and the two deploy front-end applications and back-end services. 3.3 Results and Analysis In order to quantitatively evaluate the performance of the method, 70% of the data sets are randomly selected as the training set and the rest as the test set. This paper adopts two popular evaluation methods, namely, accuracy and recall evaluation experiment. The accuracy is calculated as follows: α=

N1 + N2 M1 + M2

(6)

In formula (6), α represents the accuracy rate; N1 represents the number of actually positive samples in the predicted positive samples; N2 represents the number of actually negative samples in the predicted negative samples; M1 and M2 represent the total number of positive and negative samples. Similarly, the calculation formula of recall rate can be expressed as: β=

N1 N1 + N3

(7)

636

Z. Zhang

In Eq. (7), N3 represents the number of predicted negative samples but actually positive samples.The accuracy rate represents the proportion of students’ scores correctly classified in all sequence behaviors, while the recall rate represents the proportion of students’ scores classified as positive in all actual scores. This experiment compares the output results of educational data mining technology with the results of educational data mining technology based on model driven and homomorphic encryption privacy protection, so as to verify the effectiveness and superiority of online intelligent educational data mining technology proposed in this paper based on collaborative learning. The comparison results of the accuracy of various data mining technologies are shown in Table 2. Table 2. Comparison results of accuracy (%) Number of experiments

The educational data mining technology proposed in this paper

Educational data mining technology based on model driven

Educational data privacy protection based on homomorphic encryption Mining Technology

1

96.07

89.44

88.43

2

94.58

88.68

85.68

3

95.29

87.26

86.29

4

97.66

91.06

84.06

5

96.32

90.39

85.82

6

95.24

88.52

86.64

7

94.81

89.81

85.31

8

95.65

87.95

81.42

9

96.23

85.63

82.85

10

96.12

89.24

85.93

According to the test results in Table 2, the accuracy of the educational data mining technology under collaborative learning proposed in this paper is 95.80%, which is 7% and 10.56% higher than the educational data mining technology based on model driven and homomorphic encryption privacy protection, respectively. Therefore, the data mining technology proposed in this paper has relatively high accuracy, has certain advantages in dealing with various types of educational data, and can analyze the potential relationship between students’ behavior. The comparison results of recall rates of various data mining technologies are shown in Table 3. According to the test results in Table 2, the accuracy of the educational data mining technology under collaborative learning proposed in this paper is 90.69%, which is 9.65% and 8.51% higher than the educational data mining technology based on model driven and homomorphic encryption privacy protection, respectively. From the comparison

Research on Data Mining Technology of Online Intelligent Education

637

Table 3. Comparison results of recall rate (%) Number of experiments

The educational data mining technology proposed in this paper

Educational data mining technology based on model driven

Educational data mining technology based on homomorphic encryption privacy protection

1

91.47

81.52

84.44

2

90.88

78.36

82.68

3

89.99

78.42

83.26

4

88.65

79.83

85.02

5

90.32

82.66

81.31

6

91.09

81.22

82.95

7

92.14

80.05

80.28

8

91.48

83.11

83.64

9

90.52

84.42

78.41

10

90.33

80.85

79.82

results of recall rate, the data mining technology in this paper has better performance, can capture more purpose features on the basis of ordered behavior features, and may achieve better prediction. Based on the experimental results of accuracy and recall, the data mining technology proposed in this paper can more sensitively analyze students’ daily activities and more appropriately reflect their life and learning conditions.

4 Conclusion Big data is gradually penetrating into all aspects of human society, not only changing people’s way of thinking, work and lifestyle, but also changing social productivity and production relations. Educational data mining has become a main direction of big data application in the field of education. In collaborative learning, data mining technology can not only help to improve the school management and decision-making ability, but also predict and warn students’ abnormal behavior and academic achievement by analyzing the data of various application systems of digital campus. This paper proposes a data mining technology for online intelligent education under collaborative learning. Its innovation point is to construct the learner portrait under the cooperative learning from four dimensions and extract the characteristics of educational data. The decision tree method is used to cluster the behavioral characteristics of students, and the online intelligent education data mining algorithm is designed to analyze the correlation between students’ behavioral habits and comprehensive quality. This technology can improve the accuracy and recall of educational data mining analysis, and record the relationship between student behavior sequence and achievement in digital form.

638

Z. Zhang

However, there are still some limitations in this paper. The constructed learner learning behavior data mining model adopts the existing explicit learning behavior data, and does not analyze the implicit learning behavior data. In addition, the weight value can be determined according to the learning behavior of different learners, so as to enhance the internal correlation between learning behavior and achievement.

References 1. Chen, D., Zhan, Y., Yang, B.: Analysis of applications of deep learning in educational big data mining. E-education Research 40(2), 68–76 (2019) 2. Rawlings, D., Winsall, M., Yin, H., et al.: What is a compassionate response in the emergency department? Learner evaluation of an End-of-Life essentials online education module. Emergency Medicine Australasia: EMA 33(6), 983–991 (2021) 3. Pang, J., Sui, M.: Homomorphic encryption privacy protection data efficient intelligent mining simulation. Computer Simulation 36(6), 316–319 (2019) 4. Shen, G.: Application research of big data analysis in college wisdom education. Modern Electronics Technique 42(04), 97–100 (2019) 5. Zurawski, S.A., Pickett, K.A., Widmer, M.: Expanding OT’s role in the mental health treatment of Parkinson’s disease through high-quality online education. Am. J. Occup. Ther. 75(2), 1–11 (2021) 6. Weber, M.M., Larkin, D.J., Patrick, M.: Creating informed consumers of aquatic invasive species management programs through online education for nonprofessionals. Invasive Plant Science and Management 15(1), 41–48 (2022) 7. Networks, J.: Online education services selects juniper networks to elevate student experiences in the digital era. Commswire Magazine: Incisive, Intormed, Independent, Objective 8(11), 13–14 (2022) 8. Li, P., Jia, L., Li, W.: Research on a book recommendation model based on data mining. Journal of Zhejiang University of Technology 47(1), 80–85 (2019) 9. Zhang, P., Ding, L., Jiang, N., et al.: Data mining algorithm of the automation system of the distribution network based on the support-confidence-lift framework and its application. Electrical Measurement & Instrumentation 56(10), 62–68 (2019) 10. Gao, Y.: Homogeneous vulnerability mining simulation of online network heterogeneous fault tolerant data. Computer Simulation 37(3), 377–380 (2020) 11. Mohamed, N., Salama, M.M.A.: Data mining-based cyber-physical attack detection tool for attack-resilient adaptive protective relays. Energies 15(12), 1–15 (2022)

Resource Assisted Online Teaching Model of University Library in the Context of Ecological Civilization Qu Long1(B) and Qiong Hao2 1 Library of Sichuan University of Arts and Sciences, Dazhou 635000, China

[email protected] 2 Department of Mechanics and Electronics, Wuhan Railway Vocational College of Technology,

Wuhan 430205, China

Abstract. As one of the basic educational resources of colleges and universities, university library not only serves the needs of teaching and scientific research of teaching staff, but also an important educational base for cultivating college students. The traditional online teaching mode assisted by university library resources can not meet the needs of current teaching, and can not meet the needs of users for teaching resources at the lowest cost. Therefore, in the context of ecological civilization, A new resource assisted online teaching model of university library is designed. Firstly, the resource assisted online teaching model of university library is designed. Secondly, the resource assisted online teaching principle of university library is optimized in the context of ecological civilization. Finally, the online teaching standard of university library is planned, so as to construct the resource assisted online teaching model of University Library in the context of ecological civilization. The results show that, The designed online teaching model has good teaching effect, effectiveness and certain application value. Keywords: Ecological civilization context · University library · Resources · Auxiliary · Online teaching model

1 Introduction Since the 18th National Congress of the Communist Party of China, China has put forward a series of requirements for the construction of ecological civilization, pointing out the direction and path for building a beautiful China and moving towards a new era of socialist ecological civilization. At the same time, China has issued a series of policies and plans on the construction of ecological civilization, which shows the importance China attaches to the construction of ecological civilization, and also reflects China’s determination to the construction of ecological civilization. The construction of ecological civilization needs the full support of all sectors of our society. As college students, they are also an important supporting force to promote the construction and development of ecological civilization. Improving college students’ awareness of environmental © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 639–653, 2022. https://doi.org/10.1007/978-3-031-21161-4_49

640

Q. Long and Q. Hao

protection is directly related to the practice and implementation of the construction of ecological civilization in our country. Therefore, as one of the basic educational resources of colleges and universities, university library not only serves the needs of teaching and scientific research of teaching staff, but also an important educational base for cultivating college students [1, 2]. The construction of ecological civilization education system in university library needs the assistance of colleges and universities. The concept of ecological civilization is closely combined with the ideological education of college students, so as to promote college students’ correct understanding of ecological concept and transform the correct concept of ecological civilization into consciousness and behavior consciousness. At the same time, it is also the need of service transformation and innovative development of university library. This paper analyzes the current situation of University Libraries participating in the construction of ecological civilization, finds out the source of existing problems, carefully analyzes and studies them, and puts forward useful and feasible suggestions. At the same time, I hope to attract the attention of University Libraries and relevant departments to participate in the construction of ecological civilization, and carry out practical operation, so as to provide some useful reference suggestions for China’s participation in the construction and development of ecological civilization. According to the policy of the 18th National Congress of the Communist Party of China on vigorously developing ecological civilization, this paper proves the necessity and feasibility of China’s University Libraries’ participation in the construction of ecological civilization from the two aspects of the construction and educational functions of university libraries, and also provides a practical reference basis for China’s University Libraries to participate in the construction of ecological civilization. The participation of University Libraries in the construction of ecological civilization is an inevitable requirement of the deteriorating ecological environment and the demand for social ecological information. The participation of University Library in the construction of ecological civilization is a new type of information service, which transforms the original passive service of university library into active service. Literature [3] introduces the current situation of items in the library to readers by sampling the websites and directories of a University Alliance and using the definitions established in the framework. The qualitative data of this study will be presented in tabular form after exploring the framework. This article has implications for the information objects in each academic discipline and the continuous service of makerspace and media center. Literature [4] expounds the invincibility of the library from two aspects: Architectural Design and spatial function design. He believes that the expansion of space resources and the diversification of space functions of public libraries will become the development direction of public libraries, and will reshape the future of public libraries. In order to solve the problem that the traditional online teaching mode assisted by university library resources can no longer meet the needs of current teaching. This study is conducive to improve the embodiment of the educational function of University Library and re recognize the better positioning of the educational function of university library. Construct the theory of ecological civilization of university library. Through the research on the participation of University Libraries in the construction of ecological civilization, this paper enriches the existing content of the theoretical research on the

Resource Assisted Online Teaching Model of University Library

641

ecological civilization of university libraries, and extends the scope of the research on the participation of University Libraries in the construction of ecological civilization. Promote the construction and development of China’s ecological civilization. This paper is to improve college students’ attention to the construction of ecological civilization, enhance their awareness and sense of responsibility, integrate the relevant knowledge of ecological civilization into life, and establish a good ecological civilization society. The participation of University Libraries in the construction of ecological civilization is also conducive to the development of China’s ecological civilization construction.

2 Design of Resource Assisted Online Teaching Model of University Library in the Context of Ecological Civilization 2.1 Designing the Resource Assisted Teaching Mode of University Library The integrated construction of library digital teaching resources with the participation of users in the social network environment refers to that the University Library cooperates with all departments of the University under the guidance of the relevant concepts of social network, makes an overall planning for the digital teaching resources scattered in all departments, and goes deep into the professional teaching and scientific research process of college and department teachers and students through guidance and incentive measures [3, 4], The process of organizing teachers and students to participate in all links of the construction of digital teaching resources, organizing and integrating the obtained resources into the digital resource system of the library. The development of social network provides new opportunities for users to participate in the construction of library digital resources. Users’ participation in the construction of library digital resources has become an important way for University Libraries at home and abroad to build resources and an important means for libraries to improve service quality. The construction diagram of this model is shown in Fig. 1 below.

Resource assisted teaching model in Colleges and Universities

Library resources teaching theory

Library assisted teaching environment

User hierarchy

Construction standard

Fig. 1. Resource assisted teaching mode of University Library

642

Q. Long and Q. Hao

As can be seen from Fig. 1, at present, the research on users’ participation in the construction of Library Digital Resources under the social network environment has accumulated certain achievements, but there is a lack of systematic discussion on the whole process of teaching resource construction, and there is also a lack of in-depth research on all links of integrated construction. These research results have certain significance for guiding teachers and students to participate in the practice of the integration of digital teaching resources [5]. Firstly, this paper defines the integrated construction of library digital teaching resources participated by users under the social network environment, and summarizes the research status from the following three aspects: the research on the construction of digital teaching resources participated by teachers and students under the social network environment, and the theory and practice of users participating in the four construction stages of library digital resources planning, production, selection, organization and integration, Research on the integrated construction of digital teaching resources under the social network environment. Secondly, this paper makes a comprehensive study and Discussion on the theoretical basis, definition of relevant concepts, implementation environment, analysis of advantages and disadvantages, construction principles, construction standards and other factors of users’ participation in the integrated construction of library digital teaching resources under the social network environment. Based on the theoretical analysis of the user level, functional system planning and management mechanism of the model, This paper constructs the sub model hypothesis of user creation resources, user evaluation, library integration and library digital teaching resources service in the process of users participating in the integrated construction of library digital teaching resources under the social network environment. This paper constructs the general model map of users’ participation in the integrated construction of library digital teaching resources, and puts forward five suggestions for Chinese colleges and universities to carry out the integrated construction of users’ participation in library digital teaching resources under the social network environment: formulate overall planning, unify standards and norms, and realize optimal allocation and comprehensive utilization; Optimize the evaluation, review and later performance evaluation standards, and strictly control the quality of resources; Pay attention to the research of management mechanism and promote users to participate in the integrated construction of digital teaching resources; Taking users as the center [6, 7], build a resource construction mode dominated by user needs; Focus on solving the core problems of user participation in resource construction. In the context of ecological civilization, users’ participation in library resource assisted teaching mode contains several different factors, as shown in Fig. 2 below.

Resource Assisted Online Teaching Model of University Library

plan

643

Selection

Resource assisted online teaching factors

production

organization

Fig. 2. Factors of library resources assisted teaching

As can be seen from Fig. 2, planning is to formulate long-term, medium-term and short-term strategic plans for the development of digital resources from a macro and micro perspective, establish a digital resource co construction and sharing system benefiting users at different levels, and a digital information resource policy and construction framework. Applying information resource planning theory to library information resource management will better integrate all kinds of existing information resources in the library and realize the comprehensive planning of information collection, processing, storage, transmission and use. The core of user participation in the integrated construction mode of digital teaching resources is whether it can effectively integrate the existing social network technology, give play to the enthusiasm of user participation, and realize the construction of cross departmental information resources on campus [8]. The integrated layout mode of teaching resources not only considers the subject attribute of resources, user groups and information provided In addition to information service, we should focus on the theoretical basis, implementation environment, construction principles and standards and step-by-step sub model of the model. Moreover, the mode can be dynamically feasible adjusted within a certain period of time according to the change of user needs to ensure the maximum fit with user needs. The user fit formula is as follows (1). √ 2 F (1) G= D In formula (1), F represents user satisfaction value and D represents dynamic adjustment parameters. The mode is an intuitive and concise description of the structure and function of things and a simplified form of theory. Under the social network environment, the integrated construction mode of library digital teaching resources participated by users involves university students, teachers, librarians, academic affairs office, Graduate Office, educational technology center and other user and management subject elements, and reflects many uses such as resource creation, evaluation, integration and service [9]. The construction of model needs to be supported by relevant theories, mainly including constructivism theory, humanistic learning theory and collaborative learning theory.

644

Q. Long and Q. Hao

2.2 The Principle of Optimizing the Resources of University Library and Assisting Online Teaching in the Context of Ecological Civilization According to Article 2 of the general provisions in Chapter I of the regulations for libraries of ordinary colleges and universities issued by the Ministry of education in 2015, the university library is the University’s document and Information Resource Center, an academic institution serving talent training and scientific research, an important part of the University’s information construction and an important base for the construction of campus culture and social culture. Article 3 stipulates that the main functions of the library are educational functions and information service functions. Libraries should give full play to their role in talent training, scientific research, social services and cultural inheritance and innovation. It can be seen that the university library is the document information center of the University, which provides the guarantee of document information resources for scientific research, teaching and talent training. The university library should actively participate in the information construction, campus culture construction and socialization construction of the University. University library not only meets the function of reader information service, but also fully reflects the educational function of university library. Therefore, the university library has the responsibility and obligation to undertake part of the educational functions of the University. Through the participation of the University Library in the development of ecological civilization construction, it not only reflects the educational functions of the university library, but also plays a role of inheritance and innovation for the campus culture and social services of the University. University Libraries’ participation in the construction of ecological civilization may be supported by government funds. With the guarantee of funds, university libraries will better and more deeply and comprehensively participate in the construction of ecological civilization, have more choices and innovations in the ways and forms of participating in the construction of ecological civilization, and then achieve good social results, Therefore, in the context of ecological civilization, to optimize the principle of resource assisted teaching in university libraries, we first need to set up assisted teaching optimization indicators, as shown in the following (2). √ G V −C (2) A= D In formula (2), V represents the standard auxiliary parameter and C represents the actual optimization index. In order to further verify the optimization effect, the index difference is processed at this time, as shown in (3) below. √ G V −C A1 = 1 − (3) D In formula (3), using the above optimization indexes, the resource assisted online teaching can be further optimized. In the past, the construction of teaching resources in Colleges and universities was mainly carried out by libraries, academic affairs offices, educational technology centers and various colleges or departments. This is the traditional mode of teaching resources construction in Colleges and universities in China

Resource Assisted Online Teaching Model of University Library

645

for many years. The integrated construction of digital teaching resources is to integrate the resources of the above departments from the beginning of users’ participation in the creation of resources, which needs to break the original resource construction pattern of “each department in its own way” in the school and cooperate and participate in the integrated construction of resources. In the collaborative environment, the above departments still carry out evaluation, review and other construction work on the resources under their jurisdiction, but they need to strengthen macro coordination at the school level management level. They can set up an information resource construction team under the existing school information construction committee or office, with the curator of the library as the leader and the heads of the academic affairs office, Graduate Office and educational technology center as the deputy leaders, Carry out the construction of digital teaching resources in a unified way. All departments carry out collaborative construction of digital teaching resources in their respective fields according to the overall goal of discipline and specialty teaching of the school. The competent department of the school should introduce the corresponding dynamic mechanism and restraint mechanism in the management, so as to promote users and departments to actively participate in the construction of digital teaching resources. Users’ participation in the integrated construction of library digital teaching resources is a complex system project restricted by many factors. Therefore, in the implementation process, we must follow certain principles to provide the correct direction for the integrated construction of teaching resources. The deepening of the reform of higher education system has prompted China’s colleges and universities to gradually adjust the professional structure. While doing a good job in the construction of basic disciplines, they have gradually formed a multi-disciplinary structure development model with key disciplines as the leader and taking into account emerging disciplines, interdisciplinary disciplines, marginal disciplines and branch disciplines [10]. Discipline resource construction occupies an important position in Colleges and universities in China: first, discipline construction, especially the construction of key disciplines, is very important to meet the large-scale and high-level needs of discipline knowledge, and it is also the main symbol to measure the quality of running colleges and universities and their position at home and abroad; Second, the specific forms of discipline service of University Libraries in China include discipline resource construction, discipline resource navigation, discipline consultation, subject novelty search, learning trend analysis, discipline frontier tracking, etc., and discipline resource construction ranks first, which shows the importance of discipline resource construction in discipline service. Therefore, taking subject construction as the direction is the primary principle for users to participate in the integrated construction of library digital teaching resources. The content of resource integration construction should be inclined to the advantageous and characteristic disciplines of colleges and universities, absorb the digital teaching content in the existing discipline construction of colleges and universities, and bring the digital teaching resource integration service platform formed by users into the discipline professional resource service system. In order to obtain more advantageous teaching resources, the library has built a series of information service systems integrating the construction and sharing of teaching resources, from the original document information resource guarantee system to the

646

Q. Long and Q. Hao

library alliance and the construction of subject information portal in recent years. From the perspective of the existing digital teaching resources, the academic affairs office, graduate school, science department, modern technical education center, library, faculty and students and network resources have their own characteristics and can be complementary. Moreover, in recent years, the trend of digital resource sharing in colleges and universities with knowledge sharing as its prominent feature is becoming increasingly obvious. A representative example is the OPENCOURseware of Massachusetts Institute of Technology, which has become a benchmark for complementary sharing of university teaching resources worldwide [11]. However, users’ participation in the integration construction of library digital teaching resources can realize this construction goal to the greatest extent by mining the resources of users from the beginning of the creation of resources. In addition, the complementary advantages of resources to a large extent is conducive to mining the potential value of resources, greatly improve the output efficiency of university teaching resources. The benefits of users’ participation in the integration of library digital teaching resources under the social network environment come from the resource services provided for users in the process of teaching and scientific research. First of all, according to the principle of cost-effectiveness, the benefit of users’ participation in the integration construction of library digital teaching resources should exceed its cost, otherwise it is inappropriate to carry out resource construction. The benefits of users’ participation in the construction of library digital teaching resources integration depend on the number of users covered by its service platform. Therefore, we should maximize user participation. On the one hand, the more users participate, the more quantity and types of teaching resources users contribute, which can reduce the cost of resource acquisition. On the other hand, the more users participate in the service, the more people are willing to enjoy the service, the more benefits will be generated. Secondly, we should improve the quality of resource service in the resource service link, and take multiple measures to improve the construction efficiency of teaching resources. The ultimate goal of digital teaching resource construction is to better meet the teaching and scientific research needs of teachers and students. Users are the core of the whole resource, and the degree of user participation and cooperation is the decisive factor of user satisfaction. The integration construction of library digital teaching resources with users’ participation should meet the needs of users, especially teachers and students. Moreover, the resource demand of users will become more and more intense with the change of information environment. User-centered is the inexhaustible power to obtain sustainable development of resources. Secondly, the resources of the integration construction of library digital teaching resources that users participate in are entirely based on users. In accordance with the principle of user-centered, the basic service platform of resource integration is constructed to provide diversified information services for colleges and universities, so as to realize the “take for users and use for users” of digital teaching resources. Users to participate in teaching resource allocation is based on the demand of resources between teachers and students, with the highest efficiency and best effect of teaching resources as the direction, focus on adjusting the teaching resources of the user population distribution and resources flow in order to maintain the user higher

Resource Assisted Online Teaching Model of University Library

647

demand of the sustainable development of resources, at the lowest cost to achieve teaching resource requirements of users. In social users to participate in the digital library under the network environment teaching resources integration process, shall, first of all, in both user requirements and the goal of construction of digital teaching resources of colleges and universities, based on the structure planning according to the principle of optimal allocation of teaching resources, planning includes the focus of the teaching resources, scope, type and quantity distribution, etc., To obtain the maximum practical benefits of user participation in the construction of integrated teaching resources. To optimize the allocation of resources, one is to establish the digital teaching resource model that best meets the needs of users; the other is to take into account the advantages and disadvantages of the digital teaching resources generated by users and realize the complementarity of advantages of various types of digital teaching resources. 2.3 Planning the Standards of Online Assisted Instruction in University Libraries Under the social network environment, there are many types and types of resources for users to participate in the integrated construction of library digital teaching resources, so it is necessary to carry out unified standards and norms. At present, the standards related to teaching resources mainly come from research institutions and standards committees, mainly including China’s network education technology standard system and computer management teaching system specification of China’s education information technology standard committee, and the modern distance education resource construction technical specification of the Modern Distance Education Resource Construction Committee of the Ministry of Education, The shareable content object reference model, metadata standard and learning design specification of IMS global learning alliance between the White House Science and technology office and the Department of defense. China’s network education technology standard system is a unified technical standard supporting educational resource sharing, information exchange and system interoperability at the educational application level. It includes the following eight types of standard projects: guidance, learning resources, learners, learning environment, education management information, multimedia teaching environment Virtual experiment and learning tools, electronic textbooks and e-book bags. The standard diagram applicable to the integrated construction of library digital teaching resources participated by users in the social network environment is shown in Fig. 3 below.

Learning resources

Learner class

Learning environment Fig. 3. Standard diagram of online teaching assisted by University Library

648

Q. Long and Q. Hao

As can be seen from Fig. 3, the learning resource class can be used to define the metadata structure of any resource related to the construction of teaching resources, mainly standardize various resources involved in digital teaching resources, and unify the basic attribute structure of resources, such as the name, format, purpose, etc.; CELTS-41 technical specification for the construction of educational resources unifies the development behavior of resource developers, the production requirements of development resources and the functional requirements of management system. For the construction of digital teaching resources, it is mainly stipulated from two aspects: first, from the perspective of users, in order to use these digital teaching resources conveniently, label their attributes, standardize the data type and compilation type of attributes from the perspective of operability, and set some characteristic attributes according to the specific characteristics of different teaching resources; Second, from the perspective of managers, this paper puts forward the architecture and basic functions of the management system for managing digital teaching resources. CELTS-42 metadata application specification of basic education resources provides a teaching resource data model for current higher education in China, and provides a set of Resource Cataloging guidelines, so that users can quickly and effectively retrieve the required resources in the school teaching resource database and widely realize resource sharing. The learner class specifies the syntax and semantics of the learner model. Based on this, the information of users (students and teachers, etc.) of digital teaching resources can be subdivided according to personal information, academic information, management information, relationship information, security information, preference information, performance information and portfolio information model to create a personal learner standard information base, It can be used in the whole process of college education or learning and scientific research. Moreover, the user’s information base supports the migration and sharing of user information between different systems, so as to improve the portability of user information data; Celts-13 participant identifier defines the data type of an identifier, which is used to identify participants in the process of learning, education and training. It can be used to define the data type of the identifier of participants such as individual users and department users in the process of digital teaching resource construction. The learning environment class uses multi sectoral standards to support the communication between all participants in the construction of digital teaching resources, such as cooperative learning and communication between students and cooperative teaching and scientific research communication between teachers. Thus, various participating groups are created, and the cooperation environment, functions and tools required for mutual cooperation are provided. The cooperation spatial data model, cooperation environment data model and cooperation group data model are defined, which can be reused in the form of integration. At the same time, the data model instances are allowed to be exchanged, stored, retrieved, reused and analyzed by other systems.

Resource Assisted Online Teaching Model of University Library

649

2.4 Constructing the Resource Assisted Online Teaching Model of University Library in the Context of Ecological Civilization Under the social network environment, the users who participate in the construction of library digital teaching resources mainly include: individual users, mainly refer to the individuals who develop and use digital teaching resources in Colleges and universities such as students and teachers, mainly for the creation and evaluation of resources; Department users mainly refer to school departments related to the use, management and maintenance of digital teaching resources, such as academic affairs office, Graduate School, Academy of Sciences and modern educational technology center, participate in the creation, evaluation and audit of resources, and assist the library in formulating construction planning and management system; Resource management center mainly refers to the specific functional department of the library responsible for the planning, process control and user management of the whole integrated construction of digital teaching resources. Its work includes: the formulation and implementation of construction planning, including the type, quantity, construction process, user participation mode and construction fund budget of resource construction; Formulation and implementation of management rules and regulations related to construction, including reward and punishment system, personnel performance appraisal, etc.; The development, construction, operation, management and maintenance of the integrated construction platform of teaching resources. The resource assisted online teaching model of University Library Based on this is shown in Fig. 4 below.

Resource creation

Resource evaluation

Resource audit

Resource construction planning and management

Fig. 4. University Library assisted online teaching model

It can be seen from Fig. 4 that under the social network environment, individual users are the most active elements to participate in the integrated construction of digital teaching resources, and their new resources can best meet the needs. In the process of teaching and scientific research, teachers and students in Colleges and universities are accompanied by the production of teaching plans and scientific research resources such as learning materials, courseware and handouts. However, due to the lack of resource construction and protection mechanism, their management is relatively scarce. The survival period of resources is limited to a certain teaching and scientific research period, and their academic scientific research value can not be continuously recycled. Under the social network environment, users participate in the integrated construction mode of library digital teaching resources. Students and teachers can use the social network platform to contribute their own digital teaching resources, evaluate the existing resources

650

Q. Long and Q. Hao

and recommend the teaching resources they are interested in. At the same time, students can carry out cooperative learning, teachers can carry out cooperative teaching and scientific research, teachers and students can carry out interactive communication, integrate the teaching and scientific research process of teachers and students with the resource construction process, and constantly produce valuable teaching resources. At the same time, through the unified integrated management and open access of the library, the resources created by users can be recycled. Department users are the new resources for users to participate in the integrated construction of library digital teaching resources. Under the social network environment, department users who participate in the integrated construction of library digital teaching resources include academic affairs office, Graduate School, modern technology education center, scientific research institute, library and other departments. First of all, departmental users should give full play to their departmental advantages over individual users, actively participate in the production of subject teaching resources and contribute more standardized digital teaching resources. Secondly, department users need to review the resources produced by teachers and students according to their department administrative functions, and hand over the digital teaching resources reviewed by their department to the library for the overall integration of resources. At present, most university departments do not lack the experience of evaluating the resources contributed by individual users, and the resource evaluation of individual users is a long-term professional work. We should not only accurately grasp the social network tools of digital teaching resource evaluation, but also be familiar with the service objects of digital teaching resources. Therefore, selecting staff with professional knowledge and familiar with department business for resource evaluation is the “mainstay” of the integrated construction of digital teaching resources. The creation of teaching resources by users is an important driving force for the integrated construction of resources, which is directly related to the evaluation and communication of resources by the business layer and the service quality of user resources by the application layer. Resource creation tools can use various social software and social networking sites, such as social service sites such as face book and Youtube, and social software such as blog, Wiki and QQ. The resources created not only refer to the resources generated by social network technology, but also include the digital resources generated by users in other ways. User created resources can help the library effectively tap the current and potential demand information of users, reduce the uncertainty in the library application layer service, improve the user acceptance of the service, and truly achieve the library service based on user needs.

3 Case Analysis 3.1 Overview and Preparation In order to verify the teaching effect of the designed University Library Resource assisted online teaching model, this paper selects universities for library resource assisted online teaching. The overview of the selected samples is shown in Table 1 below.

Resource Assisted Online Teaching Model of University Library

651

Table 1. Sample overview Title/Position

Number of people

Coefficient

Research institute

61

0.565

Associate researcher

62

0.466

Staff member

81

0.365

Library assistant

45

0.145

Department director

21

0.569

Working personnel

116

0.449

The samples in Table 1 can be used for subsequent example analysis. As the carrier of digital teaching resources, the university digital teaching resources platform constructs a place for users to access, share and exchange information resources. It is a relatively mature teaching resources management organization mode, which effectively promotes the construction process of digital teaching resources. Moreover, the quality of resource platform construction is an important assessment index of the achievements of teaching resource construction in Colleges and universities. The curriculum center accounts for a large proportion of the existing digital teaching resource platform in Colleges and universities, followed by the teaching resource library, and the quality courses are the least. In addition, there are a small number of commercial teaching platforms such as blackboard or secondary development based on open source network teaching platforms such as Moodle. At this time, the calculation formula of online teaching index can be designed, as shown in (4), below. K = GH \L

(4)

In formula (4), H represents the teaching coefficient and l represents the sum of scores. This formula can be used for subsequent model performance analysis. 3.2 Application Effect and Discussion On the basis of the above preparatory work, the resource assisted online teaching mode of university library, the traditional online teaching mode and the service quality perception and evaluation teaching mode of university library designed in this paper are applied to teaching respectively. The teaching indicators of the three teaching modes are calculated according to formula (4).The application effects are shown in Table 2 below.

652

Q. Long and Q. Hao Table 2. Application effect

Teaching batch

The online teaching model and teaching index designed in this paper

Traditional online teaching model and teaching index

The teaching mode of service quality perception and evaluation in university library

1

0.954

0.541

0.641

2

0.921

0.533

0.543

3

0.923

0.612

0.712

4

0.911

0.635

0.634

5

0.859

0.598

0.515

6

0.903

0.746

0.676

7

0.866

0.558

0.548

It can be seen from Table 2 that the teaching index of the resource assisted online teaching model of university library designed in this paper is high, which proves that the teaching effect of the design method is good and has certain application value.

4 Conclusion The library can make use of its own information resources and environmental advantages to build the library into an ecological civilization education base as a grass-roots environmental public opinion information point and an ecological civilization demonstration window. First, set up an ecological civilization observation room to hire environmental protection experts and well-known scholars to talk with readers about environmental protection hot spots and focus issues, and guide and cultivate the public to participate in environmental protection correctly and orderly by holding seminars, lectures, forums and other means. Second, set up awards to promote the typical experience of environmental protection, excavate the touching deeds and typical examples of social environmental protection organizations and environmental protection volunteers in carrying out environmental protection public welfare activities, inspire and move more people through publicity and promotion, and encourage more people to participate in the construction of environmental protection undertakings. Third, do a good job in environmental research, communicate and cooperate with environmental protection experts, put forward more feasible and practical suggestions to the government, and strengthen the library’s sense of social responsibility. Only when people fully understand and agree with the ways and methods of environmental protection can they spontaneously take environmental protection actions in production and life. In addition, the existing online resource assisted teaching mode of library can not meet the current teaching needs. This paper puts forward the online resource assisted teaching mode of University Library under the background of ecological civilization, establishes a series of information service systems integrating the construction and sharing of teaching resources, carries out the construction of digital teaching resources, and maximizes the participation of users. Therefore, if we

Resource Assisted Online Teaching Model of University Library

653

can spread a large amount of scientific knowledge in advance, change pollution first to prevention, do a good job in the publicity, education and guidance of ecological civilization, and make this work long-term and normalized, there will be a basic guarantee for the implementation of building a beautiful China.

