Environmental Issues of Blasting: Applications of Artificial Intelligence Techniques (SpringerBriefs in Applied Sciences and Technology) 9811682364, 9789811682360

Citation preview

SPRINGER BRIEFS IN APPLIED SCIENCES AND TECHNOLOGY

Ramesh M. Bhatawdekar Danial Jahed Armaghani Aydin Azizi

Environmental Issues of Blasting Applications of Artificial Intelligence Techniques 123

SpringerBriefs in Applied Sciences and Technology

SpringerBriefs present concise summaries of cutting-edge research and practical applications across a wide spectrum of fields. Featuring compact volumes of 50 to 125 pages, the series covers a range of content from professional to academic. Typical publications can be: • A timely report of state-of-the art methods • An introduction to or a manual for the application of mathematical or computer techniques • A bridge between new research results, as published in journal articles • A snapshot of a hot or emerging topic • An in-depth case study • A presentation of core concepts that students must understand in order to make independent contributions SpringerBriefs are characterized by fast, global electronic dissemination, standard publishing contracts, standardized manuscript preparation and formatting guidelines, and expedited production schedules. On the one hand, SpringerBriefs in Applied Sciences and Technology are devoted to the publication of fundamentals and applications within the different classical engineering disciplines as well as in interdisciplinary fields that recently emerged between these areas. On the other hand, as the boundary separating fundamental research and applied technology is more and more dissolving, this series is particularly open to trans-disciplinary topics between fundamental science and engineering. Indexed by EI-Compendex, SCOPUS and Springerlink.

More information about this series at https://link.springer.com/bookseries/8884

Ramesh M. Bhatawdekar · Danial Jahed Armaghani · Aydin Azizi

Environmental Issues of Blasting Applications of Artificial Intelligence Techniques

Ramesh M. Bhatawdekar Department of Mining Engineering Indian Institute of Technology Kharagpur Kharagpur, India Geotropik, Centre of Tropical Geoengineering Department of Civil Engineering Universiti Teknologi Malaysia Johor Bahru, Malaysia

Danial Jahed Armaghani School of Civil and Environmental Engineering University of Technology Sydney (UTS) Sydney, Australia Department of Urban Planning Engineering Networks and Systems Institute of Architecture and Construction South Ural State University Chelyabinsk, Russia

Aydin Azizi School of Engineering, Computing and Mathematics Oxford Brookes University Oxford, UK

ISSN 2191-530X ISSN 2191-5318 (electronic) SpringerBriefs in Applied Sciences and Technology ISBN 978-981-16-8236-0 ISBN 978-981-16-8237-7 (eBook) https://doi.org/10.1007/978-981-16-8237-7 © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

About This Book

Blasting is an important operation in any mining or civil engineering projects for breaking hard rock. During blasting, only 25–30% explosive energy is utilized for breaking rock for desired fragmentation, throw, and formation of muck pile. Balance energy is converted into undesired environmental effect of backbreak, flyrock, air over pressure, and ground vibration. With the advancement of artificial intelligence and machine learning techniques, accuracy in prediction of environmental effects due to blasting has improved. This book covers the successful use of these techniques in predicting, minimizing, and controlling the mentioned blasting environmental issues. A critical and state-of-the-art review of the available artificial intelligence and machine learning models in solving the blasting environmental issues is provided in this book.

v

Contents

1 An Overview of Blasting Operations and Possible Techniques to Solve Environmental Issues of Blasting . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 Blast Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Environmental Effect Due to the Blasting . . . . . . . . . . . . . . . . . . . . . . . 1.3.1 AOp or Air Blast . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.2 Ground Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.3 Flyrock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Blasting Effect Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.1 Prediction of AOp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.2 Prediction of Ground Vibration . . . . . . . . . . . . . . . . . . . . . . . . . 1.4.3 Prediction of Flyrock Distance . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Prediction Methods by Computational Techniques . . . . . . . . . . . . . . . 1.6 Blasting Solutions Enabled by the Blastiq™ Platform . . . . . . . . . . . . . 1.6.1 Blast Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6.2 Blast Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6.3 BlastIQ™ Advanced Technologies . . . . . . . . . . . . . . . . . . . . . . 1.7 Conclusion Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 1 2 4 4 5 5 6 6 7 10 11 13 14 14 14 14 15

2 Review of Empirical and Intelligent Techniques for Evaluating Rock Fragmentation Induced by Blasting . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Rock Fragmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3 Blastability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4 Fragmentation Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5 Background of ML Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.1 Artificial Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.2 Adaptive Neuro-Fuzzy Inference System (ANFIS) . . . . . . . . . 2.5.3 Support Vector Machine (SVM) . . . . . . . . . . . . . . . . . . . . . . . . . 2.5.4 Genetic Algorithm (