References 1. Coombs, B.: Material diplomacy: a continental manuscript produced for James III, Edinburgh University Library, MS 195. Scott. Hist. Rev. 98(2), 183–213 (2019) 2. Boyce, G., Greenwood, A., Haworth, A., et al.: Visions of value: leading the development of a view of the University Library in the 21st century. J. Acad. Librar. 45(5), 102046 (2019) 3. Simpson, J.: Real world objects: conceptual framework and university library consortium study. J. Acad. Librariansh. 45(4), 332–342 (2019) 4. Wu, J., Cheng, H., Koen, D., et al.: Open, inclusive and sharing: a new example of library space reconstruction in the new era-special interview with library building experts on the opening of the Helsinki central library, Finland. Libr. J. 38, 4–12 (2019) 5. Mance, G., Rodgers, C., Roberts, D., et al.: Deeply rooted: maximizing the strengths of a historically black university and community-based participatory research to understand environmental stressors and trauma among black youth. Am. J. Commun. Psychol. 66(3–4), 256–266 (2020) 6. Wang, J., Yuan, R., Shi, H.: Quantitative representation of perception and evaluation method for service quality in university library under 4-D space. Libr Hi Tech. (2019), ahead-ofprint(ahead-of-print) 7. Ujwary-Gil, A.: Organizational network analysis: a study of a university library from a network efficiency perspective. Libr. Inf. Sci. Res. 41(1), 48–57 (2019) 8. Muthanna, A., Sang, G.: State of university library: challenges and solutions for Yemen. J. Acad. Librariansh. 45(2), 119–125 (2019) 9. Miranda-Valencia, B.L.: Satisfaction and consumption emotions of library users at a public university in Mexico: a case study. Libri 71(2), 109–121 (2021) 10. Fakoya-Michael, S.A., Fakoya, M.B.: Library usage by university accounting students: a comparison of contact and open distance learning institution in South Africa. J. Acad. Librariansh. 46(1), 102093 (2020) 11. Qin, H.J.: The reasonable collocation of digital library resource under cloud computing environment. Comput. Simul. 36(3), 375–378 (2019)

Online Education and Learning Model of Applied Optics Course Based on Artificial Intelligence Yankun Zhen1(B) and Haolin Song2 1 College of Science, Xi’an Shiyou University, Xi’an 710065, China

[email protected] 2 School of Computer Science and Technology, Beijing Institute of Technology,

Beijing 100081, China

Abstract. In the optoelectronics related curriculum system, the applied optics curriculum focuses on improving learners’ ability in practical optical design, while ignoring the online teaching link, so it is difficult to ensure the teaching quality. In order to solve the above problems, a design method of online education and learning model of Applied Optics curriculum based on artificial intelligence is proposed. After determining the design theme, through independent learning and team division, complete relevant data acquisition, optical system optical characteristic calculation, initial structure establishment, evaluation function composition, global optimization process and tolerance analysis, effectively combine artificial intelligence with the learning process, and achieve the learning goal of Applied Optics online teaching. Finally, the experimental results show that the online education and learning model of Applied Optics Based on artificial intelligence is more effective in stimulating learning motivation, improving learning efficiency, enhancing students’ design practice and operation ability, and cultivating team communication and cooperation skills. Keywords: Artificial intelligence · Applied optics · Online education

1 Introduction In recent years, driven by the progress of information technology, with the proposal of emerging online learning concepts such as large-scale open online courses, online education shows a rapid development trend in market scale and the number of learners. Online education has become a hot field in the Internet industry. It not only brings huge business opportunities, but also may lead to a fundamental change in learning mode in the near future [1]. At present, there is no exact definition of the concept of social media, but what is clear is that it has the characteristics of decentralization. People can use it to create content, share information, and get feedback from others. Most of the online learners are on-the-job adult learners, Therefore, online education not only pays attention to the transmission of knowledge, but also emphasizes the improvement of learners’ ability to © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 654–667, 2022. https://doi.org/10.1007/978-3-031-21161-4_50

Online Education and Learning Model of Applied Optics Course

655

analyze and solve practical problems. The cultivation of practical skills requires learners to participate in practice and learn in practical activities [2]. At the same time, the new online education model emerging in recent years increasingly encourages learners to learn in a collaborative way, promotes the internalization of knowledge and speeds up the mastery of skills through communication with others (other learners or teachers). Social media provides an effective way for collaborative learning in the context of online education. Using social media, learners and teachers, learners and other learners can communicate synchronously or asynchronously. This efficient communication method can improve the learning effect of online education. Aiming at the learning problems in Applied Optics, combined with the characteristics of artificial intelligence, this paper introduces the application of artificial intelligence in online teaching of Applied Optics, and describes the feedback and response of learning model. Combined with the practical characteristics of Applied Optics online teaching, artificial intelligence mode is applied to teaching design as an important learning organization form in different stages of the course.

2 Online Education and Learning Model of Applied Optics Course 2.1 Online Education Management System of Applied Optics Course The biggest difference between online learning based on social media and traditional classroom learning is the difference in the way of knowledge transmission. Classroom learning is a teacher-centered teaching method, and knowledge is mainly one-way transmission from teachers to students [3]. In online learning based on social media, knowledge can not only come from teachers, but also from learners, and be transmitted among all participants of learning activities. The figure shows the difference of knowledge transmission between the two different teaching methods (Fig. 1). student

learner

student

learner

student

learner

student

learner

student

learner

student

learner

teacher

Fig. 1. Knowledge transfer and blending mode of classroom learning and online learning

Considering the learning background and course characteristics, the learning contents will include preliminary aberration theory, application of software tools and practical optical design. The preliminary aberration theory will focus on the study of

656

Y. Zhen and H. Song

monochromatic aberration and chromatic aberration [4]. The learning content of actual optical design will focus on the experience of optical design process, verify theoretical knowledge through software operation, and guide reasonable software operation with theoretical knowledge. Different from traditional classroom education, online education based on network not only needs to impart knowledge and supervise learning, but also needs to provide online teaching resources and environment for learners. Learners need to think independently and explore actively, which requires higher self-consciousness and initiative [5]. With the continuous development of online education and the proposal of personalized learning, there are new requirements for the content and form of learning resources. It is necessary to organize learning resources according to certain norms and standards to make them have structural characteristics, so as to facilitate learners to query, recommend and share learning contents in a wide range [6]. Learning resources refer to all resources that learners can use for learning in the learning environment of teaching system and online education, including teacher guidance, resource information, material content, multimedia equipment and technology. Through the standardized organization and modeling of learning resources, it can meet the new needs of learners for dynamic and retrievable learning resources (Fig. 2).

Metadata, integration model

content

explain

practice

activity

Generative information

evaluate

tool

Learning tools

Learning circle

learning activities

Learning content

Learning assessment

Learning records

Semantic association

Fig. 2. Elements of online learning elements of Applied Optics Course

Among them, metadata and aggregation models describe the characteristics of learning elements, which are convenient to organize and search for content. Learners mainly learn by obtaining resource content. Practice and evaluation are used to test learners’ acquisition and mastery of knowledge. Activity description Standard code of Conduct and operation mechanism of learning Activities Service interface Provides a functional interface for information exchange between learning element and learning service artificial intelligence environment, and provides data information for learning tools, content, activities, exercises and other modules.

Online Education and Learning Model of Applied Optics Course

657

2.2 Online Learning Evaluation Index of Applied Optics Course In order to demonstrate students’ problems professionally and typically, ask questions within the limits of teachers, and each group selects a deviation as the problem to be understood and solved. At this stage, the team members are required to collect data and learn the basic function operation videos of the software by themselves, and regularly organize group discussions and share learning experience to prepare for future classroom demonstrations [7]. At this stage, teachers will guide students to master relevant knowledge, discuss core issues, and timely feed back problems in students’ learning process. At the same time, teachers should emphasize and guide students not to limit their learning content to the distortion of their own group, and learn all distortions comprehensively (Table 1). Table 1. AOC learning process based on CPBL model The preliminary stage

The intermediate stage

The advanced stage

Learning focus

Primary aberration theory Basic functions of optical design software

Aberration correction strategies

Complex optical system design

Group formation

Students’ self-organization

Based on the previous grouping, one person from each group is selected to form a new group

The self-recommendation of the group leader and the group members

Problem arises

Scope limited by the teacher Group self-selection

Taking a simple optical system as the object, determine the optical characteristic parameters by yourself

Find and select optical design problems completely independently

Learning resources Textbook Literature Video library

Software manual Literature Patent library Website videos

Software manual Literature Patent library Website videos

Activity venue

Smart classroom Group choice

Computer room Group choice

Computer room Group choice

Learning methods

Traditional teaching, Personal learning, Group learning

Personal learning Peer learning Group learning Online and offline communication between teachers and students

Personal learning Peer learning Group learning Online and offline communication between teachers and students

Teacher role

Gradual weakening from Supporting role the leading role to a supporting role

Supporting role

658

Y. Zhen and H. Song

In the middle stage of the learning process, one student is randomly selected from each group to form a new group on the basis of the previous group, which is conducive to the diversity of aberration control methods mastered by group members. As group members come from different aberration groups, they have different degrees of understanding of different aberrations. Teachers should weaken the role of experts, only play an auxiliary role, actively respond to questions from the students attempt to encourage students to make a different, rather than the default design and operation process, let the team members interact each other and generate dynamic learning resources, and share knowledge and skills, so that the students have more channels to solve specific practical problems, Thus, it is more likely to achieve the best design effect. After the design theme is determined, through independent learning and team division, relevant data collection, calculation of optical characteristics of the optical system, initial structure establishment, evaluation function composition, global optimization process and tolerance analysis are completed, and finally the design objectives are completed. In this stage, the teacher only plays a supporting role, answering the questions raised by the students in time and giving guidance to the mistakes in time, but the focus of this stage is to emphasize the subjectivity of students’ independent learning. In the three stages of the learning process, the learning results will be displayed and evaluated in different ways according to the characteristics of the learning stages and the difficulty of the problems, as shown in Table 2. Table 2. Evaluation indicators of learning results The preliminary stage

The intermediate stage

The advanced stage

Learning outcome exhibit

Group class presentation Questions from other groups Teacher’s summary and supplement

The competition between groups The sharing of Paperwork and operating procedure video

Defense group display Adversarial group questioning Paper report

The role of learning outcome evaluation

Teacher

45%

Teacher

Teacher

45%

Group members

35%

Group members

25%

Other groups

10%

Self-evaluation

10%

Group members

75% 25%

Other groups

25%

Self-evaluation

5%

Multi channel process evaluation and conclusion evaluation are combined to collect learning results. Among them, the subject of learning effect evaluation can be carried out not only by teachers, but also by students in the form of peer evaluation system. In the traditional collaborative recommendation algorithm, the similarity between two users is difficult to calculate because of the lack of common scoring items. However, when the user attributes are similar or even the same, their scores on the same items may also be very close. Therefore, the combination of Pearson similarity and user attribute

Online Education and Learning Model of Applied Optics Course

659

similarity is used to measure the similarity of users, so as to alleviate the problem of too sparse scoring matrix (Fig. 3).

N Algorith m start

Data preprocessing

Scoring matrix coefficient

Generate / fill user item scoring matrix

Collaborative filtering recommendation based on modified similarity measure

End of algorithm

Y Predicted scoring of non scored items

Build user attribute preference model

BP neural network training

Generate user attribute item attribute scoring matrix

Generate user item attribute scoring matrix

Fig. 3. Online teaching collaborative recommendation algorithm based on Artificial Intelligence

Since the development of students’ learning process is related to the design of learning situations and learning strategies, the learning background of this course emphasizes the application of theoretical knowledge to practice; Students combine teacher-student interaction and student cooperation throughout the learning process. In order to effectively combine artificial intelligence with learning process and realize the learning goal of Applied Optics online teaching, the whole learning process mainly includes three learning stages: primary stage, intermediate stage and advanced stage. The framework design and development of online curriculum resources is an important link in the success of the implementation of online teaching mode. It can also be said that the important factor of teaching success or failure is determined by curriculum resources. According to the different functions and uses of information-based teaching resources, our university classifies online course resources into three categories: guiding resources, content-based resources and generative resources. The overall framework design is as follows (Fig. 4):

Guiding resources

Content resources

Generative Resources

Curriculum teaching teacher

Online Q & A after class

Classroom interactive communication

Assignments and exams

Course live broadcast

Self reflection

Course supporting materials

Complete the task

Syllabus content design

Fig. 4. Online education curriculum content structure optimization framework

660

Y. Zhen and H. Song

This stage is the central link of the whole online teaching stage. The quality of classroom teaching directly affects the implementation of the whole teaching mode, and content resources are the key to determine the success or failure of classroom teaching stage. Unlike ordinary online teaching, which uses recorded video courses as course resources, our teaching adopts the way of live video. The so-called teaching behavior should be a real-time process completed by teachers and students. There is a time difference between ordinary video teachers and students when they record it and watch it. The time difference will lead to further estrangement between teachers and students, It is not conducive to the development of Teaching. At this stage, teachers’ direct teaching is still the most efficient and direct teaching method. In the process of activities, people prefer to face a living person rather than a cold machine. One advantage of online classroom is that it can carry out more communication between teachers and students. In order to ensure the classroom quality of teaching, students can be arranged to speak on behalf of students after group discussion for many times in a semester, and the students who speak can be scored. When settling the classroom academic results at the end of the term, the system calculates the students who have not spoken or have not spoken enough times, and the score of the project will be discounted or even failed. At the same time, the attendance rate of students can be recorded through the student side camera system. After students log in to the system through their account, the system can monitor students’ performance in class. If students leave for no reason, those who leave for more times can be judged as late, early leave or even absenteeism, which is more effective and accurate than taking time to roll call. Du Judai’s attendance also saves time. 2.3 Realization of Online Teaching and Learning of Applied Optics Course With the explosive growth of information resources, it has become difficult for ordinary users to find the content they need from a large number of resources. As information providers, how to make their product information stand out from the multifarious information base and get attention has also become very difficult, hence the emergence of recommendation system. Recommendation system should be based on user needs, interests, preferences, and recommended that may be of interest for the user resources, and provide personalized, differentiated content services for the general information collected by the user personalized recommendation system and modeling module, user behavior records attribute, the recommendation algorithm module and content display interface of these four parts, as shown (Fig. 5). The user information collection and behavior record module is responsible for obtaining user attributes, scoring information and behavior records from the database and web log, so as to provide data sources for the implementation of recommendation algorithm; User attribute modeling is responsible for analyzing user attributes and their behaviors, extracting effective information, and establishing user attribute preference model: recommendation algorithm is the core module of recommendation system, which needs to generate recommendation results, recommend the content in the resource database to users, meet the personalized needs of users, and the content display interface displays the resource information to users, and in the process of user browsing, Continuously collect user feedback, collect user behavior information, update user needs, and improve system interactivity and real-time. The formulaic definition of recommendation system

Online Education and Learning Model of Applied Optics Course

661

Resource pool

Content display interface

Resource pool Recommendation algorithm User attribute modeling Information collection and behavior record

User feedback

Fig. 5. Online teaching recommendation system of Applied Optics Course

is as follows: in Recommendation System. Let C represent the set of all registered users and A represent the set of all online products. In the actual online system, the number and scale of users and products are usually large, so that the utility function u(x) calculates the evaluation of user C on product s, that is, the liking u(c, s), where R is a totally ordered set. For each user c ∈ C, the recommendation system can help them find the products with the greatest preference, that is: ∀c ∈ C, ∃s ∈ A, s = u(x) − A arg max C − u(c, s)

(1)

At the same time, users can also contain a variety of user information, such as age, gender, specialty and so on. Product s also contains a variety of item information. If it is on the shopping website, it can include the unit price, style, origin and so on. The utility function represents the user’s evaluation and preference for the recommended items. Like traditional education, online education also needs a course route process. Teachers also need to prepare lessons before teaching, determine the syllabus and design the course content. On this basis, the system has a set of unique route modules to assist teachers in their course work, as shown in the following figure (Fig. 6): Preparation before class needs teachers according to the characteristics of the online mode, to determine the content of the syllabus and curriculum, not only is the original PPT courseware to pass on the server, such as over it, and the difference from the online education and traditional education teaching ways, than the general existing online education simply video video lectures, biggest advantage of this system can be taught to broadcast live. The resource display and learning module shows learners the recommended methods to realize the learning resources recommended by the module, and provides supporting functions such as learning browsing. Meanwhile, real-time acquisition of learners’ behavior information is carried out. The main users are students and teaching students who can study on the platform according to their own interests and personalized recommendation results. After learning, they can evaluate and discuss, etc. Teachers can upload learning resources, guide students in learning and manage resources and other information. Administrators can perform routine maintenance of the system (Fig. 7).

662

Y. Zhen and H. Song Course teaching video Course supporting materials Course content design

Teaching program

Course evaluation and summary

Stage test questions Live course arrangement Examination questions

Fig. 6. Online education course route design

Information modification Personal Center Password modification Course on demand Curriculum evaluation Course learning Learning plan Learning records Resource Recommendation Job resources interest group Learning platform

Learning support

Task reminder Online Q & A Learning Forum Course recommendation

Course selection management

Student course selection Course selection query Score query mail

Feedback Feedback mailbox

Fig. 7. Function module diagram of learning platform

Based on the learning problems in Applied Optics and the characteristics of artificial intelligence, the application of artificial intelligence in online teaching of Applied Optics is introduced, and the feedback and response of learning model are described. Considering the practical characteristics of Applied Optics online teaching, artificial

Online Education and Learning Model of Applied Optics Course

663

intelligence mode is applied in teaching design as an important learning organization form in different stages of the course. In the teaching design, students are gradually familiar with the artificial intelligence mode, give full play to the advantages of the artificial intelligence mode, stimulate students’ motivation for autonomous learning, realize the flexible combination of curriculum theory and design practice in the learning process, exercise students’ teamwork ability, and finally effectively complete the goal of building curriculum knowledge points based on the network.

3 Analysis of Experimental Results Function test the personalized resource recommendation module and related modules in the network education and learning platform to verify whether their functions meet the system requirements and design test objects: test the functional modules such as system resource recommendation, resource evaluation and learning support, and evaluate the correctness, reliability and integrity of the functions, And test the interaction and complementarity between the personalized resource recommendation module and the existing functions of the learning platform (Table 3). Table 3. Deployment and test equipment parameters Equipment

Model

CPU

Memory/GB

Operating system

Network broadband/M

Quantity

Web server G2153MT

Inter(R)Core(TM)[email protected] GHz

32

Windows XP

120

2

Database server

G2155MT

Inter(R)Core(TM)[email protected] GHz

16

Windows XP

120

2

Web client

G2157MT

Inter(R)Core(TM)[email protected] GHz

32

Windows XP

120

2

Convergence validity is mainly used to detect the correlation between different problem items in the same construct. It is generally believed that if the standardized factor load λi is greater than 0.6, the combined reliability is greater than 0.7 and the average extraction variance is greater than 0.5, the internal convergence of the model meets the requirements. The calculation formulas of 19.1% combined reliability and average extraction variance are as follows, where is the standardized factor load and θi is the measurement error of observation variables. CR = s − AVE = 

(λi − u(x))2 (λi )2 + (θi )

−1

λ2i  − s u(x) λi 2 + (θi )

(2) (3)

After analyzing and calculating the measurement model of the initial model with amos 17, the test results related to convergence validity are shown in the table (Table 4).

664

Y. Zhen and H. Song Table 4. Convergence validity test results

Variable

Measurement item

Normalized factor load

Composite reliability

Mean extraction variance

Academic integration

A1

0.768

0.865

0.562

A2

0.795

A3

0.698

A4

0.765

A5

0.653

B1

0.912

0.965

0865

B2

0.914

B3

0.902

C1

0.856

0.889

0.754

Perceived usefulness (PU)

Expected confirmation (EC)

The standardized factor loads of the measurement items were all above 0.6 and passed the significance test, indicating that each measurement item had a strong explanatory ability to the underlying variables to which it belonged. The combined reliability of the six latent variables was also higher than the recommended value of 0.7, indicating that the measurement items in each group had high internal consistency. AVE all meet the standard greater than 0.5, indicating that the measurement items can reflect the characteristics of latent variables. Therefore, on the whole, the measurement model has good convergence validity. By integrating the artificial intelligence model into the course teaching process, the theoretical knowledge module has problems and is specific. Students establish a close relationship between theory and practice by analyzing and solving problems. The success and failure of the design results are precious to students, so the AI model improves students’ interest in learning and increases their motivation to learn. Figure 8 shows the results of the student survey on the degree of recognition of artificial intelligence for motivating learning. Set the experimental environment to intert (R) core (TM) [email protected] GHZ, the memory is 8.0 GB, the operating system is windows7, and the implementation of data preprocessing and recommended algorithm adopts pycharm2017 1 and MATLAB r2017b experimental data sets are movielens data sets. This data set is collected and founded by the GroupLens project team of Minnesota University in the United States. It can receive users’ scores on movies and provide personalized information recommendations. 3 among them, each evaluation score ranges from 1 to 5. Each user provides its age, gender, occupation and other attribute information when registering, and each movie provides its title, release date Subject type and other characteristic information. In this paper, movielens IM data set is selected and the evaluation diversity is divided into five subsets. The subsets do not intersect. Each subset contains 80.000 scored base data set and 20000 scored test data set. The base data set and test data set complement each other. The sparsity of the base data set is 95%, which is a typical sparse matrix,

Online Education and Learning Model of Applied Optics Course

665

Cognitive level 50%

40%

30%

20%

10%

0%

Uncertain

Very agree

agree

Fig. 8. Survey results of students’ cognitive level

which can help verify the improvement of the algorithm in solving the problem of sparse scoring matrix, The number of neighbor user sets is 10, 20, 30, 40 and 50 respectively. The traditional user based collaborative recommendation algorithm and the improved method proposed in this paper are used for comparative experiments. The average absolute error between the prediction score results and the test data set is compared, and the results are shown in the figure (Fig. 9). 0.94 User-based CF

MAI

Proposed CF

0.88

0.82

0.76

0.70

5

15

25

35

45

Number of neighbor users

Fig. 9. Comparison of average absolute error of teaching information recommendation

As shown in the above figure, when the number of neighbor users is 40, the minimum average absolute error of the proposed method is 0.77, while the minimum average absolute error of the traditional user based collaborative recommendation algorithm is 0.81. Compared with the traditional method, this method has smaller average absolute error

666

Y. Zhen and H. Song

and better recommendation effect. The improved algorithm simulates the preference model of user attributes based on the dual attribute scoring matrix and adopts artificial intelligence technology to reasonably predict the scoring of non scored items, which alleviates the problem of inaccurate recommendation caused by too sparse scoring data in the traditional algorithm. In order to verify the effectiveness of this design method, simulation experiments are carried out to test the performance of this design method, and the accuracy of teaching information recommendation is compared with that of literature [2] and literature [3] algorithms. The specific results are shown in Fig. 10. 1.00 Literature [2] method Literature [3] method Proposed method

Accuracy

0.90

0.80

0.70

0.60

5

15

25

35

45

Number of neighbor users

Fig. 10. Accuracy comparison of teaching information recommendation

As shown in the figure above, the recommended accuracy of the proposed method is about 0.90, while the recommended accuracy of the two literature methods is lower than 0.90. Compared with the two literature methods, the proposed method has higher recommended accuracy and better recommended effect. This is because the proposed method emphasizes the subjectivity of students’ autonomous learning and combines multi-channel process evaluation with conclusion evaluation to collect learning results.

4 Conclusion Through the effective integration of artificial intelligence and applied optics online teaching learning process, teachers as coaches gradually guide the learning process, and students experience the learning process by putting forward and solving problems in computer software theoretical learning and practical operation. In the whole learning process, students’ learning objectives are clear, and they are motivated to study independently. At the same time, students are encouraged to study independently. Pearson similarity and user attribute similarity are combined to measure the similarity of users, so as to alleviate the problem that the scoring matrix is too sparse. In the teaching design, students are gradually familiar with the artificial intelligence mode, give full

Online Education and Learning Model of Applied Optics Course

667

play to the advantages of the artificial intelligence mode, stimulate students’ motivation for autonomous learning, and realize the flexible combination of curriculum theory and design practice in the learning process. These practices have proved that the artificial intelligence model has advantages in cultivating students’ self responsibility ability and excavating students’ autonomous learning motivation, and has a certain reference significance for the development of educational reform. Fund Project. 2017 “higher education teaching method overseas study program” of the national study abroad fund committee.

References 1. Hai, J., Xie, G., Wen, J., et al.: Implementation of multi-agent ring hunting simulation based on processing. Comput. Simul. 37(4), 7 (2020) 2. Licato, J., Zhang, Z.: Correction to: evaluating representational systems in artificial intelligence. Artif. Intell. Rev. 52(4), 1 (2019) 3. Benterki, A., Boukhnifer, M., Judalet, V., Maaoui, V.: Artificial Intelligence for vehicle behavior anticipation: hybrid approach based on maneuver classification and trajectory prediction. IEEE Access PP(99):1–1 (2020) 4. Jf, A., Lu, F.B., Jw, A., et al.: From brain science to artificial intelligence. Engineering 6(3), 248–252 (2020) 5. Dama, C., Langford, M., Dan. U.: Teachers’ agency and online education in times of crisis. Comput. Hum. Behav. 121(3),106793 (2021) 6. Deja, M.: Information and knowledge management in higher education institutions: the Polish case. Online Inf. Rev. 43(7), 1209–1227 (2019) 7. Ht, A., Xiang, Y.B., Hty, C.: Learning-related soft skills among online business students in higher education: grade level and managerial role differences in self-regulation, motivation, and social skill. Comput. Hum. Behav. 95(3), 179–186 (2019)

Online Teaching Method of Biochemistry Course of Nursing Specialty Based on Association Rules Jin Chen1(B) , Kai Yang1 , Ying Pang1 , Menglai Shen1,2 , Ping Liu1 , and Jie Lu3 1 Anhui Medical College, Hefei 230601, China

[email protected]

2 Hefei Jinyu Medical Laboratory, Hefei 230000, China 3 Shanghai Institute of Visual Arts (SIVA), Shanghai 201620, China

Abstract. The content of biochemistry course of nursing specialty is relatively abstract, which is difficult for students to understand and has low learning enthusiasm. In order to improve students’ learning enthusiasm and improve the overall teaching effect, an online teaching method of biochemistry course of nursing specialty based on association rules was studied. The problems existing in online teaching are summarized and analyzed. Based on the existing problems and on the basis of in-depth understanding of data mining theory, association rules algorithm is applied to online learning behavior analysis, online examination score analysis, teaching resource recommendation and teaching quality evaluation to mine the hidden and useful knowledge in teaching data and realize the effective combination of teaching and learning in online teaching. The test results show that under the application of this method, the accuracy of course recommendation is maintained at about 80%, which improves the overall teaching effect. Keywords: Association rules · Nursing specialty · Biochemistry course · Online teaching method

1 Introduction With the rapid development of mobile Internet, Internet of things and other new generation information technologies, data has become an important information resource in modern society. The processing of big data is inseparable from the application of data mining technology. Therefore, data mining technology plays a very important role in all walks of life. In the education industry, the mobile Internet has created a cross time and space learning method, the way students acquire knowledge has changed fundamentally, and the traditional teaching mode is gradually changing into an online based teaching mode. In the face of a large number of data resources generated in the process of online teaching, how to make full use of these teaching data and provide useful decision-making basis for online teaching is a problem worthy of discussion. Biochemistry is an important public basic course for nursing majors in medical colleges. It is a necessary basis for further study of physiology, pharmacology and clinical © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 668–682, 2022. https://doi.org/10.1007/978-3-031-21161-4_51

Online Teaching Method of Biochemistry Course

669

medicine. Biochemistry course has many contents, which is highly abstract and logical, and requires students to have a certain physical and chemical foundation to be easy to understand. However, in recent years, nursing majors in many medical colleges have implemented the enrollment model of both arts and Sciences. For liberal arts students, due to their weak physical and chemical foundation, they are subjectively afraid and resistant to the learning of Biochemistry, forming the bad habit of passive learning, and the learning effect is poor.

2 Related Work Introduction Reference [1] aimed at the problems existing in the current network teaching, put forward methods to improve the effect of network teaching of biochemistry courses of nursing specialty, including careful preparation of lessons in the early stage, interactive discussion type live teaching, enhancing the sense of learning community, etc., taking multiple measures to improve the teaching effect of network teaching and ensure the teaching quality. Reference [2] takes the course Ideological and political education as the starting point to share the experience of online teaching practice and teaching experience of the basic course “biochemistry” of undergraduate nursing major in the Medical College of Guizhou University. The main contents include: the basic situation and teaching requirements of the course, the special scheme designed for online teaching, how to teach professional knowledge and integrate ideological and political elements in online teaching implementation cases, as well as the evaluation of the implementation effect of online teaching, In order to provide reference for online teaching of related courses in other colleges and universities. The above existing research methods do not recommend courses according to the association rules of courses, resulting in low accuracy of course recommendation. In order to improve the accuracy of course recommendation, this paper proposes an online teaching method of biochemistry course of nursing specialty based on association rules. Firstly, it summarizes and analyzes the problems existing in online teaching. Then, based on the existing problems and on the basis of in-depth understanding of data mining theory, this paper focuses on the application of association rules in online learning behavior analysis, online examination score analysis, teaching resource recommendation and teaching quality review, and discusses how to find hidden and useful knowledge from a large number of teaching data to help students learn effectively, Guide teachers to teach effectively, so as to realize the effective combination of teaching and learning in online teaching, and also provide a certain reference value for the application and development of data mining technology in the field of education.

3 Problems in Online Teaching Compared with traditional teaching, online teaching shows some characteristics of the information age, but some problems have also been found in the process of implementing online teaching, mainly focusing on the insufficient use of teaching resources, the lack of online application skills of teachers, the imperfect statistical function of network teaching platform, and the lack of effective data mining methods. The specific analysis is as follows:

670

J. Chen et al.

(1) Due to the lack of understanding of the value of online teaching, some teachers carry out online teaching under the pressure of applying for a project or meeting the teaching evaluation. Once the project is completed and the evaluation is completed, the developed online courses and digital teaching resources will be idle, not fully utilized and continue the later construction. (2) Due to the lack of online application skills, some teachers’ teaching contents have remained unchanged for several years after the construction of online courses; Some teachers even put too much emphasis on students’ autonomous learning ability, throwing a pile of learning materials online for students to digest and absorb by themselves, ignoring the timely adjustment of online teaching and support services according to students’ conditions, and ignoring the analysis of online teaching effect. (3) Due to the incomplete statistical functions of online teaching platforms and the lack of sound online learning tracking and monitoring, teachers can not know the detailed learning effects of students’ knowledge points in time, and can not grasp the learning dynamics of students in time, so that teaching contents and organizational activities can not be adjusted accordingly with the dynamic changes of students’ learning status, Not to mention giving targeted guidance to students, so it is easy to separate teaching from learning, and online teaching can not achieve the expected effect. (4) Due to the lack of mining methods and means, people often ignore a large number of teaching data generated in the process of online teaching, and can not find valuable knowledge and laws in the data, resulting in a great waste of data resources and limiting the improvement and development of online teaching. At the same time, it also puts forward new requirements and convenience for teachers and students. Online teaching not only requires teachers to have the ability of online application and information processing, the presentation of teaching content should be multi-level and diverse, and the teaching resources should be rich and novel; More importantly, teachers should mine useful teaching feedback from a large number of teaching data information, and make corresponding adjustments to teaching mode, teaching strategy and teaching content in time, so as to better guide students’ learning and teach students according to their aptitude, so as to achieve the best online teaching effect. Therefore, data mining technology has important application value in online teaching. In addition to online application knowledge and skills, students also need to have a certain cognitive style, cognitive ability and autonomous learning ability, improve their self-control ability in the learning process and cultivate good online learning behavior.

4 Integration of Teaching Data Aiming at the problem that online courses and digital teaching resources can not be fully utilized, teaching data integration is carried out. In the traditional sense, data warehouse refers to the previous data stored in a single place, which is called data center. These data resources are processed in some basic ways, including transaction processing and batch processing. Data warehouse is produced to mine the relationship between data and get

Online Teaching Method of Biochemistry Course

671

the final decision. There are a large number of databases before it is produced. It has certain integration and stability. At the same time, it can reflect the changes of history and is also organized for a certain theme [3]. Data warehouse is like a process, a process of integrating, processing and analyzing the data of courses, grades and so on. The data warehouse integrates the information of different data sources in multiple databases and reorganizes them according to the set subject content. Then, the reorganized historical data will be stored in the data warehouse. Generally, all the data stored will not be modified. The data warehouse structure model is shown in Fig. 1. Final data interface

data mart

Integrator

ETL

database

database

database

Fig. 1. Data warehouse structure

The main function of data warehouse is to extract metadata from the original decentralized database and integrate it together. There is a big difference between operational data and DSS analytical data. The data source of the first database is different from that of the original database, and all the data sources of the first database are different from those of the original database; Second, the comprehensive data in the data warehouse cannot be obtained directly from the original database system [4]. Therefore, before entering the data warehouse, the data must be unified and integrated. This step is the most critical and complex step in the construction of data warehouse. The work to be completed includes: (1) All contradictions in the source data should be unified, such as the same name and different meaning of the field, synonymy of different names, inconsistent units, inconsistent word length, and so on. (2) Data synthesis and calculation. The data synthesis work in the data warehouse can be generated when extracting data from the original database, but many are generated inside the data warehouse, that is, after entering the data warehouse.

672

J. Chen et al.

5 Preprocessing of Teaching Data In order to solve the problem of teachers’ lack of online application skills, simplified data is needed, so the teaching data is preprocessed. Analysis of the teaching data preprocessing in this chapter is for data cleaning, data type conversion and data statistics.Where, data cleaning includes missing value processing and removing outliers.The obtained data is corrected and supplemented by teaching data preprocessing to obtain as much clear and effective information as possible, and to lay the data foundation for the later association rule mining. (1) Data cleaning A. Missing value processing: there are many reasons for the lack of data information. Some information comes from different databases. For example, in the process of getting teachers’ personal information, in addition to teaching information in the educational administration system, we need to transfer personal information from the personnel system, and transfer scientific research information from the scientific research system. Some information can be deleted in the process of transfer. Some are the lack of original data, such as students’ scores. The absence or delay of students will cause the lack of original score information. For these cases, we need to use the most common value to make up for them. B. Delete outliers: that is, delete the abnormal information in the data set, which may affect the quality of the model. If we want to build a model based on students’ performance information (original performance, make-up performance, deferred performance, graduation make-up performance, etc.), some students may fail in the effective performance. After the make-up test, two results will appear. The best way is to delete them directly. For the analysis of students’ performance, we will keep the students’ original performance [5, 6]. (2) Data type conversion: data conversion refers to the conversion of data extracted from relevant business systems according to the data structure of the data warehouse of the auxiliary decision-making system. These conversion operations include the conversion of data format, data content and structure, and data statistics. It includes the following aspects: The data for data type conversion comes from the existing business systems of the school. There are many kinds of business system databases, including Oracle, SQL server, mysql, DB2 and text data. The data formats of different databases are different. Therefore, in the process of data integration, it is necessary to convert the data types of different data files into the data types of the target database.

Online Teaching Method of Biochemistry Course

673

Data content conversion: in the process of data integration, it is necessary to convert the content of the extracted data according to the corresponding rules, including string operation, arithmetic operation of numerical value, date and time, etc. for example, the identity of students in the existing educational administration system includes junior college students, undergraduates, postgraduates and doctoral students, which are represented by strings, In the data integration of data warehouse, it needs to be converted into corresponding numerical types [7]. (3) Data statistics In data integration, statistics and classification results of business data are sometimes required. Therefore, the data transformation process requires statistics and classification of relevant data of business system.

6 Association Rule Mining In view of the imperfect statistical function of the network teaching platform and the lack of effective data mining methods, it is necessary to improve the data statistical function through association rule mining. The data mining process can be divided into four parts: data integration, data preprocessing, datamining, evaluation & interpretation. The process is shown in Fig. 2.

database

data warehouse

Data preprocessing

DataMining

Evaluation&Inte rpretation)

Fig. 2. Data mining process

Knowledge discovery in database, also known as knowledge discovery and data mining, appeared in the late 1980s. It is a new research field with strong vitality based on database technology and combined with artificial intelligence, machine learning, statistics, neural network and other disciplines and technologies; Since the 1990s, it has

674

J. Chen et al.

developed by leaps and bounds, and has become a very active research field involving artificial intelligence, database theory and technology, e-commerce and other disciplines; It shows good application prospects in business management, production control, market analysis, engineering design and scientific exploration, and will become one of the key technologies that will have a far-reaching impact on the society in the next few years. Among them, the research of association rule mining is an important content of knowledge discovery research. Its purpose is to find interesting association relationships or patterns between itemsets in large-scale data sets [8]. Association rule mining is an important part of data mining. Through association rule mining, we can find interesting associations or related relationships in the itemsets of a large number of data. Shopping basket effect is a typical case of association rule mining. This association relationship is considered according to conventional thinking, and there is often no law to speak of. Through the mining technology of association rules, this law becomes discoverable knowledge. The process of association rule mining generally includes two stages. In the first stage, all high-frequency project groups must be found first, and in the second stage, association rules are generated from these high-frequency project groups [9]. The first stage is to find all high-frequency project groups in all the original database data, that is, to find out the item set that all projects meet the minimum support or not less than the minimum support. Find all frequent itemsets to form association rules. The second stage is to generate association rules. That is, using the high-frequency item group in the previous step to generate rules, under the condition of minimum confidence, if the trust of a rule can meet the minimum trust, association rules are formed. For example, the reliability of the rule XY generated by the high-frequency itemset g-project group {x, y} is obtained by using the association rule formula. If the reliability is not less than the minimum reliability, XY is called the association rule. The problem of association rule mining was first proposed by R. Agrawal in 1993. Since then, many researchers have conducted extensive research on this problem, mainly focusing on improving the association rule mining algorithm to improve the efficiency of mining and popularizing the application of association rule mining. So far, one of the classical association rule mining algorithms is still the Apriori algorithm proposed by R. Agrawal and others. The main idea of this algorithm is to first find the frequent itemsets in a given large data set, and then generate strong association rules through the frequent itemsets; The core idea of finding frequent itemsets is to use the results of the previous scan of the database to generate the candidate itemsets of this scan, so as to improve the efficiency of the search. The concepts of frequent itemsets and strong association rules are determined by two indicators: support and reliability; The itemsets that meet the requirements of support are called frequent itemsets, and the rules that meet the requirements of reliability are called strong association rules [10, 11].

Online Teaching Method of Biochemistry Course

675

The basic process of mining association rules with Apriori algorithm is shown in Fig. 3. Start

Scan database

input data

Set minimum support and minimum confidence

Generate candidate set

N

Generating frequent itemsets CN and support numbers

Y Minimum support data

End

N

Y Delete data with support less than the minimum support Generate frequent itemset LN and strong rules to calculate support and confidence

Fig. 3. Basic flow of the Apriori algorithm

Where, the confidence and support formulas are calculated as follows: Confidence: X −µ Y = √ 1/5

(1)

In the formula (1), Y represents confidence; X represents average value of the teaching data; µ represents standard deviation of the teaching data. Support: F =Y ·

P(A ∩ B) n

(2)

In the formula (2), F represents support degree; P(A ∩ B) represents the probability of both transaction A and transaction B; n represents the number of things in the transaction set.

676

J. Chen et al.

7 Design the Online Teaching Method of Biochemistry Course of Nursing Specialty Based on Association Rules The association rules mined above can help reasonably arrange teaching plans, timely reform teaching methods and guide students’ learning methods. Examinations are not only an investigation of students’ learning, but sometimes more a test of teachers’ teaching level. Therefore, when candidates prepare for the examination, teachers can give targeted guidance according to the actual situation of candidates, Then make reasonable arrangements for students to pass the exam smoothly. Designing teaching method based on Association Rules refers to clarifying the relationship between various factors and online teaching through the above mined association rules, and designing online teaching method according to the obtained relationship, so as to make online teaching more reasonable. The process is shown in Fig. 4.

Start Enter the association rule collection for mining Analysis of teaching behavior

Student achievement management

Teaching resources recommendation

Teaching evaluation

Find out the related factors

Find out the related factors

Find out the related factors

Find out the related factors

Association analysis

Association analysis

Association analysis

Association analysis

Mining results

Optimize teaching methods End

Fig. 4. Design the process of teaching methods based on the association rules

Online Teaching Method of Biochemistry Course

677

(1) Analysis of teaching behavior Among many teaching behaviors, which are related to the effect of online teaching? What is the learning behavior of most students? By mining the relationship between students’ online learning behavior and learning effect, and looking at the relationship between their academic performance and which factors, we can predict what kind of learning behavior in the online environment can make personnel work twice with half effort and achieve better academic performance. (2) Student achievement management Effective student achievement management can not only improve students’ learning ability, guide teachers to reasonably arrange teaching plans and improve the overall competitiveness of colleges and universities, but also better serve students, teachers and teaching. The goal of applying association rule mining technology in student achievement management is to transform a large number of student achievement data into valuable information and knowledge. Data mining technology can not only help managers mine interested knowledge and information from the massive data in the database, but also help teaching managers study them to varying degrees, so that they can use and analyze these data more effectively. In recent years, the research perspective of student achievement management has gradually developed from the traditional management mode to a management mode based on mining the effective information of achievement data, supplemented by the traditional management mode. (3) Teaching resources recommendation When users visit the teaching system, the corresponding “access traces” are generated and actively recorded and stored by the system. Through association rule mining, the hidden knowledge and patterns in these access behaviors can be found. These knowledge and patterns can describe the general behavior law of users, which is the key to constructing user interest model. The fusion of association rule mining and information recommendation is also the key. (4) Teaching evaluation In teaching evaluation, the evaluation standard, as a part of it, affects the practical significance of teaching quality evaluation. A reasonable teaching evaluation standard can play a role of supervision and feedback on teaching. Teaching evaluation is an overall evaluation of teachers’ teaching process, content and teaching effect. It is an important way to detect and measure teachers’ teaching quality. Teachers can recognize the shortcomings in the teaching process through the feedback information of teaching evaluation, so as to continuously improve teaching methods. Teaching managers can also put forward targeted suggestions and opinions to teachers through the evaluation information fed back by students and the previous mutual evaluation information of teachers, and encourage and suggest teachers to continuously improve teaching methods

678

J. Chen et al.

and means in the future teaching process, so as to further improve the teaching quality. At the same time, It can also provide decision-making basis for college teaching managers to carry out teaching reform.

8 Example Analysis Take the application of association rule in teaching resource recommendation as an example. 8.1 Analysis Tools This study selects the data mining software Weka (Waikato environment for knowledge analysis) as the learning analysis tool. The tool is an open source intelligent analysis project supported by the New Zealand government. It is widely used in many fields such as finance, medical treatment and transportation. It can directly connect to the system database and realize the seamless integration of heterogeneous data sources. Weka contains a large number of data mining algorithms, mainly including association rules, clustering, attribute selection, classification, visualization, preprocessing and so on. In addition, with the help of Weka interface, we can develop new data mining algorithms on Weka architecture, or realize our own visualization tools by learning from Weka methods. The data format of Weka tool is ARFF (attribute relationship file format). The expression of ARFF format is similar to excel table. Each row in the table represents an instance in the database, and each column represents an attribute, that is, a field in the database. 8.2 Research Object The experimental data sample of this paper comes from the database of the network teaching system, and the learning data of the top 10 professional students in the system are selected. During the construction of the online teaching system, the website group technology is used, that is, each online course is essentially an independent website, which can be called a sub station relative to the whole course platform. Online courses have independent domain names and can be distributed on different servers. Users can log in to online courses through the main portal or directly to a course sub station, You can jump freely between courses, and single sign on technology solves the problem of repeated login. If a user logs in to an online course, a “trace” will be left on the server where the online course is stored, that is, a server log file will be generated. If web mining is used, the work of the data collection stage is to collect all the server log files that store the course website. In the next data preprocessing stage, these log files will be cleaned and identified, and then enter the mining stage for rule mining to finally get the desired course Association. The format of “log” information is greatly different from ordinary log files. It only stores and records some attribute information related to course recommendation, including user name, grade, major, access course and access time, as shown in Table 1.

Online Teaching Method of Biochemistry Course

679

Table 1. Top 10 course login logs Serial number

User name

Course

Login time

1

XX

Fundamentals of nursing A1

2020/5/3/10:25:36

2

XX

First aid nursing technology A2

2020/5/4/12:20:11

3

XX

Internal medicine nursing A3

2020/5/3/8:20:43

4

XX

Surgical nursing A4

2020/5/5/14:4:50

5

XX

Obstetrics and gynecology nursing A5

2020/6/1/17:48:35

6

XX

Pediatric nursing A6

2020/5/7/15:20:11

7

XX

Community nursing A7

2020/5/3/25:24:45

8

XX

Internal medicine nursing

2020/5/18/12:10:25

9

XX

Community nursing

2020/5/3/14:15:30

10

XX

First aid nursing technology

2020/5/3/9:11:20

When the “log” information is written into the database one by one, it is recorded in the special course login log table. The next data mining work is carried out for the corresponding database files. 8.3 Parameter Setting With the help of the data mining tool Weka platform, the association rules are mined, and the complete Apriori algorithm written in advance is imported into Weka tool. Open the data file, select Apriori algorithm and set the parameters: the lower bound of minimum support is 0.5, the upper bound is 0.5, and the minimum confidence is 0.3. 8.4 Basic Processes Based on Association Rules The recommendation process of teaching resources based on association rules is as follows: Step 1: first, find out all the association rules G ‘supported by the user in the rule set G, that is, all the courses preceding the rule G’ have been accessed by the user. Step 2: find out all courses h predicted by association rule G ‘but not accessed by users. Step 3: directly recommend all courses h to users, or sort according to the weighted value of courses in H, and select the first m to recommend (the value of M can be adjusted at any time according to the situation). The mining association rules are shown in Table 2.

680

J. Chen et al. Table 2. Association rules Rule

Confidence degree

Support degree

A1A2

100%

60%

A1A3

50%

50%

A1A4

52%

40%

A1A5

66%

60%

A1A6

70%

40%

A1A7

72%

45%

A2A1

65%

50%

A2A3

100%

65%

A2A4

58%

60%

A2A5

70%

50%

Taking A1A2 rule as an example, it can be determined that the confidence of the two rules reaches 100%, indicating that the rules are effective, which can be recommended to student users, that is, students who visit the basis of nursing are more likely to visit the emergency nursing technology course, so they can be recommended to the emergency nursing technology course; vice versa. You can also recommend relevant online course information to teachers according to the rule table. For example, the information provided in Table 2 indicates that there is a strong correlation between the two courses of basic nursing and emergency nursing technology. Teachers can arrange the links between courses and adjust the teaching contents of courses accordingly. 8.5 Evaluation Method For association rule analysis, this paper uses accuracy (Precision@k) And average the number of association rule entries to evaluate the algorithm results. At the same time, because the algorithm has been actually used in the system, it is also necessary to evaluate and explain the running time of the algorithm on the server. Since some algorithms have not been implemented in the system, the classic standard data set on association rule testing, groceries data set and system data are selected for comparison and explanation. The calculation formula of accuracy is shown in formula (3).  (3) P = |S(k) ∩ D| |S(k)| In the formula (3), S(k) represents the first k rules of association rules in the result of association rule algorithm operation, D represents all association rules obtained from the test set. 8.6 Course Recommendation Results Reference [1] and reference [2] methods compare the course recommendation accuracy of the three methods, and the results are shown in Fig. 5.

Online Teaching Method of Biochemistry Course

681

100 Reference[1]method Reference[2]method

Accuracy/%

80

The method of this paper

60

40

20

0

0

20

40

60

80

100

Support degree/%

Fig. 5. Accuracy results

It can be found from Fig. 5 that the accuracy of the method and course recommended in this paper increases with the increasing support.This is due to the high support for removing a large number of items with fewer appearances, so the results increase with higher support, and the final course recommendation accuracy remains at around 80%, significantly higher than the other methods.

9 Conclusion Biochemistry is one of the compulsory basic courses stipulated in the teaching plan of nursing specialty in higher vocational colleges in China. It is also one of the important professional basic courses of nursing specialty in higher vocational colleges. It is a marginal discipline developed from organic chemistry and physiology. The knowledge theory of biochemistry has penetrated into other professional courses of the specialty and various fields of clinical nursing. It is the basis for nursing students in higher vocational colleges to learn professional courses such as basic nursing, emergency nursing technology, internal medicine nursing, surgical nursing, obstetrics and gynecology nursing, pediatric nursing and community nursing. This discipline plays an important basic role in the follow-up professional course teaching, and integrates with other disciplines and appears in the test questions of nurse qualification examination. The biochemistry of nursing major in higher vocational colleges has the characteristics of strong theoretical content, which involves complex material molecular structure formula, chemical reaction formula, material metabolism pathway and regulation mode, transmission of genetic information and the latest progress of molecular biology. It is one of the difficult courses for nursing majors in higher vocational colleges. With the rapid development of information technology and the thrust of various needs, online education came into being. The participation of colleges and universities in the field of online education increases students’ educational opportunities. At the same time, they also use “online

682

J. Chen et al.

education” to enrich teaching methods and attract more learners. Promoting the modernization of education is an important means to meet the growing needs of lifelong learning in society. Based on the above background, an online teaching method of nursing professional biochemistry courses is proposed based on the association rules.The correlation rules algorithm was applied to online learning behavior analysis, online test score analysis, teaching resource recommendation and teaching quality evaluation, and a large number of hidden teaching data was excavated, combined with online teaching teaching and learning, recommended course knowledge popular with students.The example analysis results demonstrate the effectiveness of the studied method. Fund Project. Anhui Province University Humanities and Social Sciences Research Key Project (SK2019A0937); School-level Scientific Research Project (WJH202010t); Quality Engineering: Production Education Integration Training Base Project (2021cjrh020).

References 1. Meilan, X., Zheng, Z., Li Mengyang, X., Hongwei, L.Y., Lin, H.: Exploration of network teaching model of biochemistry course in nursing specialty. Educ. Teach. Forum 43, 294–297 (2020) 2. Xiaodong, R., Linzhao, L., Yali, Z.: Discussion on online teaching practice of biochemistry course combined with ideological and political education. Guide Sci. Educ. 28, 123–124 (2020) 3. Moslehi, F., Haeri, A., Martínez-Lvarez, F.: A novel hybrid GA–PSO framework for mining quantitative association rules. Soft Comput. 24(6), 4645–4666 (2020) 4. Tjortjis, C.: Mining Association Rules from Code (MARC) to support legacy software management. Softw. Qual. J. 28(6), 633–662 (2020) 5. Wu, Y., Zhang, J.: Building the electronic evidence analysis model based on association rule mining and FP-growth algorithm. Soft. Comput. 24(2), 7925–7936 (2020) 6. Delgado, F.: Teaching physics for computer science students in higher education during the COVID-19 pandemic: a fully internet-supported course. Fut. Internet 13(2), 35 (2021) 7. Boll, S.C., Müller, H., Lunte, T., et al.: Making, together, alone: experiences from teaching a hardware-oriented course remotely. IEEE Pervas. Comput. 19(4), 35–41 (2020) 8. Sun, J.: Research on resource allocation of vocal music teaching system based on mobile edge computing. Comput. Commun. 160(2), 342–350 (2020) 9. Yu, X.: Resource scheduling for piano teaching system of internet of things based on mobile edge computing. Comput. Commun. 158(99), 73–84 (2020) 10. Xu, N., Fan, W.H.: Research on interactive augmented reality teaching system for numerical optimization teaching. Comput. Simul. 37(11), 203–206+298 (2020)

Music Distance Education Resource Sharing Method Based on Big Data Platform Jun Zhou(B) and Hui Lin College of Art, Xinyu University, Xinyu 338000, China [email protected]

Abstract. In order to promote the rapid transmission of music distance education resources, so that students can obtain more shared data information in unit time. This paper proposes a resource sharing method for music distance education based on the big data platform. According to the connection form of the big data platform architecture, determine the function capability of the virtualization sharing technology. Then, by calculating the shared weight of educational resources, the farthest transmission distance of music distance education resources in the network environment is constrained. Complete the relevant application technology analysis based on the big data platform. On this basis, construct the IaaS resource sharing structure. Combined with the established OpenStack scheduling policy, the statistical numerical indicators of shared access are calculated. To realize the smooth application of the music distance education resource sharing method based on the big data platform. The experimental results show that under the action of this new sharing method, the total value of shared data information obtained by students in unit time increases significantly. The method can promote the rapid transmission of music distance education resources, and meet the needs of practical applications. Keywords: Big data platform · Music distance education · Resource sharing · Shared weight · Iaas structure · Openstack scheduling strategy

1 Introduction Colleges and universities are the departments that concentrate most of the digital education resources, and their digital resource sharing status largely determines the level of my country’s digital education resource sharing. However, there are still many problems in the construction and sharing of digital educational resources in various colleges and universities in my country. It hinders the further development of my country’s digital education resources co-construction and sharing project. Therefore, it has certain practical significance and application value to study how to better solve these problems and realize the effective sharing of digital resources among universities [1]. With the innovation of technological means, the application of distance education is becoming more and more extensive. Media technology enriches human life, and dynamic music rhythm fills every corner of the world. The development of the music world mainly depends on the © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 683–694, 2022. https://doi.org/10.1007/978-3-031-21161-4_52

684

J. Zhou and H. Lin

dissemination of the network. In music education, there are not only students heading for university or professional teacher tutoring. There are also many musicians with low education and music lovers who are not engaged in music-related majors but have great enthusiasm for music. The latter do not have the ability to receive professional music education on a regular basis in work and life. However, distance education can give them an opportunity to continue their studies and enable them to improve their musical quality. At present, there are still many imperfections in music education in our country. Due to the unbalanced distribution of economic education in my country, it has directly led to the disparity in the level of music education in various regions. At present, excellent teacher resources and teaching equipment are gathered in the developed areas of education. In some more remote areas, the number of teachers is small and the pressure on teachers is heavy. This method cannot effectively guide students. Moreover, there are few teaching equipment, which cannot guarantee the normal use of students. At the same time, music education is also greatly limited in the current education system. General music education only exists in music classes in primary and secondary schools, and cannot meet the needs of other people from all walks of life who love music. But with the digitalization of music education, distance education can dramatically change the current situation. Not only students can learn music through distance education. In addition, people in the community can also cultivate their interest in music and learn about music theory in their spare time. With the continuous development of informatization in recent years, cloud computing, mobile Internet, and smart Internet of Things have been widely used. The amount of information data continues to grow rapidly. In order to quickly and effectively deal with a large amount of data, the storage, reading and retrieval of distance education information, etc. Big data technology has become a hot spot of attention of universities and colleges [2]. This also brings new security challenges, and more security risks need to be dealt with. Cybersecurity risks in today’s era have become more diverse and complex. How to ensure data security in the current environment and how to provide students with more accurate security control strategies need in-depth research and discussion. At the same time, big data also provides new opportunities for the development of information security.

2 Applied Technical Analysis 2.1 Big Data Platform Architecture In order to ensure the information security in the music distance education network, it is implemented through the definition of various security responsibilities. And provide support for the organization’s safety management, safety operation and maintenance, and safety technology. It is divided into three layers: decision-making layer, management layer and executive layer. The safety management system and process are placed in the big data platform architecture. The security management framework provides the basis for managing risks and building trust in the system. And defines all security management elements, methods, objects, rules, processes, etc. For the current big data computing, the distributed computing method is mainly adopted. When using distributed computing, it must face the process of data transmission

Music Distance Education Resource

685

and information interaction. Attackers can steal and tamper with data through various means. The data information stored, transmitted and processed in the big data platform is classified according to its risk. And on this basis, for different types of data information, according to the principle of moderate protection and acceptable residual risk. The protection of different security protection levels is carried out by grading [3]. In this way, the problem that large-scale complex systems are difficult to achieve overall highlevel protection can be solved. With appropriate investment, the data and information that need to be protected can be properly protected. The complete music distance education network big data platform architecture is shown in Fig. 1. Big data platform architecture Distance education technology system

Education technology

Education management

Education institutions

Resource management

Analysis of music education resource information

Distance education strategy and resource services Music distance education big data resources are appropriate

Fig. 1. Schematic diagram of the structure of the big data platform system

At all stages of the entire data life cycle, such as data collection, storage, mining, publishing, and deletion of big data applications, there is a risk of privacy leakage. Privacy in the era of big data is to carry out effective data mining without exposing user sensitive information. In the traditional information security field, more attention is paid to the security attributes such as the privacy of files. The prior art is mainly based on static datasets. Therefore, we must also consider how to realize the privacy protection and effective use of dynamic data in the complex environment of rapid data changes in the era of big data. 2.2 Virtual Sharing Technology Virtualization is a broad and varied concept that refers to the fact that computer elements are not on real hardware facilities, but run in a virtual environment. It is a solution to optimize resources and simplify management. It simplifies the reconfiguration process while expanding hardware capacity. The main purpose of virtualization technology is to simplify the complex resource access and management process. It is not limited and constrained by physical devices and is a logical representation of resources [4]. Virtual resources manage physical resources uniformly through standardized interfaces. The interaction between physical storage and virtual resources is illustrated by some

686

J. Zhou and H. Lin

basic patterns of virtualization. This can greatly reduce the difficulty of accessing and managing physical resources. Virtualization technology is one of the important foundations for data center to realize resource sharing and big data storage. It makes the computing power of the data center more scalable, data access is more flexible, management is simpler, and it can better serve cloud computing. Virtualization dynamically maps the physical resources of the infrastructure to application drivers, and the virtualized infrastructure creates a pool of virtualized resources. Unified management of servers, storage and networking. The resources in the resource pool can be called at any time according to the needs of the application. Figure 2 reflects the complete framework of virtualized sharing of music distance education resources.

Remote application

Remote application

Remote application

Remote application

Music education virtual operating system

Music education virtual operating system

The virtual machine

The virtual machine

Virtual sharing environment of music distance education resources

Remote server

Resource sharing server

Distance learning equipment

Fig. 2. A framework for virtualized sharing of music distance education resources

A virtualized infrastructure separates the music distance education environment from the underlying hardware infrastructure, allowing multiple servers, storage, and networks to aggregate into a shared pool of resources. Dynamically provide resources in the resource pool to applications in a safe and reliable manner according to user needs. According to different application fields of virtualization, virtualization can be divided into server virtualization, storage virtualization, network virtualization and so on. The core of storage virtualization is the mapping of physical storage devices to a single logical storage resource pool, which unifies various heterogeneous storage resources into a resource that is a single view to users. 2.3 Educational Resource Sharing Weight Calculation All computing nodes are screened by the big data platform. If only one computing node passes the shared filter, the execution of this step can be skipped, and the node can be directly returned to create a virtual machine instance. Otherwise, further calculate the

Music Distance Education Resource

687

strategy through the weights, according to the size of the weights. Sort each node, select the node with the smallest weight, and create a virtual machine instance as the optimal computing node [5]. The shared system first obtains the performance monitoring data of all computing nodes. Then, according to the characteristics of the requested instance, the load status of the computing node is evaluated. That is, the weight is calculated, and the one with the smallest weight is selected as the target computing node. The load can be represented as a 3-dimensional vector < CPU, Mem, Net > related to CPU, memory, network. The algorithm should take into account these three types of resources at the same time to avoid affecting the overall performance due to optimizing a certain quantity. The specific educational resource sharing weight calculation expression is as follows. β ×Y    D=  1 − δ 2 × 1 − e1

(1)

In formula (1), β represents the sharing coefficient of music education resources based on the big data platform. Y represents the mean value of educational resource transmission per unit time. δ and e represent two different distance education scaling coefficients [6]. Music distance education resource sharing behavior is similar to CPU-intensive applications, for instances running communication-intensive applications. Network congestion due to large-scale data transmission on the network increases data transmission latency, reduces data transmission efficiency, and affects application execution. The network load is too heavy, the blocking time is prolonged, the data may be lost, and it needs to be resent. This affects the current data transfer time as well as subsequent data transfers. Even if the blocking time is not particularly long. It will also increase the total time of data transmission in the network, affecting the efficiency of data sending, receiving and processing. Therefore, when creating an instance, you should weigh the network load of the compute node to calculate the overall load.

3 Music Distance Education Resource Sharing Method 3.1 IaaS Build Build an IaaS resource sharing structure belonging to the music distance education network. It is only necessary to install the components required for deployment according to the rigid requirements of the big data platform, combined with the infrastructure connection environment. First, it is necessary to analyze the existing hardware environment and determine the type of operating system. Next, install the authentication server. Then, install and configure and compute and mirror services. Finally, install the storage service and web console. In the music distance education network, the IaaS resource sharing structure supports various types of application software such as RHEL, CentOS, Fedora, and Ubuntu. In order to fully share the resource information, the processing software of the Folsom version should be selected.

688

J. Zhou and H. Lin

Authentication services are provided by the OpenStack Keystone component. It has two main functions, music education resource management: tracking and monitoring user behavior. Service Catalog: Provides users with the location of the service catalog and API endpoints available. Authentication divides users into Users, Tenants, and roles. The three of them are bound together and can be managed with the command line or by modifying the etc./nova/policy.json file for unified management. The mirroring service is provided by the IaaS resource sharing structure. Its installation and configuration consists of two steps, configuring the mirror service backend database and the Glance configuration file. Nova node installation is the most important part. Its nodes are divided into two types: control nodes and computing nodes, the former is used to manage the latter in a unified manner. Object storage is also provided by the IaaS resource sharing structure for mirror storage of music education resources. Dashboard is provided by OpenStack Horizon. It is used to provide a web-based management interface, both components are optional. Music distance education is a form of music education resource dissemination based on big data platform. The core of distance education is to use appropriate methods and means to express boring content, express abstract concepts with vivid animations, and make students easier to accept and understand. To stimulate interest in learning and improve learning effect. Music micro-class is a long-distance educational activity with the help of media information tools. It spreads music education resources and music education functions outward. Students can check materials in the music micro-course database, realize the sharing of high-quality educational resources, and meet individual learning requirements. It is an educational activity with management, evaluation and regulation. At present, the research on establishing a micro-course distance education platform in ordinary colleges and universities in my country is still in the exploratory stage. Moreover, we are faced with many problems such as unreasonable design scheme and waste of teaching resources. Therefore, it is very urgent to develop a high-efficiency big data distance education platform. Let f denote the resource information mirror storage coefficient in the IaaS structure. φ represents a given functional treatment index. Eˆ represents the transmission characteristics of music distance education information in the IaaS structure. χ represents a resource information discrimination coefficient related to the IaaS resource sharing structure [7]. With the support of the above physical quantities, the formula (1) can be combined, and the construction principle of the IaaS resource sharing structure can be expressed as:   2 χ f ·D (2) P= φ−1 Eˆ The IaaS resource-sharing fabric is at the heart of OpenStack Compute. It provides endpoints for all shared API queries and initiates most deployment activities. Such as instance running, termination, etc., as well as implementing some management policies.

Music Distance Education Resource

689

3.2 OpenStack Scheduling Policy In the OpenStack Compute software architecture. The big data platform host completes the sharing task of music distance education resources through the interaction of message queue and Nova-database. Let r denote the minimum educational resource sharing behavioral feature. A represents the initial labeling coefficient of music distance education resources. I represents the screening coefficient of the resource information to be shared. S represents the mean value of educational resource scheduling within the shared channel. u represents the resource scheduling permission based on OpenStack policy [8]. Combining the above physical quantities, the sharing capability of OpenStack policies can be expressed as: +∞ W = r=1

  A2 I × S  dS u·P

(3)

When the scheduling policy is started, when the instance is created, each computing node already has a certain load. According to user-specified requirements, virtual machine instances are created for users on compute nodes in the shortest possible time. At the same time, make the CPU, I/O and network load of each computing node as balanced as possible. The goal of the filtering strategy is to filter unavailable compute nodes. Filter out the nodes that meet the user’s needs from the available computing nodes. Thereby, the coverage of the second-step operation execution is narrowed, and the waiting time of the user is shortened. The scheduling process is mainly divided into the following steps: According to the hardware requirements configured by the user, determine whether the shared node is available. Based on the instance type and hypervisor type specified by the user. It judges whether the music distance education resource information meets the actual application requirements. This paper adopts a user-defined scheduling structure to filter the shared resources. Let α denote a given educational resource planning metric. cmax represents the maximum numerical result of the filtering permission of the resource to be shared. cmin represents the minimum numerical result of the filtering permission of the resource to be shared. h stands for shared behavior vector. bα represents the filter criterion coefficient of music distance education resource data when the metric value α is. ω represents the resource data inductance item in the big data platform [9]. The coefficient in the formula was set with reference to the reference [9]. With the support of the above physical quantities, formula (3) is simultaneously established. The execution expression of the OpenStack scheduling policy can be defined as: Z=

hW · |bα |ω |cmax − cmin |2

(4)

After all shared computing nodes are filtered by the filtering policy. If only one compute node passes the big data platform filter. The execution of this step can be skipped, and the node can be directly returned to create a virtual machine instance for

690

J. Zhou and H. Lin

distance education. Otherwise, further through the weight calculation strategy, according to the size of the weight, the nodes are sorted. Select the node with the smallest weight as the optimal computing node to create a virtual machine instance for distance education. 3.3 Shared Traffic Statistics The statistics of shared visits is the final processing link in the design of the music distance education resource sharing method. It can discriminate the farthest transmission distance of music distance education resource information according to the actual connection form shown by the big data platform. On the one hand, it ensures the smooth implementation of the OpenStack scheduling policy. On the other hand, the music distance education network can also maintain a relatively stable connection state. In this way, various types of resource information are classified and stored to realize the planning of shared behavior paths. Music distance education, a new form of education, emphasizes that learners’ knowledge should be constructed by themselves in the interaction of the environment. The learner’s cognition plays the main role, and the student is the main body of information processing and the active constructor. Rather than passive recipients and indoctrinated objects of external stimuli. The theory of constructivism is not only the theoretical basis for the production of music micro-lectures, but also the theoretical basis for the construction of the distance education platform of music micro-lectures. At the same time, this special audiovisual medium is an integral part of the teaching system. Teaching activities need to control teaching objectives, teaching content and form, teaching methods, teaching quality, etc., and also need to compare teaching design schemes, and choose the best decision. It is also necessary to use feedback information to evaluate the teaching effect, and then modify the original teaching methods. Educational system cybernetics is the theoretical basis for the distance education platform of music micro-course. In addition, the music distance education platform considers the optimal allocation of learning resources from the perspective of the teaching dissemination process and the complete teaching system. Coding and decoding are the basic theories of instructional media preparation and dissemination. Micro-lecture is a multimedia technology that integrates various media such as text, image, sound, animation and video through a computer to establish a logical connection. And they are sampled and quantized, encoded and compressed, edited and modified, stored and transmitted, and reconstructed and displayed. It is based on the teaching communication theory. Let v0 denote the minimum information ratio item coefficient of music distance education resource sharing behavior. v represents the result of the maximum value of the coefficient v0 . q˙ and q˜ represent two different count feature values of music distance education resource data. In the big data platform environment, the inequality condition of q˙ = q˜ is always established [10]. σ represents the resource sharing pattern vector associated with the q˙ eigenvalue. ϑ represents the resource sharing pattern vector associated with the q˜ eigenvalues. With the support of the above physical quantities, formula (4) can be combined to express the statistical results of the shared access to

Music Distance Education Resource

music distance education resources as:     v   v0 q˙ −σ  1 − Z   v0 =1 M =  v 2 v0 q˜ ϑ

691

(5)

v0 =1

After a large number of music distance education resources are produced and collected, they need to be classified, organized and deployed. First of all, the various music micro-lectures collected should be classified according to the types of music majors and file types. Ordinary college music micro-courses can be divided into theoretical music micro-courses, performance music micro-courses, and appreciative music micro-courses. It can also be divided into vocal music micro-lessons, instrumental music micro-lessons, etc. If the big data platform is regarded as an absolutely stable data information storage space. And in this storage environment, the resource sharing relationship between information and information will not be affected by any external conditions. At this time, the music distance education resource data to be transmitted is processed according to the statistical result of the shared traffic. It can be considered that the longer the transmission distance matched with the data information, the stronger the data inclusiveness of the sharing mode, and vice versa. The terminal computer of the distance education platform can provide one-to-one tutoring to realize individualized teaching. You can also repeat a task tirelessly to improve learning. But the limitation of computers is that they cannot answer students’ questions as flexibly and satisfactorily as teachers. It cannot directly exchange experience and share joy with students, and cannot influence students with its own noble behavior. This is the shortcoming of realizing the sharing of music distance education resources based on the big data platform.

4 Case Analysis In order to highlight the practical application value of the music distance education resource sharing method based on the big data platform. The following comparative experiments were designed. The specific experimental procedure is as follows. Step 1: In the distance education network environment shown in Fig. 3, the single room uses the big data platform and the traditional resource scheduling method to control the education network. The former was used as the experimental group and the latter was used as the control group.

692

J. Zhou and H. Lin Resource information transmission

Student host

Education host Music distance education resource information

Campus mainframe

Resource sharing platform

Fig. 3. Distance education network environment

Step 2: The experimental numerical results of the transmission rate of music distance education resources in the experimental group and the control group were calculated separately. Step 3: Compare the variables of the experimental group and the control group, and summarize the experimental results. The transmission rate of music distance education resources can reflect the total amount of shared data information obtained by students in unit time. Generally speaking, the faster the transmission rate of music distance education resources. The total amount of shared data information obtained by students per unit time is also greater, and vice versa. Table 1 records the numerical levels of the transfer rate of music distance education resources under ideal conditions. Analysis of Table 1 shows that under ideal circumstances, with the continuous increase of the accumulation of music distance education resource information. The transmission rate of educational resources also shows an accelerating numerical trend. However, the value increase in the early stage of the experiment was significantly larger than that in the later stage of the experiment. The global maximum of 7.26 Mb/ms is an increase of 3.94 Mb/ms compared to the global minimum of 3.32 Mb/ms. Figure 4 reflects the specific numerical changes of the transmission rate of music distance education resources in the experimental group and the control group. The transmission rate of music distance education resources in the experimental group maintained an increasing numerical trend throughout the experiment. When the cumulative amount of resource information is equal to 3GB and 6GB, the transmission rate level of the experimental group is lower than the ideal value level. When the accumulated amount of resource information reaches 9GB, the transmission rate level of the experimental group exceeds the ideal value level. But its mean level is always higher than the ideal value. The transmission rate of music distance education resources in the control group maintained a numerical change state of first increasing and then decreasing. Before the accumulated amount of resource information reached 15 GB, the resource transfer rate of the control group kept increasing numerically. However, when the accumulated amount of resource information is between 15 GB and 27 GB, the resource transfer rate of the

Music Distance Education Resource

693

Table 1. The ideal value for the transfer rate of educational resources Cumulative amount of music Resource transfer rate distance education resource /(Mb/ms) information /(Gb) 1

3.32

2

4.15

3

4.68

4

5.04

5

5.53

6

5.87

7

5.96

8

6.07

9

6.11

10

6.23

11

6.29

12

7.03

13

7.14

14

7.21

15

7.26

Fig. 4. Experimental values of the transmission rate of educational resources

control group has been continuously decreasing. During the whole experiment, the mean level was relatively low, and the regional maximum value was much smaller than that of the experimental group.

694

J. Zhou and H. Lin

In summary, it can be seen that under the action of the sharing method based on the big data platform. The transmission rate of music distance education resources has been promoted to a certain extent. It can help students get more shared data information in unit time. This method is in line with the actual construction and design requirements of distance education network.

5 Conclusion The new music distance education resource sharing method analyzes the application ability of virtualized sharing technology on the basis of big data. With the help of the IaaS resource sharing structure and the OpenStack scheduling strategy, the access to shared educational resource data is counted. It realizes the sharing of music distance education resources based on the big data platform. The experimental results show that under the action of this new sharing method, the total value of shared data information obtained by students in unit time increases significantly. From a practical point of view, with the application of this new sharing method, the transmission rate of music distance education resources has shown a significantly accelerated change trend. It can help students get more shared data information in unit time.

References 1. Noor, M.V.M., et al.: Uplink resource sharing and power management scheme for an underlay D2D communication. Wirel. Pers. Commun. 110(2), 637–650 (2020) 2. Krishnamoorthy, D.: A distributed feedback-based online process optimization framework for optimal resource sharing. J. Process Control 97(6), 72–83 (2020) 3. Long, C., Wang, S.: Music classroom assistant teaching system based on intelligent speech recognition. J. Intell. Fuzzy Syst. (Preprint) 1–10 (2021) 4. Abrol, P., Gupta, S., Singh, S.: A QoS aware resource placement approach inspired on the behavior of the social spider mating strategy in the cloud environment. Wirel. Pers. Commun. 113(4), 2027–2065 (2020) 5. Yuan, Y.: Design and realization of computer aided music teaching system based on interactive mode. Comput. Aid. Design Appl. 18, 92–101 (2020) 6. Wang, N., Xu, H., Xu, F., et al.: The algorithmic composition for music copyright protection under deep learning and blockchain. Appl. Soft Comput. 112, 107763 (2021) 7. Tanveer, H., Balz, T., Sumari, N.S., et al.: Pattern analysis of substandard and inadequate distribution of educational resources in urban–rural areas of Abbottabad, Pakistan. GeoJournal 85(5), 1397–1409 (2020) 8. Zhang, L.L.: Simulation research on the integration and sharing of public library resources under cloud computing. Comput. Simul. 37(05), 416–419 (2020) 9. Saito, T., Watanobe, Y.: Learning path recommendation system for programming education based on neural networks. Int. J. Dist. Edu. Technol. 18(1), 36–64 (2020) 10. Awajan, I., Mohamad, M., Al-Quran, A.: Sentiment analysis technique and neutrosophic set theory for mining and ranking big data from online reviews. IEEE Access 9(99), 1–16 (2021)

Real-Time Tracking Method of Students’ Targets in Wushu Distance Teaching Based on Deep Learning Jie Zhang1(B) and Na Ma2 1 Shaanxi University of Chinese Medicine, Xianyang 712046, China

[email protected] 2 Sports Department, Tianjin Foreign Studies University, Tianjin 300204, China

Abstract. The development of information technology has promoted the development of distance education, and Wushu teaching has gradually changed from traditional face-to-face teaching to distance education. In order to improve the teaching effect, it is necessary to track the teaching objects in real time. In order to improve the real-time tracking of students’ goals in Wushu distance learning, a real-time tracking method of Wushu distance learning goals based on deep learning is designed. The experimental results show that the designed real-time tracking method has a short tracking delay, which proves its effectiveness. Keywords: Deep learning · Wushu · Distance teaching · Goal · Real-Time tracking

1 Introduction As one of the important directions of computer vision, the main task of target tracking is to extract the motion information and position information of the target [1], which provides the basis for the semantic analysis of later behavior. Target tracking can be defined as: selecting an area as a target in the first frame of a video stream, automatically finding the target in the next several frames, and outputting the target position. With the development of society [2, 3], people are increasingly pursuing a more intelligent life. Video surveillance exists in many places in our lives, including criminal investigation monitoring, traffic vehicle monitoring, community security monitoring, etc. At present, many of these surveillance videos are artificially tracked. Due to the uncertain factors of manual operation, important information may be missed [4], so the research on target tracking algorithm has strong research value and wide application fields. At present, many tracking algorithms are in the stage of simulation experiment, and there are few real applications. Target tracking faces a series of challenges, such as complex and changeable environment, non-rigid deformation of target, occlusion of target, illumination change, long-term tracking and so on, which leads to great limitations in the research of moving target recognition and tracking technology [5]. Therefore, the © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 695–708, 2022. https://doi.org/10.1007/978-3-031-21161-4_53

696

J. Zhang and N. Ma

research on tracking and recognition based on extended targets is of great value in practical application and indispensable significance in theoretical research. The main tasks of target tracking include acquiring video sequences, preprocessing the video sequences, giving the target to be tracked, extracting target features, matching or binary classification, and finally giving the location or behavior track of the target. The target tracking task needs to meet the requirements of stability, accuracy and real-time. Because of the limitations of algorithms, hardware and large amount of calculation, target tracking is not widely used in commercial applications. Therefore, it is of great significance to study the real-time and stable tracking of extended targets in complex background. Inspired by convolutional networks, some scholars use this advantage of convolutional neural networks to use layers with switch mechanism in the tracking process. Ma et al. have done similar work [6, 7]. They use training CNN [8] on ImageNet to improve the tracking accuracy and stability. This algorithm uses switching between layers with semantic information and fine-grained information, and fuses information from layers for tracking strategy from coarse to fine. However, these algorithms mentioned above are all pre-trained offline on ImageNet. And then directly used for online tracking, and the network will not be fine-tuned in the tracking process. This tracking network that only uses target image data for pre-training is unreliable, because the target in one video can be the background in another video. Therefore, some scholars have proposed a new tracking network [9] to solve the above problems. The network uses video data pre-training-a two-layer tracker based on CNN. This method effectively adjusts the prelearned features according to specific targets during online tracking. Nam proposed a video training CNN network [10] with common network and multiple branches to distinguish the target from the background. However, these video training trackers don’t explicitly use the semantic information of the target, that is, they don’t know the category of the object. Without knowing the category of the object, the tracker will probably fail to track. With the advent of the Internet age, many scholars began to innovate Wushu teaching methods by means of the Internet, improve teachers’ teaching quality and optimize students’ learning mode. The traditional real-time tracking method of Wushu distance teaching students’ goals has poor tracking effect and can’t meet the real-time demand of tracking. Therefore, this paper designs a new real-time tracking method of Wushu distance teaching students’ goals based on deep learning, which can be used as a reference for improving the subsequent Wushu teaching effect. With the improvement of hardware capability and the rapid progress of computing capability, the society is developing towards intelligence. It is imperative to introduce deep learning into the target tracking task, extract target features through convolutional networks, and study the use of deep learning algorithm to complete the target tracking task. By training a large number of different environments and different targets, we can adapt to the tracking of different targets in different environments and get more robust features.

Real-Time Tracking Method of Students’ Targets

697

2 Design of Real-time Tracking Method for Students’ Targets in Wushu Distance Education Based on Deep Learning 2.1 Determine the Tracking Characteristics of Wushu Distance Teaching Goals The most important step of target tracking is to extract the features of the target object. Feature extraction is to convert pixel features into semantic features, and this semantic feature can be divided into traditional artificial features and learning-based features in deep learning. Whether the extracted target features are more abundant and can represent the target determines the accuracy and stability of target tracking, while the amount of calculation in the process of feature extraction determines the speed of target tracking. The traditional feature extraction method is artificially designed according to subjective judgment, which can be divided into generative model features and discriminant model features according to mathematical models. Generative model feature is that the target is represented by template, and tracking is realized by matching the template feature with the candidate target feature in the search area, such as sparse representation, and the candidate target with the sparse coefficient and the smallest error is selected as the result. Discriminant model features refer to binary classification features, which distinguish the target from the background, such as TLD. According to the different calculation of visual features, it can be divided into pattern features, gradient features, shape features and color features. Pattern features include sliding window features obtained by Gabor filter to simulate human eye receptive field, dot-line features expressed by the sum of pixel differences in adjacent matrix areas, LBP local features for extracting illumination invariant characteristics, etc. Gradient features include SIFT features of moving targets by obtaining gradient information around key points, and HOG features of moving targets by gradient direction and intensity of spatial distribution areas. Because the traditional feature is a certain target feature representation method set artificially, the robustness is limited, and the representation effect is better for specific scenes or specific objects, but there are some limitations for complex background and transformed targets in natural environment. The method based on deep learning generally uses convolution network to extract features, which is a learning-based feature. By modifying the weights of convolution layer through the training process, a convolution network with feature extraction ability is finally obtained. This feature is different from the traditional manual setting feature, it can’t be presented in a way that can be understood by people, and it is a feature that can be understood by the network. Generally, the shallow network extracts the edge features and location information of the target, while the deep network extracts the semantic information, which is used for tasks such as classification and identification. The advantages of traditional feature representation are small amount of calculation and high speed, while the disadvantages are that feature representation ability is not strong enough and features are not rich enough. The features extracted by convolution network have the advantages of good feature robustness, rich features and strong representation ability, and are suitable for the transformation target of complex scenes. However, the amount of calculation is large, and a large enough data set is needed for training. With the development of computer technology, the computing power of hardware is gradually enhanced, and new network frameworks are put forward one after

698

J. Zhang and N. Ma

another. The future development prospect of deep learning is immeasurable, so the algorithm based on deep learning has room for development and application value. As far as the current development trend is concerned, how to enrich the features and reduce the amount of calculation is a problem that needs to be solved at present. Based on this, a neural network for judging the features of Wushu distance teaching target tracking can be designed, as shown in Fig. 1 below.

Fig. 1. Tracking feature judgment neural network

It can be seen from Fig. 1 that the multilayer neural network, that is, the number of hidden layers increases, is used for nonlinear classification problems. In multilayer neural network, the leftmost layer is the input layer, the rightmost layer is the output layer, and the middle layer is collectively called the hidden layer. The layers are connected with each other, and the data is transmitted backward in turn, and the error is transmitted backward and forward [11]. The so-called training is the process of inputting the training data set into the network, comparing the output layer with the training label, and propagating the error back, so as to constantly update the connection weights. In the neural network, the activation function exists between the output of the upper layer and the input of the lower layer, and its function is to apply the neural network to nonlinear relationship fitting. The commonly used activation functions are Sigmoid function, tanh function and modified linear function ReLU. Sigmoid function has advantages in compressing data amplitude, but there is a problem of gradient disappearance in deep neural network. At the same time, due to the exponential operation, when the training process consumes more, draw the real-time tracking decision function change image as shown in Fig. 2 below.

Real-Time Tracking Method of Students’ Targets

699

Fig. 2. Real-time tracking function change image

As can be seen from Fig. 2, since the peak value of the gradient function is 0.25, the gradient decreases by a factor of 0.25 during the propagation, which easily leads to gradient dispersion. Sigmoid function has advantages in compressing data amplitude, but there is a problem of gradient disappearance in deep neural network. At the same time, because of exponential operation, the training process consumes more time. On the basis of the above analysis of the tracking characteristics of Wushu distance teaching objectives, a real-time tracking model of Wushu distance teaching objectives is constructed by deep learning. 2.2 Design of Real-Time Tracking Model of Wushu Distance Teaching Objectives Based on Deep Learning Deep learning is to learn the inherent laws and representation levels of sample data, and the information obtained in the learning process is of great help to the interpretation of data such as words, images and sounds. Its ultimate goal is to make the machine have the ability to analyze and learn like a human being, and recognize data such as words, images and sounds. Therefore, this paper designs a real-time tracking model of Wushu distance teaching goals by using deep learning neural network. First, it is necessary to determine the training index of the model, and the calculation formula is as follows (1): E=

f at t + I

(1)

where at represents the compression amplitude of neural network, f represents the value, t represents the gradient of deep neural network, and I represents the training index. Tanh function is a zero mean function, which will not lead to all positive or all negative local gradients, so the convergence speed will be accelerated and the effect is better than Sigmoid function. However, the gradient of tanh function also disappears, and the calculation amount of exponential operation still exists. At this time, the design index optimization formula is shown in the following (2). tanh =

e − e0 e + e0

(2)

700

J. Zhang and N. Ma

In formula (2), e represents the input value of neural network, e0 represents the optimization parameter, and void convolution is often used in image segmentation, which can keep information better than pool layer in the process of down-sampling, and can restore images more completely after up-sampling. The role of void convolution in the image field is not only in image segmentation, but also can be considered when it is necessary to expand the receptive field, reduce the amount of calculation and obtain small features. The designed real-time tracking method of distance education targets adds void convolution calculation, and the schematic diagram of convolution kernel division is shown in Fig. 3 below.

Fig. 3. Schematic diagram of convolution kernel division

It can be seen from Fig. 3 that hole convolution can increase the receptive field and reduce the feature size with the same amount of calculation. Before the empty convolution was put forward, the downsampling in the convolution network was usually done through the pooling layer. The role of pooling is to reduce the image size and increase the receptive field, but the process of size reduction is accompanied by the loss of information. At present, there are many algorithms in the field of target tracking, and the performance evaluation of the algorithms is reflected in the accuracy, real-time performance and stability of the algorithms. In order to make a fair comparison of various algorithms, the academic community has given public test sets and evaluation indicators, and the tracking model based on this design is shown in (3) below. overlap =

R∪G R∩G

(3)

In the tracking model (3), R represents the real tracking logo and G represents the logo box. In the tracking field, different application scenarios have different requirements for real-time performance, so the limitation of real-time performance depends on specific occasions. In this paper, the basic requirement of real-time performance is 20FPS, and improving the speed on the basis of setting is the requirement to improve the performance of the algorithm.

Real-Time Tracking Method of Students’ Targets

701

Therefore, the real-time tracking model of Wushu distance teaching target is constructed, and on this basis, the model is further optimized through offline training and online updating. 2.3 Optimizing FDT Net Wushu Distance Teaching Goal Tracking Algorithm MDNet is a tracking network structure, using convolution layer to extract features, and full connection layer as classification layer. Different from ordinary classification, it is a binary classification, that is, it is enough to distinguish the target from the background. Multi-domain structure exists to adapt to different tracking scenes. QE trains one FC6 for each input video, thus training multiple FC6 for each calibrated tracking video. The previous Conv1-Conv3 and FC4-FC5 are shared layers. Conduct off-line training and online update on the network. The parameters of convolution layer are obtained during off-line training, while in the tracking process, the parameters of convolution layer are kept unchanged, and the parameters of full connection layer FC4-FC6 are fine-tuned online. The renewal strategy also takes into account the long-term and short-term, and at the same time, it increases the intensive training of hard-to-distinguish samples. The network structure of. NET is designed for tracking tasks. The update strategy is considered carefully and the tracking accuracy is high. However, due to online update and scoring mechanism of multiple candidate boxes similar to the detected pattern, the calculation of tracking process is too large, which reduces the tracking speed. With OTB and VOT2014 as test data sets, it can only reach 1FPS on GPU platform. The reasons for the low speed of MD. NET algorithm can be analyzed as follows: the features of the full connection layer are fully connected, resulting in a large amount of calculation; Multi-domain FC6 layer, learning parameters increase, on the one hand, the convergence speed slows down, on the other hand, a new FC6 is needed for fine-tuning when tracking; It takes time to track and update the parameters of the full connection layer online. At this time, the tracking algorithm of Wushu distance teaching goal is designed as shown in (4) below.  overlap (4) P= 1 s(t)

When the algorithm is optimized, the low-level network is trained first, and then the high-level network is initialized with the weights of the low-level network. The purpose of this measure is to shorten the network training time and make the network converge faster. The reason why VGGNet performs well in the positioning task lies in the use of multiple 3*3 convolution kernels. Two 3*3 convolution layers can be superimposed, which is equivalent to a 5*5 convolution kernel effect, but there are fewer parameters to learn. And after the superposition of several convolutions, many nonlinear transformations have been completed, which has stronger feature learning ability. The reason why VGGNet performs well in classification tasks lies in its structural features. With the deepening of network layer, the number of channels of output features increases and the resolution decreases. The advantage of this design is that the multidimensional features are converted into classification vectors, which is convenient to

702

J. Zhang and N. Ma

connect the full connection layer as the output. Conv1-Conv3 convolution layers come from VGGNet, the input and output of convolution layers. Unlike VGGNet, the input size has been modified. The input size of FDTNet is 107*107, which is set to get the 3*3 feature map after Conv3 layer. The convolution layer is followed by the full connection layer FC4-FC6. FC4 and FC5 each have 512 output units, while FC6 is the second classification layer, which is used to distinguish the target from the background. For the selection of convolution layer, the factors to be considered are the amount of calculation and feature sufficiency. The low convolution layer extracts the line features and edge features in the image, which belongs to the bottom information, while the high convolution layer extracts the semantic information in the image, which belongs to the high features. In the operations of target detection and target recognition, the convolution layer is generally deeper, which is used to extract enough features. However, the deeper the convolution layer is, the larger the calculation amount is, and the worse the real-time performance is. For the subject of target tracking in this paper, it is only necessary to judge the target and background, that is, the requirements for classification tasks are not high. In this paper, three convolution layers are selected, the first two layers are used to extract the edge information of the target, and the third layer is used to extract the global information of the target and increase the receptive field of features. Of course, if you choose a deeper convolution layer, the feature expression ability of the target will be stronger, which is beneficial to the accuracy improvement, but it will increase the computational complexity to a certain extent. In this paper, the selection of convolution layers takes into account the feature sufficiency and computational complexity. According to the analysis of the network structure of MDNet and the reasons for its low speed, the improvement of FDTNet can be summarized as follows: the full connection layer Fc6 is modified into a single domain, the convolution layer Conv1Conv3 extracts the target features, and the full connection layer performs the target and background binary classification operation. ILSVRC2015 detection data set is used as training data set, and the types and quantity of training data sets are increased, so as to ensure that convolution layer has enough feature extraction ability in the training process. The fully connected layer of a single domain reduces the amount of computation and improves the training speed and convergence. Enlarge the training data set to ensure that the improvement does not affect the accuracy. The parameters of the full connection layer are not updated online, the parameters of the convolution layer and the full connection layer are kept fixed, the parameters are not modified, the calculation memory is not occupied, and only the feedforward network calculation is performed. This improvement improves the tracking speed. A new linear update rule is added, and the final position is determined by the weight of the regression position and the position of the previous frame. Adding a superparameter increases the selection improvement of the tracking precision tracking candidate frame, limiting the selection position of the tracking frame to 2*2 times that of the target frame of the previous frame, reducing the number of selection frames, reducing the amount of calculation and improving the tracking speed, but losing the stability. After the training, keep the network parameters unchanged, and give the target frame of the first frame of video. For each subsequent frame, in the area of 2*2 times of the

Real-Time Tracking Method of Students’ Targets

703

target position of the previous frame as the center, 32 candidate frames are selected by multi-dimensional Gaussian distribution in three dimensions of width, height and scale. The candidate frames are input into the network with a uniform size of 107*107, and the output of the network is a two-dimensional vector, which indicates the probability of the target and background corresponding to the input candidate frames. The three candidate boxes with the highest goal scores are the target candidate boxes. For each tracking video, the location of the target in the first frame is known, and the candidate frame of the first frame, the network output corresponding to the first frame and the actual location of the target in the first frame are taken as the data of regression network training, and the trained regression network is used to predict the frame where the target is located in the video. 800 samples are selected from the first frame of the tracking test video sequence, and the selection rules of samples are consistent with those of tracking candidate frames. These 800 samples are taken as the training data set of regression network, and the real position of the target in the first frame of the test video is taken as the label of regression training, and the trained regressor can be directly used for the regression positioning operation of the subsequent frames of the video. Training video tracking through selected data sets, and tracking students’ goals in real time in distance Wushu teaching according to the constructed model. 2.4 Real-Time Tracking of Students’ Goals in Wushu Distance Teaching Real-time tracking of students’ targets in Wushu distance education can also improve the existing tracking algorithms. Traditional tracking algorithms such as KCF are excellent in speed, and correlation filtering is used to obtain the correlation of targets, which has the advantage of small amount of feature calculation. The disadvantages are that there is a boundary effect, the detection range is limited, and the adaptive scene is limited due to the limited feature extraction ability. Combined with KCF’s correlation thought, the Siam-FC tracking network with cross-correlation is considered to be improved. Cross-correlation is used to obtain target correlation, which belongs to sliding window detection, and has the efficient realization of full convolution. The calculation amount is mainly determined by the number and size of features involved in sliding window detection. Siam-FC has relatively good real-time performance, and its accuracy needs to be improved. Then the improvement idea based on Siam-FC is to reduce the amount of feature calculation, improve the robustness of features and obtain richer semantic features. Siam-MF is an extended target tracking network based on Siam-FC, which aims to improve the accuracy and real-time performance while ensuring the tracking stability, so that the tracking algorithm can adapt to various application environments. The convolution layer of Siam-FC comes from AlexNet, and the tracking task is completed by convolution feature correlation. The structure of the network is easy to understand. After feature extraction of the target template and the search area, similarity measurement is carried out, and the location of the target in the search area is located according to the similarity score map. The feature extraction task is completed through the convolution layer of the network, and the whole convolution layer is used as the similarity measurement function. The training set is ILSVRC2015 data set, and the shallow convolution layer is selected to extract features. The training process mainly completes the determination of convolution layer parameters, while the tracking process

704

J. Zhang and N. Ma

does not need to update the parameters. Compared with MDNet, the comprehensive performance is stronger and the accuracy is slightly lower than that of MDNet, but the speed of OTB test set reaches 58FPS. Siam-FC has strong comprehensive performance, but it needs to be improved in speed and accuracy. In-depth analysis of Siam-FC network shows that the extraction ability of convolution layer affects the tracking accuracy, and it can be considered to improve the accuracy by improving the feature extraction ability of convolution layer. The reason that affects the stability lies in the selection of tracking template. Siam-FC chooses the first frame target as the template and does not update it, and the subsequent changes of target shape will affect the stability. The main reason that affects the speed lies in the whole convolution layer, and the more characteristic channels for cross correlation, the greater the amount of calculation. The main line part of the convolution layer of Siam-MF comes from five convolution layers Conv1-Conv5 of AlexNet, and the empty convolution layers Skip1 and Skip2 are added for the output of Conv1 and Conv3 and the feature fusion of Conv5 respectively. For the selection of convolution layer, the calculation amount and feature sufficiency are considered. The lower convolution layer represents edge information and position information, extracts line features and edge features from images, and belongs to the bottom information. The high-level convolution layer extracts the semantic information in the image, which belongs to the high-level features. In the operations of target detection and target recognition, the convolution layer is generally deeper, and enough features are extracted for classification. However, for the network with deeper convolution layer, the greater the computation, the worse the real-time performance. For the subject of target tracking in this paper, only target features need to be extracted, which is not used for classification, that is, the semantic features of the target are not high. In this paper, five layers of convolution are selected to fuse the features of the outputs of Conv1, Conv3 and Conv5, so as to obtain richer target features. The network parameters of Siam-MF are obtained through training. The input of the network is preprocessed pictures, including the target template and the search area. The search area is 2*2 times of the location of the target in the previous frame. After cutting out the candidate box and transforming the size of the original picture, the template with the size of 127*127*3 and the search area with the size of 255*255*3 are obtained. After feature extraction by convolution layer, the correlation between the target and the search area is finally obtained through full convolution. For the tracking process, the feature of the target template and the search area is extracted by convolution layer, and then the cross-correlation analysis is carried out by full convolution layer to get the score map of the target in the search area, which can be considered that the position where the maximum value is located is the target position. In view of the in-depth analysis of Siam-FC, the improvement of the original network can be summarized into the following four points: using feature fusion to obtain more comprehensive features. Output Conv1 through Skip1 layer and Conv3 through Skip2 layer. At the same time, feature fusion is carried out with the output of Conv5 to obtain richer target features and improve tracking accuracy. Introduce void convolution. Void convolution increases the receptive field, reduces the amount of calculation and improves the tracking speed and accuracy. Using void convolution in Skip1 and Skip2 layers can keep the receptive field of features and reduce the amount of calculation. Number of

Real-Time Tracking Method of Students’ Targets

705

channels is reduced. Reducing the number of characteristic channels entering the whole convolution layer will greatly reduce the amount of calculation. The main calculation amount of cross correlation comes from sliding window detection of full convolution network. For Conv5 layer, the number of feature channels is reduced, the reduced features are compensated by feature fusion, and the final features are combined with different levels of data. This operation reduces the amount of calculation and improves the feature extraction ability. The convolution layer extracts the template features and extracts the search area features of each frame. The void convolution is added to the connection layers Skip1 and Skip2 to reduce the number of characteristic channels of Conv5 in AlexNet, and the output of Conv1 through Skip1 is matched with the output of Conv3 through Skip2 and the output of Conv5, so that the receptive field is increased and the amount of calculation is reduced. The process of Siam-MF network is as follows: the target and search area pass through the same convolution network to extract the features of the target and search area; The feature layer of the target and the feature layer of the search area complete the cross correlation through full convolution, and the correlation graph of the target in the search area is obtained, and the position of the maximum value of the correlation graph is the central position of the target in the search area. According to the motion orientation of the extended target in the video, the search area is set to be 2*2 times the size of the target frame. At the same time, the deep convolution network leads to the loss of position information, which is undesirable in the field of tracking. Therefore, in this paper, the output features of different layers of convolution layer are fused, combining the position information of the shallow layer, the extracted features of the middle layer and the semantic information of the deep layer.

3 Test In order to verify the tracking effect of the real-time tracking method of students’ targets in Wushu distance education based on deep learning designed in this paper, this paper compares it with the traditional real-time tracking method of students’ targets, and carries out experiments as follows. 3.1 Prepare for the Experiment The experiment is carried out on ILSVRC2015 dataset. This data set includes 3862 snippets for training, 555 snippets for verification and 937 snippets for testing. Each snippet includes 56 ~ 458 frames of images. Conv1-Conv5 used in feedforward network uses the Conv1-Conv5 layer of AlexNet. In the tracking video sequence, the target is generally not too large, so the input size of the target is set to 127*127, and the input size of the search area twice as large as the target template is set to 255*255. After convolution layer, the features of 6*6 and 22*22 are obtained respectively, and the correlation graph of 17*17 is obtained after full convolution. In order to save the debugging time of parameters, the training parameters of Simese-FC are used as the initial values of the training parameters of this network. After parameter debugging, the stochastic gradient descent method Stochastic Gradient Descent(SGD) method is finally determined to be

706

J. Zhang and N. Ma

used for training. The quantitative evaluation criteria and evaluation parameters designed at this time are shown in Table 1 below. Table 1. Quantitative evaluation criteria and evaluation parameters Teacher/Student

Overlap

Accuracy

FPS

0034004

43.54

92.65

42.56

0034005

42.62

91.54

43.54

0034006

50.34

90.26

41.12

0034007

59.33

92.45

42.54

0034008

69.45

94.16

43.15

0034009

72.33

92.49

44.66

0034010

59.41

94.56

42.41

0034011

49.38

92.61

46.65

0034012

47.64

93.46

45.34

It can be seen from Table 1 that the accuracy and stability of Siam-MF and Siam-FC are the same for the same tracking video, and the real-time performance of Siam-MF algorithm is improved to a certain extent compared with Siam-FC. According to this evaluation standard and index, subsequent real-time target tracking experiments can be carried out. 3.2 Experimental Results and Discussion In the preparation of the above experiment, we use the real-time tracking method of students’ targets in Wushu distance education based on deep learning designed in this paper and the traditional real-time tracking method of students’ targets to track, and record the delay of the two methods. The experimental results are shown in Table 2 below. From Table 2, it can be seen that the designed real-time tracking method of students’ targets in Wushu distance education has a short tracking delay, which proves that it has good tracking effect, effectiveness and certain application value. On the basis of the above experiments, the methods of references [9] and [10] are introduced as comparison methods, and the tracking accuracy is taken as the index. The results are shown in the following Table 3: From the analysis of Table 3, it can be seen that this method has the highest real-time tracking accuracy, with an average of 95.08%, while the real-time tracking accuracy of the methods in reference [9] and reference [10] are 90.02% and 88% respectively, which indicates that this method has more reliable real-time tracking results.

Real-Time Tracking Method of Students’ Targets

707

Table 2. Experimental results/ms Tracking times

This paper designs a real-time tracking Tracking delay of students’ target method for students’ targets in Wushu real-time tracking method based on distance education based on deep traditional Wushu distance teaching learning, tracking delay

1

0.56

1.65

2

0.44

1.34

3

0.43

1.69

4

0.39

1.47

5

0.61

1.33

6

0.25

1.48

7

0.44

1.94

Table 3. Tracking accuracy of different methods/% Group

Methods of this paper

The method in reference [9]

The method in reference [10]

1

96.1

90.6

88.5

2

95.3

91.1

87.6

3

94.7

89.6

89.1

4

93.8

88.7

87.3

5

95.5

90.1

87.5

4 Conclusion Wushu bears the heavy traditional Chinese culture, and schools are an important territory for inheriting Wushu culture. There are many problems in Wushu teaching. Yu Jing scholars believe that the weak teachers and single teaching content are the main problems in Wushu teaching today. The lack of teachers leads to the inability of our Wushu learning to achieve high-quality and high-level exercises, thus failing to satisfy the students’ awareness of lifelong physical education; Li Benyi, a scholar, explored the dilemma of Wushu teaching from the perspective of human studies, that is, ignored the dominant position of human beings; In the process of Wushu teaching, it is necessary to meet students’ individual needs, respect students’ dominant position and pay attention to students’ overall development. Therefore, it is necessary to build an online teaching platform to realize Wushu distance teaching. In order to ensure the effect of Wushu distance teaching, this paper designs a real-time tracking method of distance teaching objectives, and carries out experiments. The results show that the tracking time delay of the designed distance tracking method is short, which proves that its tracking effect is good and effective, and it has certain application value and can be used as a reference for subsequent Wushu teaching. However, there are few experimental data in the application

708

J. Zhang and N. Ma

of this method, so it is not certain that it is suitable for a large number of college martial arts teaching. In the future research, it is necessary to conduct more extensive experiments and improve them to improve their applicability.

References 1. Feng, J., Zhao, H.: Dynamic nodes collaboration for target tracking in wireless sensor networks. IEEE Sens. J. 99, 1 (2021) 2. Gong, Y., Cui, C.: A measurement set partitioning algorithm based on CFSFDP for multiple extended target tracking in PHD Filter. Radioengineering, 30(2), 407–416 (2021) 3. Li, Z., Chen, X., Zha, Z.: Design of standoff cooperative target-tracking guidance laws for autonomous unmanned aerial vehicles. Math. Probl. Eng. 2021(2), 1–14 (2021) 4. Wang, X., Xie, W., Luo, J., et al.: Labeled multi-bernoulli maneuvering target tracking algorithm via TSK iterative regression model. Chin. J. Electron. 31(2), 227–239 (2022) 5. Li, S., Feng, X., Deng, Z., et al.: Minimum error entropy based multiple model estimation for multisensor hybrid uncertain target tracking systems. IET Signal Processing, 14(3) (2020) 6. Sun, C., Wan, Z., Huang, H., et al.: Intelligent target visual tracking and control strategy for open frame underwater vehicles. Robotica, 1–15 (2021) 7. Ma, C., Huang, J.B., Yang, X., et al.: Hierarchical convolutional features for visual tracking. In: 2015 IEEE International Conference on Computer Vision (ICCV). IEEE Computer Society (2015) 8. Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. Comput. Sci. (2014) 9. Wang, L., Liu, T., Wang, G., et al.: Video tracking using learned hierarchical features. IEEE Trans. Image Process. 24(4), 1424–1435 (2015) 10. Nam, H., Han, B.: Learming multi-domain convolutional neural networks for visual tracking (2015) 11. Tianhua, C., Siqun, Z., Yuxiao, L.: Semantic segmentation of remote sensing images based on improved deep neural network. Comput. Simul. 38(12), 27–32 (2021)

Research on Anomaly Detection of Distributed Intelligent Teaching System Based on Cloud Computing Fayue Zheng(B) , Lei Ma, Hongxue Yang, and Leiguang Liu Beijing Polytechnic, Beijing 100016, China [email protected]

Abstract. The traditional anomaly detection method of intelligent teaching system has some problems, such as poor accuracy and response efficiency. Therefore, this paper proposes a distributed anomaly detection method of intelligent teaching system based on cloud computing. Collect the abnormal data of distributed intelligent teaching system through cloud computing method, calculate the local reachable density according to Gaussian distribution function, build a data management model, and use distributed technology to locate and manage the abnormal area of teaching data, so as to achieve the goal of data detection and identification. The experimental results show that this method can effectively improve the recall rate of anomaly detection data in intelligent teaching system, and the response efficiency has been effectively improved. Keywords: Cloud computing · Distributed · Intelligent teaching · Anomaly detection

1 Introduction With the progress of science and technology and the rapid development of network technology, the information industry and its application have been greatly developed. Enterprises and institutions such as government, finance, education and individual users are more and more dependent on the network. At the same time, it also brings hidden dangers of information security. How to ensure the security of network and information system has become a highly valued problem [1–3]. As an active security protection technology, intrusion detection can detect and identify external or internal abnormal activities or intrusion behaviors, such as malicious use or destruction of computer and network resources, unauthorized access of internal users, etc. it has become a useful supplement to the traditional computer security technology, It is a new hotspot in the field of network security. Reference [4] proposes a distributed intelligent teaching system anomaly detection method based on deep learning, which obtains the data transmitted by the intelligent teaching system through the data mining method, constructs the network training model © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 709–723, 2022. https://doi.org/10.1007/978-3-031-21161-4_54

710

F. Zheng et al.

according to the deep learning method, and realizes the anomaly detection of the intelligent teaching system. This method can improve the monitoring response time, but the anomaly monitoring recall rate is poor. Reference [5] proposes a teaching system anomaly detection method based on ZigBee technology, which uses sensors to obtain abnormal data of the teaching system, and uses ZigBee technology to repair abnormal problems in the teaching system. This method can improve the accuracy of anomaly monitoring, but it is time-consuming. Reference [6] proposes a teaching system anomaly detection method based on blockchain technology. The clustering center of teaching system anomaly is obtained through data clustering method, and the blockchain technology is used to detect the learning system anomaly. This method can improve the anomaly detection effect, but the response speed is poor. To solve the above problems, this paper proposes an anomaly detection method for distributed intelligent teaching system based on cloud computing. Collect the abnormal data of the distributed intelligent teaching system through cloud computing method, calculate the local reachable density according to the Gaussian distribution function, build a data management model, use the distributed technology to locate and manage the abnormal area of teaching data, achieve the goal of data detection and identification, and effectively improve the accuracy of abnormal detection of the intelligent teaching system.

2 Anomaly Detection of Distributed Intelligent Teaching System 2.1 Recognition of Abnormal Information in Intelligent Teaching System The data that intrusion detection needs to analyze is collectively referred to as events. It can be packets in the network or information obtained from host system logs and other ways. The purpose of event generator is to obtain events from the whole computing environment and provide this event to other parts of the system in a specific format [7]. The event analyzer analyzes the data to determine whether it is violation, anomaly or intrusion, and converts the judgment result into warning information. The response unit responds according to the warning information. It can make a strong response such as

application layer

Consensus layer

User registration login

Data local storage

POW mechanism

Data trusted storage

Data sharing on the chain

network layer

Data layer

P2P protocol

hash function

Broadcasting mechanism

digital signature

Verification algorithm

Merkel tree

Ethash algorithm

Fig. 1. The structure design of the information database of the intelligent teaching system

Research on Anomaly Detection of Distributed Intelligent Teaching

711

cutting off the connection and changing the file attributes, or it can be just a simple alarm. It is an active weapon in intrusion detection. The online learning behavior data storage and sharing model based on blockchain technology optimizes the information database structure of intelligent teaching system, as shown in the Fig. 1: The abnormal event database based on cloud computing is a place to store various intermediate and final data. It receives data from the event generator or analyzer and saves it for a long time [8]. It can be a complex database or a simple text file. In this model, the first three appear in the form of program, while the last one often appears in the form of file or data flow. The specific model is as follows (Fig. 2):

Event Analyzers

Original event source

Response unit

event generators

Event database

Fig. 2. Intrusion detection model based on cloud computing

The anomaly detection system of intelligent teaching system attempts to establish a feature prototype corresponding to “normal activities”, and then mark all behaviors that are “very different” from the established feature prototype as anomalies. Deep learning is a meaningful learning process, which requires the joint participation of teachers and students. Teachers play the role of guides. Students actively participate, experience success and grow around the challenging learning theme [9]. Students not only master the core knowledge, essence and thinking methods of the subject, but also form a high learning motivation, positive learning attitude and correct values. The table is a comparison of cognitive goals between deep learning and shallow learning (Table 1). Table 1. Comparison of cognitive objectives between deep learning and shallow learning Learning type

Target hierarchy

Connotation

Deep learning

Application

Apply the learned skills to the new situation

Analysis

Decompose the material into elements and clarify the relationship between elements and the whole

Evaluate

Make value judgment on the learned knowledge according to the criteria (continued)

712

F. Zheng et al. Table 1. (continued)

Learning type

Shallow learning

Target hierarchy

Connotation

Create

Integrate all elements into a consistent whole to form a new model or structure

Memory

Extracting relevant information from long-term and short-term memory

Understand

Understanding the meaning of knowledge from teaching information

It is assumed that all abnormal behaviors are different from normal behaviors. If the trajectory of the normal behavior of the system can be established, all system states different from the normal trajectory can be regarded as suspicious attempts in theory. As shown in the Fig. 3:

Intrusion behavior

anomaly detection

Log file / network data

Normal behavior description library

Fig. 3. Cloud computing anomaly detection model

In the process of processing multidimensional data, the isolated forest algorithm uses the method of randomly selecting attributes to build a tree, and finally integrates the results of each tree to judge the anomaly, while ignoring the problem that each instance in the multidimensional data has different anomaly degrees for the randomly selected attributes. Therefore, in general, the abnormal score detected only by random attributes will still be inaccurate, which needs to be further improved. 2.2 The Abnormal Area Location Algorithm of the Intelligent Teaching System When calculating the learning credit of online learning platform learners, it is necessary to obtain the variable values involved in the relevant formulas first. The background database will collect the data required for anomaly detection while obtaining the source data [10]. Referring to the relevant data of anomaly detection, on the one hand, modify the relevant record data of abnormal learning behavior, on the other hand, deduct the

Research on Anomaly Detection of Distributed Intelligent Teaching

713

corresponding anomaly points for the calculated five-dimensional evaluation values, and then obtain the updated five-dimensional evaluation values Pa, Po, Re, Qu and Ln. After calculating the relevant weights, you can calculate the learning credit of each learner participating in the platform learning activities. The calculation formula is: SC = Pa κ1 + Po κ2 + Re κ3 + Qu κ4 + Ln κ5

(1)

Among them, κ1 , κ2 , κ3 , κ4 , κ5 are the weights of the five-dimensional evaluation index of learning credit, namely the weight of the five dimensions of Pa, Po, Re, Qu, and Ln relative to the weight of learning credit SC. The statistics-based method is one of the simplest and basic methods in anomaly detection, and its principle is also very easy to understand [11]. This method requires that the data set e must present a known distribution or meet certain laws, and then find the laws that do not meet the requirements. Point. The most common known distribution is the Gaussian distribution. The following uses it as an example to briefly describe the idea of anomaly detection based on statistics. According to the original data set, find the expected x and variance μ to determine the Gaussian distribution function, as   2   1 − (x−μ) 2  (2) e 2σ − SC(x − μ) f (x) =  √ −e 3π σ When e < σ or e > σ , we find that the probability of its appearance is very small, so we consider the outliers in this interval. This method has some advantages, but there are also disadvantages. In terms of advantages, the outliers obtained by this method have high credibility, and the outliers are intuitively visible. However, this method is limited to conform to a certain distribution [12]. If it does not conform to known distributions and laws, this method is not suitable. Secondly, this method is more difficult in parameter selection, and because outliers are also involved in the construction of the model, if there are too many outliers, it will have a greater impact on the selection of model parameters. Furthermore, this method is not suitable for the detection of high-dimensional data. The basic idea of the distance-based anomaly detection method is to find the distances of Xi points closest to the current point and sum them. The specified distance and the largest Yi points are the abnormal points. When measuring the distance of anomalous points, it usually refers to Euclidean distance, manhattan distance or Mahalanobis distance. The calculation formulas are as follows:  ⎧  n ⎪ 

⎪ ⎪ ⎪ disED = f (x) + (Xi − Yi )2 − SC ⎪ ⎨ i=1 (3) ⎪ n ⎪

⎪ ⎪ ⎪ |Xi − Yi | + SC ⎩ disMHD = f (x) − i=1

The distance anomaly detection algorithm is k-nearest neighbor algorithm (kNN). The abnormal point (ai , bi ) can be obtained by finding the largest n points with the average distance of k nearest neighbors [13]. The formula for the average distance of k

714

F. Zheng et al.

nearest neighbors of point P is as follows:  p− dis(ai , bi ) D(p) =

qi ∈Nk (p)

|disED − disMHD |

− Nk (x)

(4)

Among them, Nk (x) represents k-nearest neighbor, that is, the set of objects whose distance from point p does not exceed k-nearest neighbor. Compared with statisticsbased detection methods, distance-based anomaly detection methods can handle multidimensional data, but there are also certain problems: first, the choice of distance calculation method and the selection of parameter lrdk (p) are difficult; second, it cannot distinguish local outliers. Point. In order to solve the problem that the distance-based anomaly detection method cannot accurately distinguish the local outliers, a densitybased outlier method is proposed. The most representative one is the LOF algorithm. The main idea is to judge the abnormal situation by comparing the density of each point with its neighboring points. The smaller the density, the greater the possibility of anomaly. This density is measured by the local outlier factor. The local outlier factor of point p in the LOF algorithm is expressed as lrd k (o)

LOFk (p) = D(p) −

o∈Nk (p)

1/lrdk (p)

|Nk (x) − f (x)|

− SC

(5)

Among them, Nk (p) is the k distance field, that is, all points within the k distance of point p; lrd(p) refers to the local reachable density, expressed as reach -  dist k (p,o)

lrdk (p) = 1/

o∈Nk (p)

|Nk (p)|

(6)

Among them, reach-distr (pO) represents the reachable distance, that is, the kth reachable distance from o to p. Taking into account the controllability of the data, referring to the existing data collection methods and common behaviors of learners, data collection is performed on several behavior indicators of the evaluation model through Web Server log analysis technology and embedded point technology. Next, the following will focus on introducing the web-side data, online operation data and discussing the collection process of interactive data. The corresponding detailed indicators are shown in the Table 2. Combined with the weight of refined behavior index elements and standardized learning behavior data, the weight is calculated by structural equation model. The figure shows various learning services provided by the application layer module for users. The main functions include registration, online learning behavior acquisition, learning credit data storage and learning credit data sharing (Fig. 4). Judge the results of anomaly detection experiments from four aspects. Further judge the accuracy of each deep learning network in the migration model and the newly-built U-CONVLSTM model to learn the normal behavior characteristics of the student; then use the loss value obtained in the training model for each test segment of the test set

Research on Anomaly Detection of Distributed Intelligent Teaching

715

Table 2. Correspondence table of data collection types and data collection items Data collection type

Data collection item

Web side data

IP address, login time, course start time, deadline, study time, number of videos watched, course study days

Online operation data

Click count, collection count, job score, time consumption after video viewing

Discuss interactive data

Original sharing and forwarding times; Times of teacher-student interaction

Watch Video

Register login

Participate in the test

Knowledge sharing Online learning behavior acquisition

Participate in online classroom discussions Local storage of credit data Application layer function

Learning credit calculation storage

Learn to upload credit data Trusted storage of credit data Teachers spit out course information Students view the evaluation report

School query chain data Learning credit data sharing

Employer application query data Regulatory data of education authorities

Fig. 4. Application layer intelligent detection function design diagram

to determine at which moment the student is Abnormal behavior occurs; after the loss value is converted into an abnormal score, it can be judged whether the student has abnormal behavior; finally, the test results of all test set data in the migration model and the U-shaped model are counted. In summary, it is concluded that the learning effect and detection effect of the new U-shaped autoencoder is better than the migration model (Fig. 5).

716

F. Zheng et al. Space encoder

Input video data

Time self encoder Convolution: 6 * 6 filter: 65 step 3

Convolution: 11 * 11 filter: 130 step size: 5

Convlstm: 4 * 4 filter: 65

Convlstm: 4 * 4 filter: 33

Output configuration result

Convolution: 6 * 6 filter: 65 step 3

Convolution: 11 * 11 filter: 130 step size: 5

Convlstm: 4 * 4 filter: 65 Space encoder

Fig. 5. The learning effect and detection effect are better than the migration model

Unbalanced data means that the number of positive samples is far less than the number of negative samples. Due to this imbalance in number, a more reasonable method needs to be found to analyze these data. Therefore, the analysis method of accuracy, which does not distinguish the importance of each class, is obviously not suitable for analyzing unbalanced data sets. In unbalanced data sets, a small number of analogies and a large number of classes have more research value. Therefore, in binary classification, rare classes are usually marked as positive classes, while most classes are marked as negative classes. The following table shows the confusion matrix of the number of positive and negative samples correctly predicted and incorrectly predicted by the model (Table 3). Table 3. Confusion matrix Real class

Prediction class Positive example (+)

Counterexample (-)

Positive example (+)

f + + (AY)

f + + (IN)

Counterexample (-)

f- + (IY)

f--(AN)

The detection results of isolated forest algorithm are fuzzy processed by using the idea of fuzzy logic. The idea of the so-called fuzzy logic is based on the theory of fuzzy mathematics and fuzzy relationship synthesis principle, and uses scientific means to accurately describe some things with fuzzy conceptual boundaries and difficult to quantify. Therefore, it can be detected from multiple angles to ensure the accuracy and effectiveness of anomaly detection results. 2.3 Implementation of Anomaly Detection in a Distributed Intelligent Teaching System According to the analysis of business requirements, this network traffic security system consists of four subsystems: domain name anomaly detection, forum access monitoring, unhealthy information release traceability, and network abnormal behavior detection.

Research on Anomaly Detection of Distributed Intelligent Teaching

717

The functional structure diagram of the entire system is shown in the figure. Collecting and reading network traffic logs and other information are their common functions, and different subsystems will process different contents differently to obtain the required results (Fig. 6). Classification feature extraction Establishment of classification model Domain name exception detection

DNS log analysis and processing Malicious domain name detection Test result output Wandering log analysis and processing Frequent itemset establishment

Forum access monitoring

Word segmentation tool processing Hotspot keyword output

safety system Traffic log analysis and processing

Release and tracking of bad information

Word segmentation tool processing Match with sensitive Thesaurus Corresponding IP output Traffic log analysis and processing

Network abnormal behavior detection

Short time data analysis and summary Statistics of suspicious users Threshold data output exceeded

Fig. 6. Functional structure of network traffic security system

The malicious domain name detection module first preprocesses the network traffic log data collected by the traffic collector, extracts the characteristic information that needs to be verified, and stores it on the ES, and then uses the previously established classification model to read the data for classification, and calculate whether these domain names are Malicious domain name, output the malicious domain name and the IP information associated with it, stored in the cluster, and can be downloaded to the machine for use (Fig. 7). Malicious domain names and normal domain names can be judged based on many factors. These judgment factors are human experience. They are expressed as features, and data feature identification and detection are performed to ensure detection accuracy and efficiency.

3 Analysis of Experimental Results In order to verify the effectiveness of the algorithm proposed in this paper, this paper uses a real data set in the UC1 data set to conduct a comparative experiment. The processing time is compared and analyzed, and the RoC of each algorithm is drawn. For the method in this paper, the commonly used setting k = 10 is used in the experiment. This domain

718

F. Zheng et al. Training set reading Classification feature extraction

Establishment of classification model

Model establishment Model evaluation

Domain name anomaly detection system

Network traffic log collection Malicious domain name detection

Domain name data detection Malicious domain name and IP output

Fig. 7. Functional structure diagram of domain name anomaly detection system

name anomaly detection system is a subsystem of the network traffic security system. The network traffic security system needs to be built on the company’s big data cluster. The big data cluster built to test the system functions this time consists of 3 nodes, using the company’s CAS The platform creates three virtual machines to build a domain name anomaly detection system. Because the system is developed in Python language, in order to ensure the existence of Python dependent libraries, anaconda2 will be installed, and its specific detailed configuration information will be installed. Use a university’s 2019 grades in 2020 and 2020 and 2019 grades in 2019 and 2020 into several groups as data sets (20,000 data), and take the top 1.5% of the larger T(x) data for false detection rate statistics, The parameters are as shown in the Table 4: Table 4. Experimental parameter setting table Experiment No

1

2

3

K-means

1

0

0

PFT-OI A

0.7

0.3

0.3

Paper method

0.5

0.4

0.4

The experimental data adopts the teaching evaluation data of students in a university. The original data is composed of four fields. The clustering effect of abnormal detection distance of teaching information is as follows (Fig. 8): The traditional k-means clustering method has a higher false detection rate for anomaly detection. Compared with the traditional k-means clustering method, the false detection rate of this method is significantly lower, especially when a = 0.4, β = 0.3, y = 0.3, the false detection rate reaches a low level. Based on this, the false detection rate

Research on Anomaly Detection of Distributed Intelligent Teaching

719

Cluster value 10

5

-5

-10

-15 -60

-50

-30

-40

-20

-10

0 information content

Fig. 8. Distance clustering effect of abnormal detection of teaching information

of data detection under different methods is compared and recorded, and the details are as follows (Fig. 9):

30%

20%

10%

0%

0

3 mean value PFT-OI1

6

9 Data volume (GB) Paper method

Fig. 9. Statistical chart of system data false detection rate

Using the same data set, it can be seen from the above two experiments that the proposed method has a lower false detection rate compared with the traditional method, which greatly improves the accuracy of anomaly detection. After the establishment of the classification model, the indispensable evaluation of the effect of the classification model shall be carried out. The evaluation sub module mainly uses manual verification to statistically calculate the accuracy rate and recall rate of the classification model. The

720

F. Zheng et al.

main content of its function test is to verify whether the accuracy rate and recall rate are within the expected range to determine whether the classification model meets the requirements. The design of its test cases is shown in the Table 5. Table 5. Functional test cases of the classification model establishment module Case number

2.0

Function description Read the collected feature data, train the classification model, read the domain name data to be tested, and judge whether the output result is correct Design purpose

Verify whether the classification model can be trained correctly to ensure the correctness of the domain name detection classification model

Preconditions

The python environment configuration is normal, and the server has the characteristic data collected by the training model and the test domain name data needed to verify the rationality of the model

Use case design

Using the collected features to train the classification model, the correctness of the trained model is verified by the test data

Expected results

The program can run normally, and the training can get the classification model, which can classify the domain name

Test result

Generate the classification model correctly and output the test results normally

Test status

Adopt

Through the statistics of 7000 DNS log data used in the test, a total of 173 malicious domain names are obtained. According to the classification models trained on training sets of different sizes, the statistics of the detection output results of these log data are compared and counterattack traditional clustering. The method and the images of accuracy and recall rates published in this article, the specific results are shown below (Fig. 10). For the same sample data set, four different algorithms are used to analyze it, and the sensitivity and specificity of different models are calculated according to the result distribution, and four ROC curves are obtained. The ROC curve can graphically describe the relationship between the true rate and false positive rate of a classifier, so it is often used to compare the classification accuracy of different classifiers, and the closer the curve is to the top of the ordinate, the sensitivity and specificity of the sample The higher the degree, the higher the accuracy of the classifier. As shown in the figure, the method proposed in this paper is closest to the top of the ordinate, and the local anomaly factor algorithm is closest to the diagonal, so this paper has better classification performance and higher classification accuracy (Fig. 11). In the analysis stage of the experimental results, the conventional graphical method is also used to show the situation of the experimental results. The figure shows the experimental results of anomaly detection false detection rate experiment. The statistical

Research on Anomaly Detection of Distributed Intelligent Teaching 1 Paper method

Accuracy rate

0.9

traditional method 0.8

0.7 50

100

150

200

Training set size 1 Paper method

recall

0.9

traditional method

0.8

0.7 50

100

150

200

Training set size

Fig. 10. Comparison test results of precision rate and recall rate 1.5

True positive rate

1.2

0.9

0.6

Paper method Zigbee technology Deep learning

0.3

0

Blockchain Technology 0

0.2

0.4

0.6

0.8

False positive rate

Fig. 11. ROC curve of anomaly detection algorithm

1

721

722

F. Zheng et al.

value of false detection rate is obtained by comparing experimental designs with different weights (Fig. 12). noise factor

30%

20%

10%

0% 0

3

9

6

mean value

Paper method

PFT-OI1

Fig. 12. Comparison test results

In the experimental stage, this method can clearly show the abnormal degree judged by the algorithm, and can more quickly and accurately achieve the research goal of accurate identification and rapid positioning of the number of teaching systems, so as to fully meet the research requirements.

4 Conclusion Taking anomaly data detection as the research object and using cloud computing and fuzzy membership function as tools, this paper mainly makes an in-depth research on the widely used isolated forest algorithm. When the existing algorithms carry out anomaly detection, they randomly select attributes to build trees, while ignoring the anomaly degree of each data for the selected attributes. A data anomaly detection method based on cloud computing algorithm is proposed. From multiple dimensions, the membership of the detection results of each one-dimensional attribute is judged, and finally the final evaluation result is obtained by fuzzy operation with the fuzzy matrix. This method has achieved good detection accuracy in distributed intelligent teaching system anomaly detection, but the complexity of the algorithm needs to be further simplified to effectively improve the detection time.

Research on Anomaly Detection of Distributed Intelligent Teaching

723

References 1. Lei, H.: Exploration and practice of abnormal flow detection in intelligent campus construction based on data mining. Taiwan Strait Sci. Technol. Ind. 35(09), 55–57 (2022) 2. Zhe, C., Chong, W., Zhiqiu, H.: Program design course teaching system based on dynamic analysis. Comput. Syst. Appl. 29(10), 114–119 (2020) 3. Yangyang, F.: Design of malicious tamper detection system for teaching system based on web crawler. Digital Commun. World 04, 122–123 (2020) 4. Lin, P.H., Chen, S.Y.: Design and evaluation of a deep learning recommendation based augmented reality system for teaching programming and computational thinking. IEEE Access, 99, 1–8 (2020) 5. Jun, L., Liyan, Z., Xiaoyuan, L., et al.: Cyclic redundancy check method of serial communication data flow based on ZigBee. Comput. Simul. 38(1), 226–230 (2021) 6. Jin, R., Wei, B., Luo, Y., et al.: Blockchain-based data collection with efficient anomaly detection for estimating battery state-of-health. IEEE Sens. J. 99, 1 (2021) 7. Wójcik, K., Piekarczyk, M.: Machine learning methodology in a system applying the adaptive strategy for teaching human motions. Sensors 20(1), 314–326 (2020) 8. Xiaoming, L., Yi, Y., Yue, Z.: Simulation of large data multi-resolution acquisition method based on Java3D network. Comput. Simul. 37(2), 5–18 (2020) 9. Kumar, R., Patil, O., Karthik, N.S., et al.: A machine vision-based cyber-physical production system for energy efficiency and enhanced teaching-learning using a learning factory. Procedia CIRP 98(5), 424–429 (2021) 10. Ll, A., Yu, P.A., Lwb, C., et al.: Improving EGT sensing data anomaly detection of aircraft auxiliary power unit. Chin. J. Aeronaut. 33(2), 448–455 (2020) 11. Wu, L.: Student model construction of intelligent teaching system based on Bayesian network. Pers. Ubiquit. Comput. 24(3), 419–428 (2020) 12. Jiong, G., Qian, R., Jianjiang, H.: A summary of foreign research on the application of artificial intelligence in teaching. Audio Vis. Educ. Res. 41(2), 9–18 (2020) 13. Jian, M., Ligang, N., Xin, L.: Design of laboratory intelligent teaching system in key universities based on multimedia and network technology. Mod. Electron. Technol. 44(20), 6–16 (2021)

Abnormal Traffic Detection Method of Educational Network Based on Cluster Analysis Technology Lei Ma(B) , Jianxing Yang, and Fayue Zheng Beijing Polytechnic, Beijing 100016, China [email protected]

Abstract. Due to the long detection time of the existing education network abnormal traffic detection methods, the detection accuracy of individual abnormal traffic information is relatively low, which is easy to threaten the operation security of the education network. Therefore, an education network abnormal traffic detection method based on cluster analysis technology is proposed. According to the standardization principle, the key abnormal traffic information is processed, and then according to the definition of subspace clustering, the specific numerical results of the cluster similarity index are calculated to complete the clustering and analysis of the abnormal traffic of the education network. On this basis, execute the abnormal flow information extraction instruction, combine the known median absolute deviation measurement conditions, analyze the minimum covariance results of the detection results, and realize the smooth application of the education network abnormal flow detection method based on the cluster analysis technology. The experimental results show that, compared with traditional detection methods, under the effect of cluster analysis technology, the maximum value of abnormal traffic information of education network can reach 14.1 × 10−7 T per unit time, which is in line with the reality of rapid detection of abnormal traffic information of education network Application requirements can better avoid the threat and impact of abnormal information parameters on the security of the education network. Keywords: Cluster analysis · Network flow · Anomaly detection · Standardization of specifications · Median deviation · Minimum covariance

1 Introduction Flow detection refers to the monitoring of data flow, usually including the speed of outgoing data, incoming data and total flow. Traffic detection sometimes refers to monitoring and filtering the user’s data traffic to effectively master the bad information within the monitoring range, which is commonly used in the professional term of network security. From the perspective of the whole, application and other aspects of network traffic. Through the analysis of data and indicators reflecting the real situation of the network, © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 724–736, 2022. https://doi.org/10.1007/978-3-031-21161-4_55

Abnormal Traffic Detection Method of Educational Network

725

such as exit bandwidth utilization, application traffic ranking, application traffic ranking, main traffic flow direction, main use ports, network quality, connection success rate and data retransmission rate, it provides a basis for setting traffic management strategy [1]. All real-time analysis is refreshed every fixed time to facilitate the timely discovery of network problems; It can quickly locate the host or application with abnormality or failure through several mouse clicks. Multiple management devices can be virtualized on one device, each virtual device manages a dedicated line, and one device can realize independent analysis, guarantee and management of multiple dedicated lines at the same time. Cluster analysis refers to the analysis process of grouping a collection of physical or abstract objects into multiple classes composed of similar objects. It is an important human behavior. The goal of cluster analysis is to collect data to classify on the basis of similarity. Clustering comes from many fields, including mathematics, computer science, statistics, biology and economics. In different application fields, many clustering technologies have been developed. These technologies and methods are used to describe data, measure the similarity between different data sources, and classify data sources into different clusters. Clustering is a process of classifying data into different classes or clusters, so the objects in the same cluster are very similar, while the objects in different clusters are very different [2]. For the abnormal traffic detection and screening behavior in the education network, although the traditional application methods can define the location of data information, the mutual measurement relationship between information and information is not emphasized. The detection time for individual abnormal traffic information is long and the accuracy level is relatively low, which makes the education Internet vulnerable to the threat of abnormal traffic information during the operation. In order to solve the above problems, this paper proposes a new abnormal traffic detection method based on cluster analysis technology.

2 Cluster Processing and Analysis of Abnormal Traffic in Education Network Starting from three perspectives: standardization of abnormal traffic information specifications, subspace clustering processing, and cluster similarity calculation, the clustering processing and analysis of abnormal traffic in the education network are completed. 2.1 Standardization of Abnormal Flow Information For random data objects in the education network, the corresponding traffic data may have multiple attributes at the same time, and the dimension levels between these attribute values are not completely different, which indirectly leads to the quantitative indicators of different attributes. The difference in numerical values is very large, which in turn causes those specifications with larger orders of magnitude to have a great impact on the clustering results, while specifications with smaller orders of magnitude have little effect on the clustering results, which is also caused by clustering errors. The direct cause [3]. In order to solve the problem caused by the non-uniform dimension, the abnormal traffic information of the education network to be detected should be dimensionlessly

726

L. Ma et al.

operated before clustering, so that the value of each attribute is kept within a uniform numerical range. It is stipulated that the value of abnormal traffic information s of the education network always belongs to the set of natural numbers N , Ps represents the clustering execution intensity index of the information s, v0 represents the information attribute discriminant value under the action of the clustering algorithm, and vs represents the attribute discriminant value of the information s. Combining the above physical quantities, the standard processing principle of the clustering specification of abnormal traffic information of the education network can be expressed as: +∞ 

Js =

s=1

Ps

+∞ 

||vs − v0 ||2

s=1

rs − r0

(1)

In the above formula, r0 represents the random variable definition coefficient of the abnormal flow information of the education network under the effect of clustering, and rs represents the variable definition coefficient of the abnormal flow information s of the education network under the effect of clustering. Cluster analysis is neither a formula nor a random application idea, but a complete data information processing process. Input the data into the established processing environment, and the expected results will be obtained after the operation of clustering law. Clustering processing is a complex process with many steps, and each step needs to be repeated constantly to achieve accurate results. Because the clustering analysis algorithm only has the ability of hard division, each data can only belong to one cluster category in the operation process. To measure whether the clustering processing method can distinguish the abnormal traffic information of education network, we should not only standardize, but also retain the initial resolution of the original attributes. 2.2 Subspace Clustering Subspace clustering is also called the selection of information parameter feature space. For the education network environment, the initial abnormal traffic information storage space is divided into many different subspaces. On this basis, only the important subspaces need to be clustered. The clustering analysis algorithm usually uses greedy value to define different feature subspaces, then evaluates these subspaces storing abnormal traffic information of education network through corresponding measurement criteria, and finally obtains the required clustering conditions. To sum up, the idea of these subspace processing is to find those sparse regions in the feature subspace, then remove the sparse regions, and the remaining data is dense regions, which is the desired clusters [4]. In other words, clustering is only carried out in the feature subspace, and those unimportant spaces are artificially ignored. Therefore, the final operation result of clustering algorithm may not be complete, but the ignored result can only reflect the existence form of a small part of abnormal traffic information in education network. The subspace is carried out on the basis of considering the impact of each dimension on the clustering results. It not only does not lose the impact of any dimension on the clustering results, but also ensures the real-time storage capacity of abnormal traffic

Abnormal Traffic Detection Method of Educational Network

727

information in the education network. At present, the clustering processing of some subspaces is to map the data to each one-dimensional space for clustering, and then merge the clustering results of each space according to some law to obtain the final clustering results. However, due to the clustering of each space, the amount of calculation is ten times large, especially for high-dimensional data, and when the data is reduced to one-dimensional, The authenticity of abnormal traffic information in some educational networks may be affected. Therefore, in order to make the final detection results have strong reference value, the subspace parameters of data information should be clustered in a high-dimensional environment [5]. Let a denote the dimensionality definition coefficient of the clustering subspace, e˙ denote the normal vector of the abnormal flow information of the education network to be checked, and e denote the inverse function of the normal vector e˙ . Combining the above physical quantities, the subspace definition condition can be expressed as: D=

+∞ 

+∞ √ 2  2 (˙e)|a−1| / (e ) a +1

a=1

a=1

(2)

It is stipulated that c1 and c2 represent two unrelated educational network abnormal flow information definition items, and the improper condition of c1 = c2 is always established, d represents the dense planning coefficient of the data information parameter in the subspace environment, and α represents the data information in the subspace environment The sparse planning coefficient of the parameter. With the support of the above physical quantities, the simultaneous formula (1) and formula (2) can express the result of subspace clustering as: 

˜ = [Js ]2 · ||c1 − c2 ||2 /D D

α d −1 −1

2

(3)

The cluster analysis algorithm regards a certain point in the subspace as the center. Under the condition that the unit storage amount of abnormal traffic information in the education network does not change, the parameter value less than the data information storage density at the point can be defaulted as one kind of cluster element, while the parameter value greater than the data information storage density at the point can be defaulted as another kind of cluster element, The two elements are never equal, but they can be transformed into each other through the established clustering function. 2.3 Cluster Similarity Calculation The physical definition based on cluster similarity is as follows: given a data set with large enough sample space, it is specified that each abnormal traffic information of education network can maintain an independent corresponding relationship with a given similarity marker coefficient. On this basis, a clustering tree of data objects is constructed, and then hierarchical decomposition is carried out until certain conditions are met [6]. According to whether the hierarchical decomposition is formed from bottom-up or top-down, the similarity index can be further divided into condensed and split hierarchical clustering. Hierarchical clustering is a bottom-up strategy. Firstly, each object is regarded as a

728

L. Ma et al.

cluster, and then these atomic clusters are merged into larger and larger clusters until all objects are in a cluster or meet a termination condition; Split hierarchical clustering is a top-down strategy. Contrary to condensed hierarchical clustering, first, all objects are placed in a cluster, and then gradually subdivided into smaller and smaller clusters until each object forms a cluster or meets a termination condition. Its specific definition form is as follows. (1) Cohesion similarity Firstly, by analyzing the abnormal traffic information of education network, a complete clustering condition is established, which is regarded as the basic clustering principle of information data; Then, the cohesion index of abnormal traffic information in educational network at hierarchical nodes is analyzed; Finally, the aggregation similarity coefficient is calculated. Let x1 , x2 , · · · , xn denote the scale values of n different abnormal flow information nodes of the education network, f1 denotes the cohesion coefficient of the information index, β1 denotes the cohesion index of the information parameter, and λ1 denotes the cohesion processing authority of the information parameter. With the support of the above physical quantities, the simultaneous formula (3) can define the clustering analysis expression of agglomerated similarity as:      1 − D ˜ · f1 (x1 + x2 + · · · + xn )   (4) K1 = 2 2 β1 + λ1 (2) Split similarity Firstly, the ability to split and change the abnormal traffic information of education network in the clustering environment is determined; Secondly, the matching relationship between information variables and known clustering behavior is studied; Finally, the influence of split similarity on the detection results of abnormal traffic information in education network is analyzed. Suppose f2 represents the splitting coefficient of the information index, β2 represents the splitting index of the information parameter, and λ2 represents the splitting processing authority of the information parameter. When the scale value of the abnormal traffic information node of the education network is always x1 , x2 , · · · , xn , use The above physical quantity can be defined as the expression of cluster analysis of split similarity as: β2 · λ2  K2 =   ˜  f2 D/(x1 + x2 + · · · + xn )2 

(5)

In short, the basic application idea of cluster analysis technology is to divide data objects into relatively small clustering indexes through an information hierarchical way; Then, a condensed hierarchical clustering algorithm is used to find the real result cluster by repeatedly merging subclasses. In this process, not only the interconnection, but also

Abnormal Traffic Detection Method of Educational Network

729

the approximation between clusters, especially the internal characteristics of clusters, are considered to determine the most similar sub clusters, so that it does not depend on the model provided by a static node, It can automatically adapt to the internal characteristics of the merged clusters; Finally, the global expression ability of abnormal traffic information in education network is summarized, so that the Internet host can accurately detect these information parameters [7].

3 Abnormal Traffic Detection in Education Network With the support of cluster analysis technology, according to the processing flow of abnormal traffic information extraction, median absolute deviation calculation and minimum covariance analysis, the design and application of a new education network abnormal traffic detection method are completed. 3.1 Abnormal Flow Information Extraction In the process of detecting abnormal traffic in the education network, due to the excessively large data base, there will be phenomena such as complex data processing, slow detection process, and inaccurate detection results. In order to avoid the occurrence of the above situation, the abnormal flow information should be accurately extracted. First, collect the data generated by the education network attack to obtain the abnormal traffic data set. The abnormal traffic data generated is then stored in the database. Design a data pipeline for processing flow data, which is used to classify network flow data and analyze the classified flow to search for flow data with common characteristics [8]. Next, extract the data parameters that meet the cluster analysis criteria, and design

Fig. 1. The principle of the extraction of abnormal traffic information in the education network

730

L. Ma et al.

the interface to establish a connection with the database. After receiving the data, perform abnormal flow feature selection on the detection pipeline, design a network flow data framework to process the flow, and then perform flow classification operations on the flow. The detection host performs classification operations based on the metrics set by the user, and the classification operation detects abnormal traffic in the network data. Subsequently, the classified traffic data is subjected to a stream analysis operation to analyze the attribute combination of abnormal traffic and normal traffic that is most likely to become abnormal traffic, and analyze and explain the characteristics of the traffic data. Finally, statistically analyze the data generated by the operation, provide an analysis list sorted by abnormal support, and generate a static report to display to the user. The specific extraction principle is shown in Fig. 1. Through the cluster analysis method, the abnormal traffic information of the education network can try to create a denial of service in the detection host or service object, and the traffic is much lower than other DDoS attacks. Volumetric attacks designed to saturate the network infrastructure around the target do not only require SYN attacks that are larger than the backlog available in the target operating system [9]. If the attacker information can determine the size of the backlog and the length of time each connection remains open before timeout, the attacker can locate the exact parameters required to disable the detection host, thereby reducing the total traffic to the minimum required to create a denial of service This is also the main reason why the abnormal traffic data of the education network can be accurately extracted. 3.2 Median Absolute Deviation For the individual abnormal traffic data of education network, the variation of clustering analysis algorithm is to use the median and median absolute deviation instead of the mean and standard deviation as the measurement of the position and scattering of detection instruction distribution. The median absolute deviation method counts the median of the absolute distance from each point in the data information sample to the median of the sample [10]. Since the median itself is resistant to outliers, each peripheral data point has a limited impact on the median absolute deviation scores of all other points in the sample. From a statistical point of view, the median absolute deviation of abnormal traffic data in education networks is a powerful measure of the distribution of univariate quantitative data samples. If the test data is normal, the standard deviation is usually the best choice for statistical deviation. However, if the test data is abnormal, the median absolute deviation statistics can be used. For univariate data set M , m1 , m2 , · · · , and mn represent n unequal abnormal traffic information of education networks. The larger the subscript value, the greater the value of abnormal traffic information at the current location., The median absolute deviation is defined as when the absolute deviation of the median of the data is equal, that is, the value can represent the median data of the abnormal traffic information of the education network. The specific median absolute deviation definition expression is as follows: UM = |K1 − K2 |

(m1 + m2 + · · · + mn )2 − ξ m √ ˆ n × g˙ φ2 k

(6)

Abnormal Traffic Detection Method of Educational Network

731

Among them, m represents the average value of abnormal traffic information of n education networks, ξ represents the median check index, φ represents the predetermined deviation coefficient, kˆ represents the most reasonable deviation measurement result of abnormal education network information, and g˙ represents the median feature definition Permissions. Starting from the median deviation of abnormal traffic data in education network, the median absolute deviation is the median of its absolute value. Under the function of cluster analysis algorithm, the formula is the change of average absolute deviation formula. It is less affected by outliers because outliers have less impact on the median than on the mean. In practical application, the absolute deviation of digits refers to the statistics calculated from samples. However, it can be used to estimate population parameters. Median absolute deviation is a measure of statistical deviation. In addition, the median absolute deviation can more stably count the abnormal traffic data of education network, and the statistics of abnormal values in the data set is more flexible than the standard deviation. In the standard deviation, the distance from the mean value is square, so it has a large deviation, and the abnormal value will seriously affect it. In the median absolute deviation, the deviation of a few outliers is irrelevant. 3.3 The Minimum Covariance of the Test Results After the detection host identifies the abnormal traffic information of the education network, it needs to analyze and mine the characteristics of the data, summarize the characteristics of the abnormal data, and the frequent itemset mining method can extract the data characteristics. The combination of items that always appear together can be extracted by cluster analysis algorithm. The combination of these items is called frequent itemset. After the frequent itemset is extracted, if there are entries in a transaction, other entries can be recommended to the detection host, that is, to obtain accurate minimum covariance calculation results. The detection of abnormal traffic in educational network refers to mining the network data information that does not conform to the normal behavior in the network flow data through the processing methods of machine learning, statistical analysis and cluster analysis. By combining the main idea of network anomaly traffic detection with the idea of big data detection and monitoring, a network anomaly detection platform based on big data can be constructed. Specifically, the data collection tasks such as log information and network traffic data information in various abnormal behaviors of educational network are completed in the data collection module; Storing heterogeneous data, and processing the burst of data through cache is realized by the storage management module; Feature extraction, statistical analysis, model training, representation extraction and full-text retrieval are completed in the intrusion behavior discovery module; Real time display, data interaction visualization, report management and system operation and maintenance management are realized in the configuration management and display module [11]. The network anomaly detection platform based on big data is shown in Fig. 2.

732

L. Ma et al.

Fig. 2. Network anomaly detection platform based on big data

Let ξ represent the covariance statistical scale value of abnormal traffic information of the education network, v0 represent the initial assignment result of the detection data indication, χ˜ represent the characteristic value index of the abnormal traffic information of the education network that can be distinguished by the detection host, l1 and h1 represent two Different data information detection scalar. With the support of the above physical quantities, the simultaneous formula (6) can define the minimum covariance expression of the detection result of abnormal traffic information of the education network as: 2          |U | ξ · M   δmin =   (7)    +∞    v χ˜ · ||l1 + h1 ||d χ˜  0

So far, the calculation and processing of various index parameters are completed, and the smooth application of abnormal traffic detection method in education network is realized with the support of cluster analysis technology.

4 Case Analysis In order to highlight the practical application value of the abnormal traffic detection method of education network based on cluster analysis technology, the following comparative experiments are designed. Two Internet hosts with identical configurations were selected as the experimental objects. The hosts in the experimental group were equipped with the detection method of abnormal traffic in the education network based on cluster

Abnormal Traffic Detection Method of Educational Network

733

analysis technology, and the hosts in the control group were equipped with the traditional detection method. The hosts of the experimental group and the control group were placed in the same Internet environment; The same amount of abnormal traffic data of the education network will be obtained and input into the host components of the experimental group and the control group respectively; Record the real-time detection speed of the host in the experimental group and the control group for the abnormal traffic information of the education network, and compare the experimental results with the ideal values. The detection speed of the host component for the abnormal traffic information of the education network can reflect the degree of threat of the abnormal information parameter to the operation security of the education network. Generally speaking, the faster the detection speed, the ability to reflect the threat of the abnormal information parameter to the operation security of the education network The weaker, the stronger otherwise. The following table records the ideal numerical situation of the detection speed of abnormal traffic information of the education network. Table 1. Ideal value of detection speed Time/(min)

Detection speed/(×10−7 T)

5

9.8

10

10.1

15

10.5

20

10.7

25

10.9

30

11.4

35

11.6

40

11.5

45

11.3

50

11.0

55

10.6

60

10.8

65

11.0

70

10.8

75

10.6

80

11.5

Analyzing Table 1 shows that with the extension of the experiment time, the ideal detection speed of the abnormal flow information of the education network shows a numerical trend that first increases and then decreases. When the time value is 35 min, the detection speed reaches the maximum value of 11.6 × 10−7 T; when the time value

734

L. Ma et al.

is 5 min, the detection speed reaches the minimum value of 9.8 × 10−7 T, the physical difference between the two The numerical difference is 1.8 × 10−7 T. The following figure reflects the experimental numerical changes of the experimental group and the control group’s educational network abnormal traffic information detection speed (Fig. 3).

Fig. 3. Experimental value of detection speed

Experimental group: For the Internet host of the experimental group, the detection speed of abnormal traffic information of the education network shows a numerical trend of increasing, stable, rising, and decreasing. The fluctuation state in the later stage of the experiment is more obvious than that in the early stage of the experiment. When the time value reaches 15 min, the abnormal flow information detection speed of the experimental group education network is always lower than the ideal value. When the time value is equal to 45 min, the abnormal flow information detection speed of the experimental group education network reaches its maximum value, which is 14.1 × 10−7 T. Compared with the ideal maximum value of 11.6 × 10−7 T, this is an increase of 2.5 × 10−7 T. Control group: For the Internet hosts in the control group, the detection speed of abnormal education network traffic information shows a fluctuating numerical trend. During the entire experiment, when the time value is equal to 30 min, the detection of abnormal education network traffic information The speed reached the maximum value of 7.9 × 10−7 T, which was 3.7 × 10−7 T lower than the ideal maximum value of 11.6 × 10−7 T, and the overall average level was much lower than the experimental group. In order to test the superiority of this method, 500 abnormal traffic information are set in the education network, and the abnormal traffic detection method based on clustering analysis and the traditional network abnormal traffic detection method are used to detect them respectively. Taking the detection time as the experimental index, the shorter the

Abnormal Traffic Detection Method of Educational Network

735

detection time is, the higher the detection efficiency of the method is proved. The specific results are shown in Fig. 4.

Fig. 4. Test time comparison results

It can be seen from Fig. 4 that the detection time of the two groups of methods increases with the increase of abnormal traffic data. The experimental group spent 3.0 s on detecting 500 abnormal flow data, while the control group spent 8.0 s on detecting 500 abnormal flow data. The time of the detection method proposed in this paper is shorter, which shows that the detection efficiency of this method is higher. In summary, with the detection method based on cluster analysis technology, the Internet host’s resolution speed of abnormal traffic information of the education network has been appropriately promoted, which can better solve the operational safety problem of the education network caused by abnormal information parameters. Actual application requirements.

5 Conclusion Compared with the traditional detection methods, the detection method based on cluster analysis technology carries out cluster analysis on the subspace by standardizing the abnormal traffic information, and combines the median absolute deviation index to realize the accurate calculation of the minimum covariance of the detection results. From the practical point of view, the abnormal traffic information of education network has been detected quickly, which can avoid the threat of abnormal information parameters to the operation security of education network.

References 1. Zhao, J., Yang, Y., Xin, Y., Zhu, H.: Unsupervised network anomaly flow detection algorithm based on CGAN-LSTM. Softw. Guide 21(03), 170–175 (2022)

736

L. Ma et al.

2. Zhan, P., Chen, L., Cao, L., Li, X.: Network abnormal traffic detection algorithm based on characteristic symbol representation. J. Zhejiang Univ. (Eng. Ed.) 54(07), 1281–1288 (2020) 3. Gao, M.: Network data flow anomaly detection algorithm based on mathematical model. Changjiang Inf. Commun. 34(11), 42–44 (2021) 4. Sagatov, E., Lovtsov, K., Sukhov, A.: Identifying anomalous geographical routing based on the network delay. Int. J. Netw. Secur. 21(5), 760–767 (2019) 5. Liu, Y., Wang, Y., Qiang, Y., et al.: Network traffic anomaly detection based on random projection and clustering. Comput. Simul. 36(3), 289–293 (2019) 6. Passas, V., Miliotis, V., Makris, N., et al.: Pricing based distributed traffic allocation for 5G heterogeneous networks. IEEE Trans. Veh. Technol. 69(10), 1 (2020) 7. Liu, T., Abouzeid, A.A., Julius, A.A.: Traffic flow control in vehicular multi-hop networks with data caching and infrastructure support. IEEE/ACM Trans. Netw. 28(1), 1–11 (2020) 8. Shridhar, V.S.: The India of Things: Tata Communications’ countrywide IoT network aims to improve traffic, manufacturing, and health care. IEEE Spectr. 56(2), 42–47 (2019) 9. Mesquita, L.A.J., Assis, K.D.R., Almeida, R.C.: Multi-period traffic on elastic optical networks planning: alleviating the capacity crunch. J. Supercomput. 77(6), 5468–5491 (2020). https://doi.org/10.1007/s11227-020-03493-7 10. Bianchin, G., Pasqualetti, F.: Gramian-based optimization for the analysis and control of traffic networks. IEEE Trans. Intell. Transp. Syst. 21(7), 3013–3024 (2020) 11. Al-Najjar, A., Khan, F.H., Portmann, M.: Network traffic control for multi-homed end-hosts via SDN. IET Commun. 14(19), 3312–3323 (2020)

Simulation Operation System of Civil Aviation Professional Electromechanical Equipment Based on Human-Computer Interaction Dan Zhao(B) and Xin Zhang College of Aeronautical Engineering, Beijing Polytechnic, Beijing 100176, China [email protected]

Abstract. Traditional equipment simulation operating system adopts the principle of combining real object with simulation, which has the problems of poor feedback accuracy and slow operation feedback speed. In view of the above problems, this research designs a civil aviation professional electromechanical equipment simulation and operation system based on human-computer interaction. Taking FPGA and arm as the core, the hardware part of the system is designed by connecting Kinect sensor. After the electromechanical equipment simulation model is established by using 3D Max software, the key frames in the actual operation interactive action video are extracted by using k-means clustering algorithm, and then the simulation operation gestures are recognized by combining with the expert judgment module, so as to realize the electromechanical equipment simulation operation training under human-computer interaction. Experimental results show that the feedback accuracy of the system is higher than 95%, and the feedback is sensitive, interactive and the simulation scene is clearer. Keywords: Civil aviation · Simulation practice of electromechanical equipment · Human computer interaction · Gesture recognition

1 Introduction Driven by the rapid development of automation and intelligent technology, all kinds of electromechanical equipment are emerging and playing an important role in many fields. The electromechanical system of civil aviation is very complex. The use of professional electromechanical equipment requires users to invest a lot of time and energy. Due to the complexity of the structure and operation of civil aviation professional electromechanical equipment, operators need to spend a lot of time learning during the training operation of civil aviation professional electromechanical equipment, and the operation process is not monitored, and it is difficult to control errors [1]. In addition, in the practical operation of civil aviation professional electromechanical equipment, due to the high cost of equipment and consumables, the shortage of equipment and tools, and the danger of operation to newly trained personnel, the practical operation efficiency is very low. Therefore, the operation practice of professional electromechanical equipment © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 737–750, 2022. https://doi.org/10.1007/978-3-031-21161-4_56

738

D. Zhao and X. Zhang

assisted by the actual operation simulation system can reduce the improper damage to the equipment or the impact on personnel safety during the actual operation of the equipment. At present, the practice system of electromechanical equipment in civil aviation specialty mainly uses electromechanical equipment model and training prototype for modular training, or learns and operates through computer simulation system under the guidance of teaching personnel after learning relevant theories. The system combining physical object and virtual platform in reference [2] reduced the cost investment to a certain extent. However, with the continuous development and maturity of virtual reality technology, integrating virtual reality technology into the virtual operation process of civil aviation professional electromechanical equipment obviously has more important practical significance for cultivating professional students’ practical operation ability and reducing training operation cost. Therefore, reference [3] designed a practical operation system based on virtual reality, which improves the students’ practical operation efficiency of diesel engine. However, when virtual reality technology is applied to the practical operation simulation system of civil aviation electromechanical equipment, the interaction with system users is poor, lack of realism, and the real-time performance of the system is poor. The emergence and development of human-computer interaction technology provides a new idea for the practical operation of aviation electromechanical equipment. This new interaction technology can provide a new interactive experience in the teaching and training of electromechanical equipment of civil aviation specialty, and can also provide technical assistance for operators by using a new display mode, which can effectively reduce human errors in the process of practical operation, reduce the probability of wrong operation during real operation and avoid damage to civil aviation professional electromechanical equipment. According to the above research and analysis contents, in order to improve the talent training speed of civil aviation and improve the familiarity and mastery of civil aviation professional electromechanical equipment, this paper will design a civil aviation professional electromechanical equipment simulation and operation system based on human-computer interaction. Through the application of the system, the training conditions can be improved in the training teaching, and the human-computer interaction can be realized assemble the accessories in the virtual environment to reduce the operational risk in the practical operation of aviation special electromechanical equipment and improve the skill level of learners.

2 System Hardware Design The hardware structure of the simulation practical operation system of civil aviation professional electromechanical equipment based on human-computer interaction is shown in Fig. 1.

Simulation Operation System

Human-computer interaction equipment

739

Flash

SDRAM Low pass filter

AD7656

FPGA JTAG

Flash

RAM

Ethernet RS232

ARM

Power supply module

SDRAM

USB

Reset circuit

Fig. 1. Schematic diagram of the overall system hardware architecture

In Fig. 1, the embedded real-time operating system is located at the core. As the logic processing unit of the whole system, the core unit has network interface, ARM hardware and FPGA, and has the function of real-time multitasking. According to the analysis of the human-machine function requirements of the simulation actual operation system, the human-machine application layer mainly realizes the functions of file input and output, parameter setting and running status monitoring. Users can intuitively observe the operating status of the equipment and find problems through the human-machine application layer And the operation to deal with emergencies [4]. The ARM hardware platform uses Samsung S3C6410 as the core processor to implement hardware support for embedded Linux systems and hardware support for data communication. Select through the OM[I:O] pin. When OM[1:0] = 00, the processor will boot from NAND Flash; when OM[1 0] = 01 or 10, the processor will boot from ROM [5]. Peripheral The human-computer interaction interface mainly includes USB interface, LCD interface, keyboard interface, touch screen interface and SD card interface, which mainly realizes the input and output of files and data, as well as data storage. The voltage conversion circuit mainly uses the ACT6311 power conversion chip and the +10 V, +16 V, −7 V and +3.7 V voltages of the LM358DR. The parameter acquisition module is responsible for converting the strong electric signal generated by the human-computer interaction device into a weak electric signal suitable for AD sampling. It consists of a voltage acquisition circuit, a current acquisition circuit, a low-pass filter circuit, a frequency measurement circuit, and a phase-locked frequency multiplication circuit. The filtered signal is sent to the analog signal input end of AD7656 through a voltage follower for AD conversion. The AD sampling module samples the

740

D. Zhao and X. Zhang

six signals output by the parameter acquisition module through the externally expanded AD7656 chip [6]. The phase-locked frequency multiplication circuit is used to obtain a square wave signal with 256 times the frequency of the input signal as the driving pulse for AD sampling, so that it can ensure that the AD converter can sample 256 points per cycle. The FPGA will lose its configuration every time it is powered down, so every time it is powered on, it needs to load the startup code from the external storage. The CY22394 clock generator from CYPRESS is used. The chip integrates three programmable phase-locked loops and independently outputs 4 channels of 2000 MHz clock signals. The data communication of the system is mainly based on dual-port RAM, supplemented by serial communication. Among them, dual-port RAM is mainly used for the communication between the upper computer and the lower computer, and RS232 is mainly used for the communication between the upper computer and the PC. The main function is to print information of the system and serve as the debugging serial port of the system and the man-machine. RS485 mainly realizes the input of keys and can also be used as a data communication interface between the upper computer and the lower computer. In order to improve the scalability and versatility of the system and realize the distributed control, this article uses Ethernet as another method of data communication. The communication in this article mainly adopts the format of the virtual sCAN protocol, encapsulating the source address, destination address, instruction code, function code and function parameters into a data frame to realize the communication between the upper computer and the lower computer. The definition of the communication data frame format is shown in Table 1. Table 1. System communication data frame format definition Data frame bit number

Address

Transfer content

1

Ad0

Script

2

Ad1

Function code

3

Ad2

Data1

4

Ad3

Data2

5

Ad4

Data3

6

Ad5

Data4

7

Ad6

Data5

8

Ad7

Data6

9

Ad8

Data7

10

Ad9

Data8

11

Ad10

Request code/response code

In order to improve the realism and interactivity of using the system for the practical operation of civil aviation professional electromechanical equipment, the system uses the Kinect somatosensory device to capture the operation of the system user. The internal

Simulation Operation System

741

processor chip of the Kinect sensor selected in this design is PS1080, which can process the reflected speckle image collected by the infrared camera and generate depth and distance information. The main sensor parameters of Kinect sensing equipment are shown in Table 2. Table 2. Kinect sensor equipment parameters Numbering

Sensor parameters

Parameter data

1

Measurable distance

0.8–4 m

2

Visual field

Vertical: 45°; Horizontal: 55°

3

Frame rate

50 fps/s

4

Depth camera

720 * 480

5

Color camera

1280 * 1080

Through the setting of the above parameters, the Kinect sensor can collect 3D data of each joint point of human motion in real time for modeling. Through the difference between the angle and depth of each joint and the fingertip of the limb, the posture of the human body can be judged, and the position of each node can be determined at the same time. Time difference to track the direction of human body movement. In the process of human-computer interaction, DS90CR288A is selected as the chip to decode the operator’s behavior image. The chip supports up to 85 MHz clock and the maximum transmission bandwidth of the chip is up to 100 Mbps [7]. The image data is converted in real time through the PCI-E interface for data transmission. In order to ensure the PCI-E transmission efficiency and increase the transmission bandwidth, the extended DDR3 SDRAM is used to cache the data. Expand all data and control signals of DDR3 through BANK34. With the support of the above-designed system hardware framework, design the software part of the simulation and actual operation system of civil aviation professional electromechanical equipment to realize the use of system design functions.

3 System Software Design 3.1 Simulation Modeling of Electromechanical Equipment When users use the civil aviation professional electromechanical equipment simulation practical operation system based on human-computer interaction for practical exercises, the system needs to have the correct civil aviation electromechanical equipment model and corresponding parameters. Therefore, this article uses 3D MAX software to establish a professional electromechanical equipment model for civil aviation. According to the geometric dimensions of the electromechanical equipment, use the standard primitives in the 3D MAX software, such as spheres, cylinders, cubes, etc., to select the corresponding standard primitives. The size is basically the same as the actual object, and the unit is millimeters. After the basic standard body is created, adjust the

742

D. Zhao and X. Zhang

length, width and height according to the actual size [8]. Super Boolean objects combine two or more other objects by performing Boolean operations. Super Boolean adds a large number of functions to the traditional 3D MAX Boolean object, using different Boolean operations, the ability to combine multiple objects at once to get mesh smoothing and turbo smoothing effects. In actual modeling, by transforming into editable objects, taking editable polygons as an example, 3D MAX provides different levels of modification tools from vertices to elements. When building a complex model, a simple cube can be used to continuously modify its vertices, edges or polygons to create a variety of complex graphics. According to the actual material and texture characteristics of electromechanical equipment, the established model is processed in detail. According to the design parameters and related basic principles of the professional electromechanical equipment for civil aviation, the operation of the electromechanical equipment model is expressed by mathematical simulation, so that the subsequent use of the system for actual operation can obtain correct feedback. 3.2 Key Frame Extraction The user carries out the interactive action of simulation and practical operation through the camera and civil aviation professional electromechanical equipment. The key frame in the camera imaging video reflects the key video information in the video. Therefore, before practical operation recognition, this study uses K-means clustering algorithm to extract key frames. K-means algorithm is the most widely used clustering analysis algorithm because of its simple principle, easy implementation, good clustering effect and fast convergence speed. In this paper, for 25 bone points in the action video collected by Kinect V2, combined with the cognition of human motion, weighted k-means algorithm is used to extract key frames. In human motion, the positions of different bone points are different. For example, the weight of spine points is greater than that of wrist points, and the swing range of wrist points is larger in action, which will interfere with the cluster center. For different weights of each bone point in motion, the weight of bone point is defined according to formula (1):  Bi = Bi

n 

−1 Bi

(1)

i=1

where, n 

Bi =

  dist ai , bj

i=1,j=1

n

(2)

  where, dist ai , bj is the Euclidean distance between ai and bj . The closer the distance is, the smaller the weight is. Otherwise, it is the opposite.

Simulation Operation System

743

In the user’s practical action video collected by Kinect V2, each frame image contains the three-dimensional coordinate values of 25 bone points of the human body to form a 75 dimensional vector. An action sequence consists of N frame images, expressed as {p1 , p2 , . . . , pn−1 , pn }. pi represents the i-th frame in the N -frame image, and P N represents the vector set of an action video composed of 75 dimensional vectors per frame. The vector set P N cluster is divided into K clusters, and the frame similarity in each cluster is high. The specific steps are as follows: Step 1: randomly select K frames from the P N frame set as the clustering centroid, representing c1 , c2 , . . . , ck , k ≤ N . According to the following formula, the Euclidean distance between pi and K cluster centroids in each frame is calculated and classified into the corresponding cluster of the nearest cluster centroid.   (3) Di = arg minpi − cj  Step 2: update the weighted average clustering centroid of each cluster according to the following formula: n 

cj =

Bi γij pi

i=1

n 

(4) γij

i=1

where, Bi represents the weight of the data object in cluster cj ; γij is the standard for judging whether pi belongs to class j. If yes, the value of pi is 1, otherwise it is 0. Step 3: repeat steps 2 and 3 until the old and new centroids are equal or the difference is less than the threshold. Through the above steps, a segment of electromechanical equipment operation video is divided into K clusters, and the cluster quality ci of each cluster is obtained. Each frame in the action video is composed of 75 feature vectors with the three-dimensional coordinates of 25 bone points. Calculate the Euclidean distance between the cluster centroid and the three-dimensional coordinate value of the frame 7 {same bone point in the action sequence, and find the key frame. For example, calculate the Euclidean distance between the bone point L and the bone point l of the cluster centroid, and so on. The bone point with the smallest distance value is marked as l, and the others are marked as 0. If the frame pi in the video passes, the calculation result is {1,l, 0, 0, 0, 0, 0, 1, l, 1, 11, l, 1, l, l, 0, 0, 01, l, l, 0, 0, 1}. The specific steps are as follows: Step 1: use k-means algorithm to obtain K clustering centroids of an action sequence, expressed as: 

 Bi = (xi1 , yi1 , zi1 ), (xi2 , yi2 , zi2 ) . . . , xij , yij , zij i ∈ (1, 2, · · · , k), j = 25 (5)   where, xij , yij , zij is the three-dimensional coordinate of the j-th bone point in the i-th cluster centroid. A continuous electromechanical equipment operation video is shown as: 

 Pn = (xn1 , yn1 , zn1 ), (xn2 , yn2 , zn2 ) . . . , xnj , ynj , znj

744

D. Zhao and X. Zhang

n ∈ (1, 2, · · · , N ), j = 25 (6)   where, xnj , ynj , znj represents the three-dimensional coordinates of the j-th bone node in frame n of the action video. Step 2: calculate the Euclidean distance between the cluster centroid Bi and the three-dimensional coordinates of the 25 bone points corresponding to the operation in frame n of the action video.   the distance  for each frame in the action video, mark the minimum After calculating dis xij , yij , zij , xnj , ynj , znj as l, otherwise mark it as 0. The minimum Euclidean distance of key points corresponding to each frame and K clustering centroids is obtained through the Euclidean distance formula, and the first K frames are selected as key frames according to the number of l in B, and then sorted according to the index of each key frame in the original action video, so as to ensure that the order of K key frames is consistent with that in the action video. The k-means algorithm is used to obtain the effective and key image information in the actual operation video of aviation professional electromechanical equipment, in which the selection of K value is also the key part. The K value is small, the information obtained is too small, and the recognition accuracy is low; When the K value is large, there is more redundant information, and the recognition accuracy will also be affected. In this paper, when the K value is set to 10, better accuracy and recognition speed can be obtained. 3.3 Practical Gesture Recognition of Electromechanical Equipment The gesture area is extracted from the depth image obtained by the Kinect sensor, and the number of fingers is recognized and compared with the preset number of gesture fingers to realize the recognition of gestures and complete the algorithm framework of static gesture recognition. First, establish the coordinate transformation of the operator’s gesture in the Kinect sensor, the human-computer interaction device and the space. The coordinate of the hand coordinate system is {H }lr = {XH , YH , ZH }, the coordinate of the hand in the space coordinate system is {P}lr = {XP , YP , ZP }, and the coordinate of the hand in the humancomputer interaction device coordinate system is {W }lr = {XW , YW , ZW }. According to the posture of the gesture in different coordinate systems, combined with the law of rotation, the position Y HP and posture of the hand in the applied coordinate system are obtained:

 Y HP = {H }lr {P}lr + α {W }lr + {W }l (7) {Z}lr = αβ{W }lr

(8)

In the formula, α and β are respectively the rotation and translation amount of the applied coordinate system relative to the hand coordinate system. After determining the coordinates of the gesture in the image, segment the gesture area. When the image depth value is in the range of 600 ~ 800, the recognition rate is significantly improved, and the value is relatively stable; after the depth value reaches 800, the recognition rate begins

Simulation Operation System

745

to decrease; when the depth value is between 700 and 900, the recognition rate can reach more than 90%; depth value When it is greater than 900, the recognition rate drops significantly [9]. Considering that in the process of gesture recognition, the distance between the hand and the device will change to a certain extent, so this article uses 700 to 800 as the depth range threshold. The K-Means algorithm is used to complete the next segmentation of the gesture area. The purpose of gesture area filtering and morphological operations is to remove noise in the image and fill in some small holes, so as to better retain the edge information of the object. In this paper, the traditional median filter is improved, and the calculation is simplified by removing only the zero-value noise points. First define the filtering window, this article uses a 5 × 5 neighborhood window. According to the center value of the window, it is judged whether it is a value of 0, if it is not a value of 0, no processing is done; if it is a value of 0, all pixels with a value of 0 in the window are removed, and the remaining pixel values are searched for the median value as The center point of the window effectively filters out the noise in the image. If the gray value of a pixel (x, y) in the image is f (x, y). The window size of the median filter is 5 × 5. When the window moves to the vicinity of pixel (x, y), the gray value of the center pixel of the window can be used to replace the gray value of all pixels in the window [10]. Then the pixel gray value of the median filter window can be sorted by the following formula: {Fi−v , · · · , Fi−1 , · · · , Fi , Fi+1 , · · · , Fi+v }

(9)

In the formula, Fi is the output of the median filter, v = L − 1/2 and L are the width of the median filter window. After the image to be recognized is processed by the median filter, the noise generated by the external interference during the image acquisition and conversion process can be removed. In this paper, the threshold is set according to the difference of V (brightness) in HSV. If the difference of V is less than 100, the pixel is set to black; when V is greater than 100, the pixel is set to white. Next, scan each pixel point in sequence from different directions; if the value is different from the adjacent two pixels on the left and right, the point is regarded as the boundary point, and the process is looped to find the contour of the gesture. The depth image obtained by Kinect calculates the three-dimensional coordinate information of each bone point of the human body, thereby calculating the moving distance of each bone point in the two images before and after. If the distance is less than the preset threshold, it is judged as a static state, which means that it is the same action. The distance between the bone points in the two images before and after the detection is phased, including the start point and the end point. Therefore, it is very important to determine the start and end of the gesture. In the process of dynamic gesture recognition, the researcher first randomly selects the Euclidean distance of the hand points of the two images before and after. If the continuous M (M is greater than or equal to 5) images are in a static state, the next M + 1 image starts When moving, the M + 1 image is judged as the initial action of the dynamic gesture; if the continuous motion is N (N is greater than or equal to 10), the N + 5th image is in a static state until the N + 1 image value, The M + Nth image is judged to end the action. The GetHandState function included in the Kinect SDK can be used to determine when the operator’s hands are in a fisted state. In order to better determine the start of

746

D. Zhao and X. Zhang

the gesture and reduce the constraints on people, this method adds a fisted state, that is, when the operator’s hands are in a fisted state. When the hand is in a fisted state, the discrimination method becomes effective. Proceed as follows: 1) Call the GetHandState() function to judge whether the operator’s hand is in a fisted state, if it is in a fisted state, continue to judge; otherwise, re-check whether it is in a fisted state; 2) Determine whether the M continuous images containing the hand are greater than or equal to 5 and are in a static state, and the hands are all in a fisted state, if yes, proceed to the next step of judgment; otherwise, re-detect the images; 3) Determine whether the N images containing continuous hand movement are greater than or equal to 10 and the hand is in a fisted state. If so, determine the Mth image as the start and M + N images as the end; otherwise, return to step 1. The use of hidden Markov model needs to meet the following three premise assumptions: 1) The state has nothing to do with time; 2) The observed value at any time is only related to the current hidden state; 3) The current state is only related to the previous state. Assuming that the parameters in the HMM and the observation sequence of the gestures are known, a matching HMM model needs to be found, and the corresponding gesture is the recognized gesture. The Baum-Welch algorithm is used to train a large number of gesture samples, and the result is the gesture corresponding to the unknown gesture. Naive Bayes regards the dynamic behavior characteristics of different hands as independent of each other. According to Bayesian theory, there are: P(ai )P(a|bi ) = P(ai )

n    P bj |ai

(10)

j=1

In the formula, bj is the characteristic attribute of hand dynamic behavior; ai is the characteristic of hand dynamic behavior to be classified. Follow the flowchart shown in Fig. 2 to train and recognize gesture recognition. The set gesture discrimination expert module is used to judge the operation of the recognized gesture, so as to guide the user to carry out the simulation operation of electromechanical equipment. Transplant the operation simulation software part designed above to the system hardware framework, that is, complete the design and research of civil aviation professional electromechanical equipment simulation operation system based on human-computer interaction.

Simulation Operation System Training process

747

Get sample gestures Calculate the probability for each sample Calculate the conditional probability for each sample

Enter the gesture features to be recognized

Naive Bayes model

Calculate the conditional probability for each characteristic attribute

Classification and recognition process

Classification to the category with the maximum conditional probability

Fig. 2. Naive Bayes training and recognition process

4 Test Experiment In order to ensure the stable operation of the system, it is necessary to test the civil aviation professional electromechanical equipment simulation operation system based on human-computer interaction. 4.1 Experiment Content A stable hardware platform is the basis for the realization of system functions. First, the hardware platform needs to be tested, which includes the basic power on debugging, power timing quality testing and key signal testing of the system. Then, the software and hardware functions are jointly debugged; Finally, the power consumption test is carried out. System debugging requires defect detection system circuit board, power supply, serial port and network connected with debugging computer, oscilloscope, multimeter and other test tools. After confirming that all aspects of the system operate normally, conduct performance test on the system. In the system performance test, the system designed in this paper is compared with the traditional mechanical and electrical equipment operation system. The actual power consumption, operational feedback speed, feedback accuracy and simulation scene clarity were taken as indexes to compare the application performance of the two systems. The actual power consumption of the system is reflected by the clock test result.

748

D. Zhao and X. Zhang

4.2 Experimental Results Table 3 shows the result data of the system’s clock test when testing the operation of the designed system hardware platform. Table 3. System hardware clock test results Signal

Theoretical clock frequency/MHz

Actual system power clock measurement value/MHz

Signal duty cycle

Signal-1

300

298.75

0.5212

Audio-25

25

24.62

0.4961

RTC_CLK

35.754

35.051

0.4798

Signal-12

12

12.00

0.5103

ETH_S

12.88

11.98

0.5044

Signal-Y

45

44.63

0.4876

1.0

1.6

0.9

1.4

0.8

1.2

0.7

1.0

0.6

0.8

0.5

0.6

0.4

5

10

50

60

Feedback time/s

Feedback accuracy

According to the data analysis in Table 3, the actual measured clock frequency of system operation is close to the theoretical clock frequency. Moreover, the duty cycle of the clock signal also meets the operation requirements of the actual system. After the system hardware test is normal, the comprehensive performance of the system is tested. Invite the same number of students to use the two systems to test the operation feedback speed, feedback accuracy and actual power consumption of the system under different use scenarios. Figure 3 shows the comparison results of operation feedback speed and feedback accuracy of the system.

0.4

20 30 40 The number of real operations System in this paper

Feedback accuracy

Traditional system

Feedback time

Fig. 3. Comparison of system feedback speed and accuracy

By analyzing the data in Fig. 3, it can be seen that the operation feedback speed of the system in this paper is always higher than that of the traditional system. With the

Simulation Operation System

749

increase of the use time and operation complexity of the system, the operation feedback accuracy of the traditional system continues to decline. The operation feedback accuracy of this system is higher than 95%. Finally, the application performance of the two systems is verified with the simulation scene clarity as an indicator, and the results are shown in Fig. 4.

System in this paper

Traditional system Fig. 4. Contrast the sharpness of simulation scenes

The results shown in Fig. 4 show that compared with the simulation scene images of traditional systems, the simulation scene images of this system have higher color saturation and clarity and better simulation effect. By summing up the above experimental analysis, it can be judged that the simulated real operation system of civil aviation mechanical and electrical equipment based on human-computer interaction designed in this paper has good practical application performance.

5 Conclusion The practical operation of professional electromechanical equipment for civil aviation has the problems of long preparation time, difficulty in carrying out, and many dangerous steps and procedures. Based on the virtual reality platform to carry out safe operations, set up the error operations that are easy to occur during the actual operation in the virtual environment, and combine the human-computer interaction technology to help students carry out the simulation actual operation, improve the sense of reality, and reduce the

750

D. Zhao and X. Zhang

investment and the risk of actual operation. For this reason, this paper designs a simulation and practical operation system for civil aviation professional electromechanical equipment based on human-computer interaction. This study takes FPGA and ARM as the core, and designs the hardware part of the system by connecting Kinect sensor. After establishing the simulation model of mechanical and electrical equipment with 3D MAX software, the key frames in the interactive action video of real operation are extracted, and then the gestures of simulated operation are recognized, so as to realize the real operation training of mechanical and electrical equipment under human-machine interaction. Through the system functional test, it is verified that the system has good interactivity. It can select the practical links and difficulties according to the actual situation, and ensure the training effect of the students through the implementation of multiple simulation training. In the following research, we will consider the introduction of graph neural network to further improve the recognition effect of simulated operation gestures. Fund Project. Scientific research project of Beijing Polytechnic, project name: Simulation Operation System of Civil Aviation Professional Electromechanical Equipment Based on Human-Computer Interaction.

References 1. Zhou, Y., Chen, L., Wang, X., et al.: Outlook on digital intelligence aviation electromechanical technology. J. Sichuan Ordnance 42(05), 14–19 (2021) 2. Liu, N.: Design of extensible data acquisition system for semi-physical fully mechanized mining operating platform. Ind. Mine Autom. 46(01), 95–99 (2020) 3. Li, H., Lu, P., Li, M.: Design and implementation of virtual reality training system of a diesel engine. Comput. Simul. 36(12), 267–271 (2019) 4. Han, Z., Yue, W.: Modelling & simulation of system reliability and maintenance cost for electromechanical equipment. Mach. Tool Hydraul. 49(02), 110–112+116 (2021) 5. Hao, Y., Dai, X., Cui, X.: Design of integrated modular avionics architecture avionic system flight management module. Sci. Technol. Eng. 21(16), 6923–6929 (2021) 6. Zhao, Y., Wu, T., Gu, S., et al.: Multi-index quantitative evaluation model of gunner’s operation posture based on human-machine simulation technology. Comput. Integr. Manuf. Syst. 27(08), 2350–2361 (2021) 7. Wang, J., Guan, S., Lei, M., et al.: Modeling and simulation of master manipulator based on the remote center motion mechanism. Mech. Electr. Eng. Mag. 36(02), 179–184 (2019) 8. Cheng, R., Huang, P., Liu, Z., et al.: A human-robot interaction method of on-orbit serviceoriented space teleoperation. J. Astronaut. 42(9), 1187–1196 (2021) 9. Song, Y., Fei, Y., Sun, G., et al.: Immersive three-dimensional simulation and visualization of microelectromechanical systems. J. Syst. Simul. 26(09), 1956–1960+1968 (2014) 10. Zhang, Y., Chang, Q., Wang, D., et al.: Research and development of 3D simulation training system for basic skills of electric energy metering device. Autom. Instrum. (04), 204–207 (2019)

Design of Online Teaching Method for Subject Knowledge of Mathematics Teachers in Higher Vocational Colleges Based on Convolutional Neural Network Chunyan Yu(B) and Junyan Wang College of Humanities and Information, Changchun Technology University, ChangChun 130000, China [email protected]

Abstract. Online teaching is developed on the basis of distance education. From the perspective of teacher-student relationship, online teaching should be the unity of five elements: teachers, students, technology, courses and activities. The level of interaction in the process of online teaching affects learners’ knowledge construction and learning quality. The existing network education platform focuses on the adaptive learning of knowledge content. Unable to give appropriate feedback based on current learning status. Convolution neural network can reduce the feature dimension, compress the amount of data, reduce the number of network parameters, and prevent over fitting. Therefore, this paper puts forward the design of online teaching method of subject knowledge for mathematics teachers in Higher Vocational Colleges Based on convolutional neural network. Based on the online teaching characteristics, the convolution neural network algorithm is improved, the convolution neural network algorithm structure is optimized, and the loss function is built. Based on convolution neural network algorithm, integrate online teaching materials and optimize the examination system of mathematics online teaching courses in higher vocational colleges. Finally, an example shows that the online teaching method of mathematics teachers’ subject knowledge in higher vocational colleges can effectively improve students’ performance. Keywords: Convolution neural network · Neural network · Higher vocational colleges · Mathematics teacher · Subject knowledge · Online teaching

1 Introduction Nowadays, education has entered the stage of industrialization development. Online learning has become the main way, which is spread in many fields such as educational technology, distance education and library science. In the education industry, the continuous development of Internet technology provides learners with a learning platform that is not limited by region and time. Online education uses the convenient environment of the Internet [1], allowing learners to acquire knowledge anytime and anywhere in new © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 751–764, 2022. https://doi.org/10.1007/978-3-031-21161-4_57

752

C. Yu and J. Wang

ways, breaking the restrictions of fixed teaching places and fixed teaching time in the past traditional teaching process, So that learners can freely arrange learning time and flexibly choose learning places, which promotes the development of lifelong learning. With the rapid development of information technology in the new century, a new generation of information technology came into being. Many new technologies have also become familiar keywords for the public: big data, cloud computing, Internet of things, artificial intelligence, etc. these new technologies continue to innovate, break through and improve, change all aspects of public life and provide new ideas for the improvement of learning environment. For teachers, mastering learners’ classroom participation and understanding learners’ learning experience and learning effect of this section are indispensable links in the teaching process. From the previous online teaching activities, we find that online education is a one-way learning process, and there is almost no direct communication between teachers and learners. This one-way learning process leads to the separation of teaching activities and learning activities [2], and teachers and learners lose communication and communication. The existing online education platforms focus on the adaptive learning of knowledge content. The communication between online education platforms and learners is also reflected in after-school learning evaluation, after-school Q & A, knowledge recommendation and so on. Learners’ emotions have not been paid enough attention, educators can not know the emotional changes of learners, and can not give appropriate feedback according to the current learning state. Literature [3] takes the first grade primary school teachers as samples to investigate the relationship between mathematics teaching knowledge and teaching quality. Ten teachers completed the mathematics teaching knowledge (MKT) survey at the end of teacher preparation. In their first year of teaching, three math lessons were recorded and graded using math teaching quality. The results repeat previous studies with more experienced teachers. Literature [4] discusses the relationship between teachers’ beliefs and teaching experience. Using the validated MTBs, this study assessed the beliefs of four groups of Chinese high school mathematics teachers with different professional teaching experience: 1–5 years (n = 25), 6–10 years (n = 70), 11–20 years (n = 48) and more than 21 years (n = 28). MTBs consists of 26 items, distributed in five sub scales, covering beliefs about mathematics, learning, teaching, students and teachers. Literature [5] aims to understand mathematics teaching by asking questions through the analysis of 22 teaching cases. Teaching mathematics by asking questions begins with the task of asking questions. This study not only provides specific examples of problem posing tasks used in the classroom, but also provides relevant task variables that need to be considered when developing problem posing tasks. This study also helps us to understand how teachers deal with students’ questions in class. Convolutional neural network originated in the 1980s and is one of the first deep learning methods. This method does not need to design classification methods and specific features for a specific class of image sets, so as to achieve the purpose of classifying pictures of different categories in a group of pictures. Moreover, convolutional neural network has been proved to have excellent performance in classification task and regression task [3], which is significantly improved compared with traditional image classification. At the same time, after 2006, with the rise of deep learning, the representation learning

Design of Online Teaching Method for Subject Knowledge

753

ability of convolutional neural network has been continuously improved with the renewal of numerical computing equipment, so it has been widely valued. Therefore, one of the most important research contents of computer vision in recent years is convolutional neural network. The main content of this paper is to apply convolution neural network technology to the online teaching method design of subject knowledge of mathematics teachers in higher vocational colleges.

2 Improved Convolutional Neural Network Algorithm Based on Online Teaching Features 2.1 Optimizing the Algorithm Structure of Convolutional Neural Network As one of the classical algorithms of deep learning, convolutional neural networks can be divided into three categories according to dimension. CNN with different dimensions has different application fields. For example, one-dimensional CNN is often used for sequence data processing, two-dimensional CNN is often used for image text recognition, and three-dimensional CNN is generally used for medical image and video data recognition. Nevertheless, the basic structure of convolutional neural networks is the same [4], including input layer and convolutional layer Pooling layer, fully connected layer and output layer. This section introduces the important role of convolution layer, pooling layer and full connection layer in CNN. The convolution layer of the algorithm is one of the most important modules in convolution neural network. The convolution layer consists of multiple feature maps and a set of convolution kernels with learning ability. The feature map is composed of multiple neurons. Each neuron is connected to a small area in the previous layer. This small area with the same size as the convolution nucleus is called the receptive field. In convolution operation, the existence of redundant parameters will affect the efficiency of convolution operation, so each convolution layer should try to control the amount of parameters in the operation process. In convolutional neural network, there are two ways to reduce the parameters of the network: weight sharing and local perception field. The local perception field can reduce the parameters of the local network to a certain extent [5], but it makes little contribution to reducing the overall parameters of CNN. Weight sharing is an important concept in convolutional neural network, and it is also an effective method to reduce the amount of overall parameters. The pool layer is the next layer of the convolution layer, which is composed of multiple feature maps and corresponds to the number of feature maps of the previous convolution layer. The input of the pooling layer is the feature map extracted from the previous convolution layer. Its substantive function is to extract the features twice. In addition, it can also reduce the dimension of the features, compress the data, reduce the number of network parameters and prevent the occurrence of over fitting.

754

C. Yu and J. Wang

Common pooling methods include maximum pooling, average pooling and random pooling. The selection of pooling method is based on the characteristics of feature map [6]. For the separation of very sparse features, the method of maximum pooling should be selected; When the linear classifier is used for classification, the maximum pooling method has better feature extraction performance than the average pooling method. 2.2 Build Loss Function Currently, target detection algorithms can be divided into two categories according to deep learning methods: one is the two-step detection algorithm represented by FASTRCNN and R-FCN; Second, SSD and YOLO are typical single-step detection algorithms. RoI pooling layer is a crucial convolution layer in Fast R-CNN network, and its substantive role is to transform RoI features into feature maps of fixed space size. In the operation of RoI pooling, the feature graph with input size of H × W is first divided into h × W size grid (h and W are hyperparameters and set as 7 × 7), and then the maximum pooling operation is carried out for each RoI to further obtain the output feature graph with fixed size [7]. Then, the whole connection layer is processed to obtain two branch vectors. RoI is a rectangular box (r; C, h, w), where (r; C) represents the coordinates of the upper left vertex, and (h, w) represents the height and width. The Fast R-CNN network contains two different output layers, so the loss function adopts the method of multi-loss fusion (classification loss and regression loss). The total loss function is expressed in Formula (1) as follows: L(r, c, h, w) = Lcls (r, c) + λ[h ≥ 1]Lcls (h, w)

(1)

In the training process, each RoI has an accurate category information and a real boundary regression box, and multi-task loss L is used to conduct joint optimization training for classification and boundary boxes, so that the loss error is smaller, and the trained model has better performance and detection effect.

3 Integration of Online Teaching Materials Based on Convolutional Neural Network Algorithm The resource integration method based on convolutional neural network mainly consists of two parts. The first part is the technical part, that is, the Arduino device recognition program based on convolutional neural network, which realizes the classification and recognition of Arduino devices [8]. The second part is the learning resources part, learning resources include text, pictures and other resources, through the collection and screening of existing learning resources and making videos, to complete the construction of Arduino device learning resources library. The overall process of resource integration method based on convolutional neural network is shown in Fig. 1:

Design of Online Teaching Method for Subject Knowledge

Learner

Use

Arduino device recognition program based on convolutional neural network

755

Arduino device

Learning resource library

Fig. 1. Overall process of resource integration method based on convolutional neural network

In the learning process, learners use the Arduino device recognition program based on convolutional neural network. The recognition program tells learners the name of the recognized device and pushes learning resources related to the recognized device to learners, including text resources and video resources. Many text resources and video resources form the Arduino device learning resource library, Learners are connected to the learning resource base through the identification program. The construction process of Arduino device recognition is divided into three parts: data set construction process, model training and generation process and picture classification and recognition process. Next, the construction process is described in detail. 3.1 Data Set Construction Process The training and generation of deep learning neural network model is based on learning a large number of picture features. Therefore, this study mainly classifies and recognizes images. This paper studies the deep learning model of Arduino device recognition, which is based on a large number of previously selected pictures of 10 Arduino devices [9]. The pictures of all Arduino devices are collected with the same image acquisition equipment to ensure that the resolution of the pictures is the same, so as to build the image data set of 10 devices. After that, the code program will divide the images into three categories: training set, test set and verification set. The number of images in each set will be automatically divided according to the percentage set in the code. The CNN model structure is shown in Fig. 2. 3.2 Model Training and Generation Process After completing the image acquisition and data set construction, the image format needs to be transformed. In this process, the image file in JPEG format needs to be transformed into a file format that can be recognized by convolutional neural network. In the training process of convolutional neural network model in this paper, the picture is transformed into tfrecord file as the initial data input of the network. After processing the image data, the convolution neural network realizes the output of the network model through multilayer convolution and pooling operation By comparing the error information between

756

C. Yu and J. Wang

Fig. 2. CNN model structure diagram

the actual output and the expected output, the test accuracy value and error loss function describing the advantages and disadvantages of the model are observed and analyzed. Then, the network parameters of the model are changed. The training phase of the model is a process of constantly changing parameters and optimizing. This process continues until the output data value converges and tends to a stable state or the number of iterations ends, and the model training ends. At this time, a classification model can be obtained and used in the image classification process. 3.3 Picture Classification and Recognition Process In the process of image classification, the verification image is used to test the network model obtained in the training process. First, the picture must be transformed by format, then the model generated by training will be used to classify and identify the pictures and test the effect of recognition. In the training process of the whole model, the image data set processed by format is first input into the network as the underlying data. The network will divide the data set into training data, verification data and test data according to the set proportion. The input image is convoluted in the convolution layer to extract image features and generate multiple feature maps. The human nervous system is very complex. In order to simulate the working process of neurons in the human visual nervous system, the convolutional neural network model allows each output feature to be calculated by the activation function and added with a bias value before outputting the features of the image. After passing through 2-layer convolution layer and 2-layer pooling layer, the data is then transmitted to the full connection layer, which completely converts the features into one-dimensional vector output. The convolution neural network model uses the loss function to describe the parameters to be optimized and the effect of the model. Through the loss function, the gap between the actual output result and the expected result is determined, and then the

Design of Online Teaching Method for Subject Knowledge

757

network parameters are continuously adjusted through the back propagation algorithm. When the value of the loss function is large, it indicates that the network parameters have not reached the optimum and have to be trained again, return to the beginning of the training and learn the features of the input picture again. When the value of the loss function is small and the network reaches the convergence state, it can be considered that the current state of the network parameters is the optimal state, and then decide whether to end the model training process. The number of iterations of training is also an indicator of whether the model training is completed. When the predetermined number of iterations is reached, the model training will end even if the model has not reached the optimal state, and the model at the end will be output. The final output model is the convolution neural network model constructed by us. The type and feature information of the image extracted through layer by layer operation are saved in this model. The above is a simple process to realize the convolution neural network model. It can be seen that the hierarchical structure of CNN is relatively clear. Each layer has its specific role and responsibility. Through continuous optimization again and again, the optimal model is finally output. Arduino is a platform for intelligent hardware design and development. It can connect various sensors and set specific coding instructions for products to adapt to different environments and realize specific functions. Arduino is open source, easy to learn, and with little knowledge, you can enter the colorful electronic world and experience the passion and fun brought by DIY. Because of this, there are abundant learning resources related to Arduino, but too many learning resources also make learners don’t know what to choose. Through the collection and screening of online learning resources, the author constructs the Arduino device learning resource library. The resource library mainly contains video resources. Video resources are the process of using Arduino device recorded by the author, and each step is accompanied by explanation. The Arduino device recognition program based on convolutional neural network will push the learning resources related to the identified device in the learning resource library after the successful recognition of the device.

4 Optimizing the Examination System of Mathematics Online Teaching Courses in Higher Vocational Colleges Online teaching has created a new teaching space for college teachers. In the past, face-to-face knowledge transfer has become an exchange on the Internet. Therefore, the previous offline teaching methods are not suitable for online teaching. In the process of online teaching, we need to give full play to the advantages of the Internet [10]. Teachers need to combine the characteristics of network teaching, focus on the network, and explore the characteristic content suitable for students’ listening habits. The teaching contents should keep pace with the times and the teaching methods should have network characteristics, so as to attract the attention of college students. By exploring the production of characteristic content, arouse students’ interest and improve learning efficiency. The design of teaching courseware needs to highlight the system design. A complete system design can bring the completion of the learning process and improve students’ learning experience. Therefore, schools and teachers need to combine the advantages of

758

C. Yu and J. Wang

the Internet and pay attention to the production of high-quality teaching content, which will be more practical in network teaching. Exploring the production of characteristic content also requires the state’s policy support for schools to promote the production of characteristic content in the process of internet teaching through financial support. Based on convolution neural network algorithm, this paper constructs loss function, integrates online teaching materials, and optimizes the online teaching course examination system of higher vocational mathematics. Through the whole process assessment, students do not dare to slack off in the classroom. They will treat online learning with a more focused attitude and maintain a high degree of concentration at all times. However, teachers can not give students too much pressure, improve the teaching effect by turning the classroom, and borrow domestic high-quality educational resources to improve students’ interest in learning. Promoting cooperative learning among students can cultivate students’ team consciousness and improve their interpersonal skills in the process of cooperation. For some contents that students can master through self-study, teachers can also let students experience the feeling of being a teacher and let them deepen their understanding of knowledge through classroom teaching. In terms of course homework, teachers can arrange more cooperative homework, enhance the communication between students, improve their practical ability, break the limitation of their single thinking in the process of cooperative learning, and let more ideas collide in the communication to produce new sparks. Learning motivation is an important factor affecting students’ learning. In the process of basic education in China, middle school students have a strong utilitarian learning goal, that is, in order to enter a better university and find a better job in the future. In the process of interview, the author found that some students lost their learning motivation after entering the University. Some courses won’t listen carefully in offline class and online. The loss of clear learning objectives leads to their laziness and laziness in the process of learning, resulting in low learning efficiency. Although it has been advocated that teaching should focus on “student-centered”, there are still many phenomena of teachers’ teaching and students’ passive learning in the process of classroom teaching in Colleges and universities. In the process of learning, students need to think about what their motivation is and what they study for. Preview in advance can better integrate into the learning environment. Students should be the subject of learning, not dominated by the outside world, and gradually clarify their learning motivation through independent exploration and self thinking [11]. In the offline teaching process, teachers and students coexist in the same specific learning environment. In this relatively closed learning space, students can feel a particularly strong sense of learning immersion in the learning process. Therefore, it is particularly important to create a classroom with active atmosphere and clear learning objectives. Teachers can cause students to interact by assigning tasks and asking questions, so that students can maintain a high degree of concentration and have a strong sense of learning immersion in the learning process. Teachers can add diversified elements in the process of curriculum design, such as flipped classroom design. First, let the students group freely online, put forward open questions, have interactive discussions between different groups, comment on the performance of each group, and encourage students to carry out inquiry learning. Second, make use of the advantages of the network to communicate with students online. Students are “digital natives”, who are more accustomed to the expression methods on the

Design of Online Teaching Method for Subject Knowledge

759

network. Teachers can interact with students through expression packs and catchwords, and conduct different interactions through the unique “like” and “bullet screen” of the online teaching platform. The third is to increase the presentation of short videos in classroom content. Short videos have a fast rhythm and can attract students’ attention. Making full use of the application of short video in classroom teaching can improve students’ learning concentration. Empathy is a psychological concept, which refers to the ability to understand their emotions and intentions from the perspective of others and express them accurately. In the critical period of epidemic prevention and control, there are differences in students’ region, background and learning conditions. Building a empathic classroom is conducive to promoting the development of teacher-student relationship and improving students’ learning effect. First, teachers need to integrate thoughts and emotions into teaching activities and design rich and attractive teaching contents. Empathy classroom has built a good communication platform between teachers and students. If teachers want to better integrate empathy into teaching practice, they need to flexibly adopt teaching methods and observe students’ learning enthusiasm and attitude, so as to improve the “temperature” of the classroom. Secondly, teachers should increase the process of interaction with students, trigger students’ thinking by asking questions, let students hear their own voice and the voice of others in the process of learning, and produce empathy in research and discussion. Realizing mutual learning and inspiration between teachers and students through empathy classroom is not only conducive to the transmission of knowledge in the teaching process, but also conducive to helping students shape a sound personality. Students constantly improve their ideological realm in the process of learning and communication. Such a empathy class will be more warm and deep.

5 Case Analysis The online teaching method of mathematics teachers’ subject knowledge in Higher Vocational Colleges Based on convolutional neural network designed in this paper is applied to practical teaching, and its application effect is verified. 5.1 Analysis of Current Teaching Situation Although higher mathematics in higher vocational education plays an irreplaceable function and role in vocational education and talent training, it is difficult to effectively integrate the two. It is unable to improve students’ mathematical ability according to the development demands of vocational education, so that the quality of mathematics teaching in higher vocational education can not be effectively improved. The main reasons are the following two aspects. First, the deviation of educational concept. Due to the obvious differences between higher mathematics and professional education in teaching content, teaching methods, teaching objectives and teaching system, higher vocational colleges ignore the relationship between the two in principle, mechanism and logic, which leads to the separation of mathematics teaching from the category of vocational education. The mathematical ability contained in mathematics teaching has not been “taken care of” and “set off” in the talent training system of higher vocational colleges, resulting

760

C. Yu and J. Wang

in the lack of close relationship between mathematics teaching and talent training. This “lack of connection” or “loose connection” stems from the deviation of the concept of vocational education. If we can enhance the relationship between the two at the macro level and top-level design, we can effectively improve the value of Higher Vocational Mathematics Teaching in vocational education and talent training. Secondly, the demand of enterprises for talents with logical reasoning ability, thinking ability, spatial imagination, mathematical operation, logical analysis and material generalization ability is not obvious, which makes higher vocational colleges pay more attention to the cultivation of teamwork ability, problem-solving ability, practical ability and learning ability, and the value of mathematics teaching in talent cultivation is not obvious. The teaching design of mathematics in Higher Vocational Colleges usually takes teachers as the guidance and students as the object, which is difficult to stimulate students’ initiative and enthusiasm in mathematics teaching, ignores the emotional communication between students and teachers, students and students, makes knowledge teaching the main body of mathematics teaching, and weakens students’ development and emotion in mathematics learning. In addition, due to the obvious textbook standard problem in higher vocational teaching design, teachers’ understanding process of teaching plans and teaching materials replaces students’ understanding and experience process, which affects the role of mathematics teaching in the cultivation of thinking ability, spatial sense and data generalization ability, and can not help higher vocational colleges better connect mathematics teaching with vocational education. In order to solve these problems, we need to take the modern educational concept as the starting point, take the students as the guidance, innovate the traditional teaching design content, process and form, and make the mathematics teaching design better reflect the professionalization, informatization and application characteristics of higher vocational education. In addition, we should make changes and innovations in the teaching process, teaching methods and teaching modes. Only in this way can mathematics teaching really meet the needs of vocational education and talent training. 5.2 Application Process and Results Because each dimension of the current situation of learning investment and each dimension of influencing factors are continuous variables, product moment correlation is adopted. If the correlation coefficient is positive, it indicates that they are positively correlated, and if the correlation coefficient is negative, it indicates that they are negatively correlated. The absolute value of the correlation coefficient indicates the strength of the correlation. The greater the absolute value of the correlation coefficient is, the stronger the correlation is, and the smaller the absolute value of the correlation coefficient is, the weaker the correlation is. Analyze the relationship between students’ learning input, and the analysis results are shown in Table 1:

Design of Online Teaching Method for Subject Knowledge

761

Table 1. Analysis results of influencing learning input Dimension

Overall learning engagement

Study preparation

learning motivation

learning environment

Learning organization and management

Overall learning engagement

1

–

–

–

–

Study preparation

0.445

1

–

–

–

learning motivation

0.468

0.547

1

–

–

learning environment

0.512

0.525

0.541

1

–

Learning organization and management

0.560

0.538

0.562

0.554

1

The data analysis shows that there is a significant positive correlation between the influencing factors (learning preparation, learning motivation, learning environment, learning organization and management) and learning investment, P The values are less than 0.01, and the correlation coefficient is between 0.445 and 0.562. In addition, from the absolute value of the correlation coefficient, the correlation from strong to weak is learning organization and management, learning environment, learning motivation and learning preparation. There is also a significant positive correlation between the influencing factors of learning investment, P All values are less than 0.01. Convolution algorithm is applied in online learning, which is an optimization algorithm for data set training. The main problem solved by the algorithm is that it can reduce the imbalance between positive and negative samples during training. By suppressing a certain number of negative samples, the loss curve converges faster and the training model is better. The core idea is to screen according to the loss of input samples, that is, to screen out the negative samples that have a great impact on classification and detection, and then train the model of the screened positive samples through stochastic gradient descent (SGD). The advantages of the algorithm are mainly reflected in two aspects: one is the online selection of negative samples, which is more targeted to the imbalance of data categories; Second, with the increase of data set samples and categories, the detection effect of the algorithm will be improved more obviously. The architecture of the algorithm is shown in Fig. 3:

762

C. Yu and J. Wang

Rol Pooling Layer

Diverse &Hard Rol Sampler

Fully Connected Lavers

Convolutional Feature Maps

Image

Fully Connected Lavers

Selective-SearchRoIs (R)

Fig. 3. Architecture of algorithm

As shown in Fig. 2, in order to test the practicability of the teaching method, the above algorithm is used to identify the students’ classroom performance. The hardware equipment of the experiment is a GTX 1080ti independent graphics card server with Intel Xeon CPU, 32 GB RAM and 11 GB video memory. The software device is Ubuntu 16 under Linux 04 operating system, the deep learning framework used is Darknet, and the image display function is realized by calling opencv and related libraries. There are two experimental data sets: one is a self built student behavior data set, which includes five kinds of behaviors: raising hands, listening, answering, sleeping, writing (hand. Listen. Answer. Sleep, write); The other is the standard voc2007 + 2012 dataset. The data set includes two subsets of voc207 and voc2012, which are a benchmark set for image classification, image recognition and target detection. The comparison of students’ scores before and after using this teaching method is shown in Table 2: Table 2. Application results Experimental class

Before application

After application

Class A

72.14

86.12

Class B

74.23

87.45

Class C

75.14

88.52

The application results are shown in Table 2. After applying the teaching method designed in this paper, the scores of the three classes have been significantly improved. It shows that the method designed in this paper has the effect of improving students’ performance. In order to verify the convolution neural network algorithm proposed in this paper, the methods in literature [2] and literature [3] are used as comparison methods for comparison experiments. The comparison results are shown in Table 3.

Design of Online Teaching Method for Subject Knowledge

763

Table 3. Comparison results of academic achievements Experimental class

Convolutional neural network algorithm

Literature [2] method

Literature [3] method

Class A

86.20

70.01

76.15

Class B

90.35

74.51

73.14

Class C

89.94

73.04

72.84

It can be seen from table 3 that after the application of the teaching method designed in this paper, the scores of the three classes are more than 85, while the scores of the two literatures are less than 80. The results show that compared with the two literature methods, the scores of the three classes are higher after the application of the teaching method designed in this paper.

6 Conclusion This paper grasps the opportunities brought by online teaching and bravely meets new challenges. It improves the efficiency of online teaching. Through the early large-scale online teaching practice, it has accumulated many problems in online teaching. This paper can seize the opportunity brought by online teaching with a more open mind. Based on convolution neural network algorithm, this paper constructs loss function, integrates online teaching materials, and optimizes the online teaching course examination system of higher vocational mathematics. Through the integration of Internet and education, we can realize more advanced, more human and more perfect educational ideas, and help us better deal with the challenges from the future. In the future development, this paper discusses the application of artificial intelligence technology and advanced teaching methods in the design of online teaching methods for Higher Vocational Mathematics Teachers’ subject knowledge, and discusses the distance teaching methods. Fund Project. Jilin Province Vocational Education Research Project: Research on the Ways to Improve the Subject Knowledge and Teaching and Research Ability of Mathematics Teachers in Higher Vocational Colleges (2021XHZ043).

References 1. Bragg, L.A., Walsh, C., Heyeres, M.: Successful design and delivery of online professional development for teachers: a systematic review of the literature. Comput. Educ. 166(6), 104158 (2021) 2. Martínez, S., Guíñez, F., Zamora, R., Bustos, S., Rodríguez, B.: On the instructional model of a blended learning program for developing mathematical knowledge for teaching. ZDM Math. Educ. 52(5), 877–891 (2020). https://doi.org/10.1007/s11858-020-01152-y 3. Santagata, R., Lee, J.: Mathematical knowledge for teaching and the mathematical quality of instruction: a study of novice elementary school teachers. J. Math. Teach. Educ. 24(1), 33–60 (2019). https://doi.org/10.1007/s10857-019-09447-y

764

C. Yu and J. Wang

4. Xie, S., Cai, J.: Teachers’ beliefs about mathematics, learning, teaching, students, and teachers: perspectives from Chinese high school in-service mathematics teachers. Int. J. Sci. Math. Educ. 19(4), 747–769 (2020). https://doi.org/10.1007/s10763-020-10074-w 5. Zhang, H., Cai, J.: Teaching mathematics through problem posing: insights from an analysis of teaching cases. ZDM Math. Educ. 53(4), 961–973 (2021) 6. Zahner, W., Aquino-Sterling, C.R.: Are the words as important as the concepts? Using pedagogical language knowledge to expand analysis of mathematics teaching with linguistically diverse students. Math. Educ. Res. J. 34, 457–477 (2020). https://doi.org/10.1007/s13394020-00352-9 7. Ndlovu, M., Ramdhany, V., Spangenberg, E.D., Govender, R.: Preservice teachers’ beliefs and intentions about integrating mathematics teaching and learning ICTs in their classrooms. ZDM Math. Educ. 52(7), 1365–1380 (2020). https://doi.org/10.1007/s11858-020-01186-2 8. Pan, J., Wen, L., Qi, L., Xing, D., Zhao, Y.: Integrated reform of electronic technology courses in higher vocational colleges based on the OBE concept. J. Contemp. Educ. Res. 5(9), 20–25 (2021) 9. Zhang, X.: Teaching reform and research on art design majors in higher vocational colleges under the background of “double high plan.” J. Contemp. Educ. Res. 5(1), 4 (2021) 10. Jiang, L.: Analysis and enlightenment of daily education and training of student party members in higher vocational colleges. The Science Education Article Collects (2020) 11. Mi, Q.C., Zhao, H.M., Lin, L.P.: Unstructured data annotation based on multichannel convolutional neural network. Comput. Simul. 38(6):400-404 (2021)

3D Reconstruction Method of Virtual Teaching Laboratory Model Based on Akaze Features Mingxiu Wan1(B) and Yangbo Wu1,2 1 College of Mathematics and Computer, Xinyu University, Xinyu 338000, China

[email protected] 2 Faculty of Computing and Informatics, University Malaysia Sabah, 88000 Sabah, Malaysia

Abstract. In order to improve the accuracy of 3D modeling of virtual laboratory, a 3D reconstruction method of virtual teaching laboratory model based on AKAZE feature is proposed. In the gradient image, all Gaussian weighted vectors in the region are superposed with the feature points as the center to match the features of the virtual teaching laboratory. Logarithmically increasing according to AKAZE’s scale level. The 3D points are searched through breadth-first traversal, and an association graph based on AKAZE features is constructed. Estimate the global rotation matrix, optimal translation matrix and global position information, and obtain all the information of the 3D model library. Through the steps of material acquisition, 3D graphics construction, 3D device construction, virtual object determination and virtual teaching laboratory model release steps, the 3D reconstruction of the model is completed. It can be seen from the experimental results that the maximum matching error between this method and the actual extracted features is 6, which can accurately match the three-dimensional features and achieve the ideal reconstruction effect. Keywords: AKAZE features · Virtual teaching · Laboratory model · 3D reconstruction

1 Introduction With the continuous development of virtual reality technology and network communication technology, many colleges and universities have put forward their own virtual laboratory construction plans. In the virtual laboratory, students can either operate on the virtual experimental platform or design their own experiments. Compared with traditional teaching methods, students can understand the learning content more intuitively. 3D simulation model is an important part of virtual laboratory. For students, there are usually three sources for the 3D models in the virtual lab. They are self-developed, downloaded from the Internet, and provided by teachers. The 3D models obtained usually have many problems. For example, the types of models are limited, the format is not uniform, and the adjustment parameters are inconsistent. For this reason, there are also some researches in the field of simulation model resource library at home and abroad. © ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2022. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 765–777, 2022. https://doi.org/10.1007/978-3-031-21161-4_58

766

M. Wan and Y. Wu

Someone proposed a virtual laboratory reconstruction method based on the network 3D model library. This method reduces the coupling between the three-dimensional model and the virtual scene by distinguishing the model library and the database, and storing the model and data separately. The model library is to draw the 3D models repeatedly used in the virtual laboratory with professional modeling tools, and store them into the model library according to the model classification. The database stores the information related to the establishment of the scene according to the specific virtual scene. This completes the three-dimensional reconstruction of the virtual teaching laboratory model [1]. Others have proposed a virtual reality-based virtual laboratory reconstruction method. Using virtual modeling tools to achieve realistic 3D scenes, users can roam in the virtual scene. Learning resources are obtained by interacting with the 3D model in the scene through the keyboard and mouse [2]. However, most researches mainly focus on the management of the model itself, without considering the data information generated by the interaction between the 3D model and the scene. For a virtual scene with interactive functions. In addition to the description information of the model itself, it should also include the parameter information of the model in the scene, as well as the data entered by the user. Moreover, in the process of 3D reconstruction of the model, the accuracy of feature matching needs to be improved. Therefore, a 3D reconstruction method of virtual teaching laboratory model based on AKAZE features is proposed. Centered on the feature point on the gradient image, all Gaussian-weighted vectors in the region are superimposed to match the features of the virtual teaching laboratory. According to the scale level of AKAZE, it increases logarithmically, and searches for three-dimensional points through breadth-first traversal, thereby constructing an association graph based on AKAZE features. The three-dimensional reconstruction of the model is completed through the steps of material acquisition, three-dimensional graphics construction and virtual teaching laboratory model release.

2 Feature Matching of Virtual Teaching Laboratory Based on AKAZE Features Virtual laboratory is an open networked virtual experiment teaching system based on VR virtual reality technology. It is the digitization and virtualization of various existing teaching laboratories [3]. The Virtual Lab is designed to create a real virtual environment. This environment not only has realistic experimental scenes, experimental instruments, and experimental equipment, but also the virtual experimental equipment constructed must be able to achieve a fine-grained three-dimensional display that is operable and observable, so as to truly restore the experimental process. AKAZE (Accelerated KAZE) feature extraction algorithm is a local feature descriptor algorithm, which can be regarded as an improvement of the SIFT algorithm. It uses nonlinear diffusion filtering iterations to extract and construct the scale space, and uses a method similar to SIFT to find feature points. In the descriptor generation stage, a similar method of ORB is used to generate descriptors, but the descriptor has more rotation invariance features than ORB. The construction of the existing virtual reality system is slow and the delay is large. A laboratory often includes buildings, buildings, ceilings, floor tiles, water and electricity, instruments, equipment, operating desks, blackboards, lecterns, chairs, cabinets and

3D Reconstruction Method of Virtual Teaching Laboratory Model

767

many other elements. And its elements influence each other. To complete an experiment, it is necessary to comprehensively use each element, its attributes, and functions to achieve a good experimental teaching effect. After determining the virtual teaching laboratory model, AKAZE can use the Hessian matrix to detect local extreme points. The Hessia matrix of a point in a two-dimensional image describes the gray gradient changes in the neighborhood of the point in all directions:   (1) Ri = λ2 Rxxi Ryyi − Rxyi Rxyi In formula (1), λ is the normalized scale factor in the corresponding image group. The maximum value is obtained by comparing each pixel processed by the Hessian matrix with its 8 neighboring pixels of the same scale and the pixels in the 3 × 3 window on

(a) The first-order differential

Rx

and

Ry

are in the same quadrant

(b) First order differential

Rx

and

Ry

in different quadrants

(c) The first-order differentials

Rx

and

Ry

are not exactly in quadrants

Fig. 1. AKAZE computes eigenvalue directions

768

M. Wan and Y. Wu

two adjacent scales. The pixel point where the maximum value is located is the feature point. If the two points are in the adjacent layer and the distance is within the scale range, the duplicate point is considered to be deleted [4]. The main direction of the feature point is determined before the feature description. On the gradient image, with the feature point as the center and the radius λ as the statistical range, the Gaussian weighting operation is performed on the first-order differential Rx and Ry of the neighborhood of the feature point. Then take a 60° fan-shaped window to traverse the entire circular area around the feature point, and superimpose all Gaussian weighted vectors in the area. The direction represented by the fan-shaped area with the highest superposition value is the main direction of the eigenvalue, as shown in Fig. 1. It can be seen from Fig. 1 that the feature point neighborhood is divided into several grids such as 3 × 3 or 4 × 4. And rotate in the main direction. Discrete points are obtained by resampling at intervals of scale λ in each grid. Calculate the average gray value of discrete points and the first-order gradient information in the horizontal and vertical directions. The descriptor information of 3-bit dimension is generated, which can make the descriptor have rotation invariance and distinguishability [5].

3 3D Reconstruction of Virtual Teaching Laboratory Model Based on AKAZE Features 3.1 Association Graph Construction Based on AKAZE Features After feature matching on the image sequence, the spatial relationship of objects in different images is usually related to the sequence of the image sequence. Therefore, it is necessary to establish the neighbor relationship between images: if there are a large number of matching pairs between the two images, it means that there is an adjacent spatial relationship between the two images. In a multi-view image sequence, each image has adjacent pictures. The relationship between common feature points in an image sequence can be described by establishing a trajectory track. For example, suppose that a point in the space corresponds to the feature point in the three images f1 f2 f3 . The feature f1 in the first image matches the feature f2 in the second image. The f2 ‘s in the second image match the features in the third image. These features can then be concatenated to form a set {f1 , f2 , f3 }[6]. A track vector is formed by combining the number of the image and the number of the feature points. A track describes a common feature point in an image sequence and also corresponds to a point in 3D space [7]. According to the scale level of AKAZE, it increases logarithmically, there are n groups, and each group contains m layers. The layers in each group use the same scale as the original image. Based on this, the resulting scale values are: αi (n, m) = α0 2n+m/m

(2)

In formula (2), α0 represents the initial value of the scale. The nonlinear diffusion filter model is in units of time. Hence the need for a scale parameter in pixels [8]. For an input image, after Gaussian filtering, the gradient histogram of the image can be calculated to construct the conduction matrix Wi . The step size ι is obtained from a

3D Reconstruction Method of Virtual Teaching Laboratory Model

769

set of evolutionary time differences. All images in nonlinear scale space can be obtained through the AOS algorithm, and the scale space of image A is expressed as: Wi+1 = (H + τ Wi )Wi

(3)

In formula (3), H represents an identity matrix. τ represents the step size. Search for a track of 3D points by breadth-first traversal. First sort the feature points of all image sequences. Then, for each feature point, it is searched by means of breadthfirst traversal to find out whether the neighbor image of the image has its corresponding feature point. If found, it is considered that the feature point belongs to a track. Then new neighbors are added to the queue, and the neighbor image search continues until all images are found [9]. 3.2 Image Model Building The virtual teaching laboratory model based on AKAZE features consists of three parts: model library and database, scene management system, and visual management, as shown in Fig. 2. Network model library

Local model library

Model base management system

Scenario database

Database management system

Scenario Data Resources

Scene organization

Apply colours to a drawing

Fig. 2. A virtual teaching laboratory model based on AKAZE features

3D Model Library Global location information exists in the 3D model library. Extracting information related to feature matching from this information requires accurate estimation. The detailed steps are: Step 1: Global Rotation Matrix Estimation In the undirected association graph obtained by the improved method, there are three-view constraints between images, which are better than two-view constraints. The global matrix can be estimated from the relative rotation matrix between the images in the correlation graph.

770

M. Wan and Y. Wu

In the correlation graph, an image Ii , Ij is provided. The relationship between its relative rotation matrix Eij and their global rotation matrix Ei , Ej is: Eij = Eij Ej

(4)

where, the relative rotation matrix is obtained by decomposing the eigenmatrix obtained by matching pairs. Assuming that there are n pairs of matching images in the image set, there should be n − 1 relative rotation matrices, which are substituted into the above equation. The estimated value of the global rotation vector can be obtained by the least squares method, and the estimated value should satisfy the orthogonality [10]. Step 2: Optimal translation matrix estimation To solve the global translation matrix, the local translation matrix under three views should be obtained first. The solution method of the local translation matrix under three views is described as follows [11]. First, the translation vector of the camera in the three-view local coordinate system is calculated. Then calculate the reprojection error of the matching points, and delete the points with too large errors. Finally, the optimal translation matrix in the local  coordinate system is obtained. , I , I For three-view I i j k in a three-view set. Their respective rotation matrices are   Ei , Ej , Ek . The set of spatial points corresponding to the three-view matching points  is O. There is a point O1 in O, and the corresponding pixel point in image Ii is gi , gj . Calculate the reprojection error for this point. Step 3: Global location information estimation   Through the above steps, the relative rotation and translation Eij , Tij of each image in the image set I are obtained. Then the global position information of all image feature points can be obtained. Ideally, the relationship between the spatial position of the image Ii , Ij and their relative pose transformation is as follows: Qj − Wij Qi = εij Tij

(5)

In formula (5), εij represents a scale factor. Q represents global location information. The above equation is an equation under ideal conditions, but in fact it cannot be completely eliminated because of the noise data. As well as the physical errors generated by the camera process, the above formula cannot be strictly established. The actual situation is as follows: There is a point A in the space at the global position. Its pixel positions in the two photos are point B and point C, respectively. These three points should represent the same point, but in fact there is a certain deviation. The optimal solution of the scale factor can be considered to minimize the Euclidean distance between point B and point C relative to point A. Scene Management System The scene management system includes two parts: the model library management system and the database management system. The model management system is independent of the specific application field. A software system that classifies and maintains models and supports model generation, storage, query, execution, and analysis applications. Experts and users in various fields can use the model library management system to find the 3D

3D Reconstruction Method of Virtual Teaching Laboratory Model

771

models they need. Put into the virtual scene, the database management system realizes the storage and reading of the virtual scene data. The model base management system and the database management system jointly manage the objects in the virtual scene. When the user selects an existing virtual scene, the database management system first accesses the scene database. Take out each piece of data, get the index information, position parameters, and interactive data of the 3D model in the model library. Pass the index number to the model repository management system to find the model. Restore the model to a specific location in the scene based on parameter information. When the user clicks on the model, the interactive information of the model is displayed. When the scene is reproduced, the model library management system provides two ways to call the model. One is to directly call the 3D model in the network model library to the virtual scene. The other is to download the models in the network model library to the local model library in advance and then call them. Visual Management Virtual reality takes full advantage of scientific visualization techniques. Use virtual modeling tools to achieve realistic 3D scenes. Users can roam in the virtual scene, interact with the 3D model in the scene through the keyboard and mouse, and obtain learning resources. The visual management part is mainly responsible for establishing virtual scenes, importing 3D models, realizing scene roaming, and managing interactive operations. The rendering of the scene is realized by virtual reality technology. 3.3 Calculation of Global Rotation Matrix and Translation Matrix After using the RANSAL algorithm to filter out the wrong two-view matching relationship, it is necessary to select a reliable two-view rotation matrix as a parameter, and based on this, the global rotation matrix can be solved. First, the maximum spanning tree is used to calculate the initial global rotation matrix. Then calculate the distance between the multiplication of the rotation matrix in a ring and the identity matrix to judge whether the relationship between the two views is correct. After the two-view relationship is obtained, the global rotation matrix can be solved based on this. To solve the global translation matrix, the local translation matrix under three views should be obtained first. The solution method of the local translation matrix under three views is as follows. First, the translation vector of the camera in the three-view local coordinate system is calculated. Then calculate the ghosting error of the matching points, and delete the points with too large errors. Finally, the optimal translation matrix in the local coordinate system is obtained. 3.4 Rebuild Process Design According to the scale and characteristics of the virtual laboratory, as well as the tools and conditions of the developers, the system in this paper mainly uses field photography to collect data. 3DMax is used as the 3D modeling tool, and Photoshop is used for texture processing of the material collected from the field. Import the model into Unity as a development engine, and import the model into the engine to add components to add

772

M. Wan and Y. Wu

interactive features. Finally, use the test and release system. There may occasionally be an issue in VR environments opening 3D models whose textures are not correctly mapped to the underlying model mesh. Poorly effective texture mapping can produce noticeable distortions in the appearance of the model’s surface, such as unwanted seams, stretched or squeezed areas in the texture pattern. It is therefore necessary to use Photoshop to select the reparameterization option, edit and create the optimal surface coverage model. Material Collection The materials collected by the virtual laboratory materials mainly include the topography and elevation data of the entire main building and the campus, electronic photos of the main building, sidewalks, squares, pools, greening, other facilities, and experimental equipment. Take pictures on the spot and use photoshop for image optimization. The main thing is to crop and denoise the image first. And use the magic pen to cut out the parts that need to be textured to meet the needs of model building. 3D Graphics Construction Use 3DMax to construct the graphic image to form the terrain near the main building. Plan and satellite maps for the main building. First, make topographic contours to provide height information for topography. Use ArcMap to generate virtual layer elevation grades, marked with different levels of grayscale. Create a new drawing plane in 3DMax, and make the x and y directions according to the degree of detail. In the UVN map editor, add a picture taken in the field. 3D Device Construction Laboratory devices and instruments are mainly photographed and combined with advanced modeling tools, and the color and texture of the oscilloscope are photographed. Shapes and effects are obtained using the advanced modeling tools Mesh, Patch and NURBS available in 3DMax. Different lines have different thicknesses and need to be rendered in layers to get the best realistic effect. Complete the construction and synthesis of small objects such as oscilloscopes, power supplies, spectrum analyzers, and resistors. In the implementation process, the camera view can be combined with the 3D view for modeling, which can reduce the complexity of modeling as much as possible. Virtual Object Determination Combined with AKAZE features, the 3D model created by 3Dmax is imported into the 3D engine, making it a virtual object that can be called and processed. Then add dynamic components and custom components to add interactivity to virtual objects. Complete human-computer interaction, element interaction and visual interaction between various virtual objects. To build a virtual laboratory system. Virtual Teaching Lab Model Released After the processing of AKAZE features, the virtual laboratory can run in the 3D engine. As shown in Fig. 3.

3D Reconstruction Method of Virtual Teaching Laboratory Model

773

Fig. 3. 3D model of virtual teaching laboratory

As can be seen from Fig. 3, in order to improve the applicability and compatibility of the system, the virtual laboratory needs to be transformed into an executable file, which can be applied to each user’s system. Therefore, combined with the publishing function of AKAZE, a virtual laboratory platform that can be used by multiple platforms can be realized.

4 Simulation The test of the 3D reconstruction method of virtual teaching laboratory model based on AKAZE features proposed in this paper is an important link in the development of virtual laboratory. Only through complete system testing can the final effect of virtual laboratory development be verified. Ensure that the developed platform can fully realize the design goals and meet the needs of users. 4.1 Simulation Process The functional tests of the virtual laboratory mainly include basic functional tests (such as student login test, administrator login test, teacher login test) and main functional tests (scene roaming test, experimental teaching test, experimental operation test, online question answering test). Through the tests of these three aspects, the function of the virtual laboratory is explicitly output, and it is tested whether the actual operation of the virtual laboratory meets the functional design and user needs. The specific test flow chart is shown in Fig. 4.

774

M. Wan and Y. Wu Stude nt Login test

Teac her login test

Admi nistra tor Login Tests

Students enter the virtual lab login page

Scene roaming test Experimental demonstration test Experimental operation test Online quiz

Fig. 4. Virtual laboratory simulation process

4.2 Feature Matching Result Analysis The virtual laboratory reconstruction method based on network 3D model library (Reference [1]), the virtual laboratory reconstruction method based on virtual reality (Reference [2]), the learning signed distance 3d object reconstruction from static images (Reference [3]) and the 3D reconstruction method of virtual teaching laboratory model based on AKAZE feature are used respectively. The feature matching results are compared and analyzed, as shown in Table 1. Table 1. Comparison of feature matching results of three methods Actual extraction quantity/piece

Network 3D model library

10951

10098

20290 23452

Virtual reality

Learning signed distance methods in static images

AKAZE features

9892

10021

10951

19982

19082

19652

20290

22983

23009

22086

23452

23435

22203

23123

22725

23435

36754

35784

34278

34029

36760

It can be seen from Table 1 that the virtual laboratory reconstruction method based on the network 3D model library, the virtual laboratory reconstruction method based on virtual reality and the signed distance learning method in static images cannot completely match the features, and the maximum matching errors are 1232, respectively. 2476 and 2725. However, using the 3D reconstruction method of the virtual teaching laboratory model based on AKAZE features, the maximum matching error is 6.

3D Reconstruction Method of Virtual Teaching Laboratory Model

775

4.3 Analysis of Reconstruction Renderings The virtual laboratory reconstruction method based on network 3D model library, the virtual laboratory reconstruction method based on virtual reality and the 3D reconstruction method of virtual teaching laboratory model based on AKAZE feature are used respectively. The reconstruction effect diagram of the virtual teaching laboratory model is compared and analyzed, as shown in Fig. 5.

Fig. 5. Three methods of virtual teaching laboratory model reconstruction renderings

776

M. Wan and Y. Wu

It can be seen from Fig. 5 that the reconstruction method of virtual laboratory based on the network 3D model library, the reconstruction method of virtual laboratory based on virtual reality and learning the symbolic distance in static images have wrong matching features, resulting in poor reconstruction effect. The 3D reconstruction method of the virtual teaching laboratory model based on AKAZE features can accurately match the 3D features, so that the reconstruction results can achieve ideal results.

5 Conclusion The proposed 3D reconstruction method of virtual teaching laboratory model based on AKAZE features. The AKAZE algorithm is used to detect and extract the features of the image sequence, and the RANSAC algorithm is used to further optimize the matching results and save the coordinate information of the correct matching pairs. Then build an undirected correlation graph and obtain a global rotation and translation matrix, thus completing the three-dimensional reconstruction of the virtual teaching laboratory model. For example, using CUDA parallelization to speed up the point cloud construction process. Using this model in experimental teaching can effectively exercise students’ logical analysis ability and practical ability. Carry out research based on 3D reconstruction and carry out a large number of experimental verifications, obtain target scene pictures through digital cameras, and go through a series of processes. Compared to incremental 3D reconstruction, a more satisfactory reconstructed model is obtained, but there is still a lot of room for improvement. For problems that can be improved and researched, it boils down to the following points: (1) The 3D reconstruction is a dense point cloud, and the model is not realistic enough. In the follow-up, we can also study the implementation methods of point cloud registration, meshing, texture mapping and other steps. (2) Compared with the incremental reconstruction method, the speed of the method in this paper is greatly improved. In addition to algorithm optimization, hardware acceleration can also be considered. (3) In the 3D model reconstructed by this method, the dense point cloud will contain points other than the target. These points are generated successfully but are not needed. At present, these clutter points can only be removed manually, and it is one of the directions worth researching to make the machine automatically identify and remove these points.

Fund Project. Science and technology project of Jiangxi Provincial Department of Education: Application Research of 3D reconstruction algorithm based on akaze in Virtual Laboratory (Project No.: 003011485).

References 1. Wen, P.Z., Lei, Y.Q., Sun, M.L.: Defective hole identification and hole-filling for 3D reconstruction mesh models. Appl. Res. Comput. 37(04), 1234–1238 (2020)

3D Reconstruction Method of Virtual Teaching Laboratory Model

777

2. Peng, Y., Wang, A.D., Wang, T.T., et al.: Three-dimensional reconstruction of carp brain tissue and brain electrode for biological control. J. Biomed. Eng. 37(05), 885–891 (2020) 3. Lin, C.H., Wang, C., Lucey, S.: SDF-SRN: learning signed distance 3D object reconstruction from static images. In: Advances in Neural Information Processing Systems 33, pp. 11453– 11464 (2020) 4. Huo, P.P., Hou, Q.M., Zhou, Q., et al.: Research on 3D reconstruction method based on multiple data sources: a case study of the Ming Great Wall in Beijing. Bull. Surv. Mapp. S1, 262–267 (2020) 5. Zhang, H.X., Fang, Y.T., Li, M.: Robust reconstruction method of 3D room layout with visual-inertial module. J. Comput. Aided Des. Comput. Graph. 32(02), 262–269 (2020) 6. Qi, H.F., Wang, Y.: Research of web-oriented simplifying mechanical products’ 3D model. Comput. Simul. 38(11), 280–283+289 (2021) 7. Jin, Y.T., Lv, J., Pan, W.J., et al.: Optimization of virtual interactive interface layout based on visual attention mechanism. Comput. Eng. Des. 41(03), 763–769 (2020) 8. Sucar, E., Wada, K., Davison, A.: NodeSLAM: neural object descriptors for multi-view shape reconstruction. In: 2020 International Conference on 3D Vision (3DV), pp. 949–958. IEEE (2020) 9. Xie, X.: Three-dimensional reconstruction based on multi-view photometric stereo fusion technology in movies special-effect. Multimedia Tools Appl. 79(13–14), 9565–9578 (2019). https://doi.org/10.1007/s11042-019-08034 10. Malik, A., Lhachemi, H., Ploennigs, J., et al.: An application of 3D model reconstruction and augmented reality for real-time monitoring of additive manufacturing. Procedia CIRP 81, 346–351 (2019) 11. Yang, X., Kahnt, M., Brückner, D., et al.: Tomographic reconstruction with a generative adversarial network. J. Synchrotron Radiat. 27(2), 486–493 (2020)

Author Index

C Cai, Jiajing II-363 Cai, Jiang II-146, II-160 Chao, Yan I-463, I-589 Chen, Chen I-463, I-589 Chen, Jin I-668, II-94 Chen, Lin II-3, II-199 Chen, Wei II-175, II-573 Chen, Zhiwen II-524 D Deng, Fangyan II-509 Dong, Xiuying I-129, II-186 Du, Fuyu I-351, I-365 F Fan, Qinpei II-270, II-654 Feng, Baoling I-378, II-671 Feng, Junying II-363 Fu, Weina II-410, II-423 Fu, Zhaoqi I-39, I-50 G Gao, Caifeng I-118, I-565 Gao, Feng I-324 Gao, Jie II-334, II-347 Gu, Hongmei II-3, II-199 Guo, Zhongwen II-436 H Hao, Qiong I-639, II-524 Hu, Yuexing I-324 Hua, Wen II-495, II-509 Huang, Hongping II-132 Huang, Li II-465 Huang, Yanming II-318 Huang, Yi II-334, II-347 Huang, Yingying I-3, I-601

J Jia, Yueguo I-91, I-103 Jian, Zhixiong II-175, II-573 Jin, Wei I-324 L Lei, Hongmei II-16, II-436 Li, Chenyuyan II-451 Li, Huijuan I-129, II-186 Li, Jin II-451 Li, Jingbo II-305, II-401 Li, Li I-229 Li, Meifu II-305, II-401 Li, Min I-185, I-211 Li, Nan I-185, I-211 Li, Zan I-513 Lin, Hui I-683, II-132 Lin, Ying II-696 Liu, Fuxia I-338, II-627 Liu, Gaoming II-465 Liu, Kun I-338 Liu, Leiguang I-709 Liu, Lu I-324 Liu, Mengyang I-3, I-601 Liu, Ping I-668 Liu, Shiliang I-118 Liu, Shiyuan I-338 Liu, Shuai II-410, II-423 Liu, Wen I-50 Liu, Yi II-495 Long, Qu I-639, II-524 Lu, Jie I-668, II-94 Luo, Yingji I-281, II-216 Luo, Yuchan I-391, II-293 Lv, Changmin I-3, I-601 M Ma, Lei I-709, I-724 Ma, Na I-695, II-562 Ma, Qinghui I-430, II-318

© ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering 2022 Published by Springer Nature Switzerland AG 2023. All Rights Reserved W. Fu and G. Sun (Eds.): eLEOT 2022, LNICST 453, pp. 779–781, 2023. https://doi.org/10.1007/978-3-031-21161-4

780

Author Index

Mao, Lingli I-197, I-229 Meng, Lingjun II-654 Meng, Shangyu II-363 Miao, Li II-231, II-388 N Ng, Giap Weng II-696 Ning, Xiaojie I-28 P Pang, Ying

I-668

Q Qi, Jianli II-482 Qiao, Daifu II-44, II-482 Qu, Mingfei I-415, I-445 S Shao, Lili II-588 Shen, Menglai I-668 Shi, Jinmei II-363 Song, Chao I-229 Song, Dawei I-542 Song, Haolin I-542, I-654 Sun, Rong I-39, I-50 T Tang, Wen

I-62, I-513

W Wan, Mingxiu I-765, II-696 Wang, Fei I-430, II-318 Wang, Haijun II-81, II-614 Wang, Jiajie I-615, II-69 Wang, Junyan I-751, II-685 Wang, Lina II-671 Wang, Linan I-378 Wang, Ni II-16, II-436 Wang, Rui II-482 Wang, Shuang I-430 Wang, Song I-578, II-56 Wang, Xin II-108, II-120 Wang, Yue II-246, II-257 Wu, Jiaofeng I-487, I-499 Wu, Yangbo I-765, II-696 X Xiao, Lin I-338 Xiao, Ying I-445

Xin, Qiyuan II-654 Xiu, Anteng I-529, II-31 Xu, Changdong I-578, II-56 Xu, Pan II-120 Xu, Zhichao I-197, I-229 Xue, Xiaobo I-296, I-311 Y Yan, Jichao II-160 Yan, Yipin II-495, II-509 Yang, Hongxue I-709 Yang, Jianxing I-724 Yang, Kai I-668 Yang, Lei I-91, I-103 Yang, Qiuhui I-281 Yao, Jianfeng II-363, II-465 Ye, Ying I-269, II-451 You, Dunke I-50 Yu, Chunyan I-751, II-685 Yu, Dan I-158, I-170 Yu, Hongbo II-436 Yu, Rong I-14, I-474 Yu, Yi II-81, II-614 Z Zang, Peng II-588, II-601 Zhan, Jing II-282, II-640 Zhang, Jianhua I-281, II-216 Zhang, Jie I-695, II-562 Zhang, Lan I-296, I-311 Zhang, Lirong I-415, I-445 Zhang, Lizhen I-324 Zhang, Mingming II-146 Zhang, Pan II-108, II-120 Zhang, Shujun I-3, I-601 Zhang, Weiwei I-487, I-499 Zhang, Xin I-403, I-737 Zhang, YaJuan II-363 Zhang, Yiqian II-246, II-257 Zhang, Zhaohu I-62, I-513 Zhang, Zhi I-243, I-627 Zhao, Dan I-403, I-737 Zhao, Haiyan I-338, II-627 Zhao, Na I-255 Zhao, Yonghao II-573 Zhen, Yankun I-542, I-654 Zheng, Fayue I-709, I-724 Zheng, Jingya II-146, II-160 Zheng, Zhongwen I-255, II-377

Author Index

Zhou, Haixia I-28 Zhou, Jun I-683, II-132 Zhou, Ping II-465 Zhou, Qian II-231, II-388

781

Zhou, Zhiyu II-537, II-549 Zhu, Chaojun I-145, I-324 Zhu, Jing I-75, I-351, I-365, I-553 Zou, Xiang I-75, I-553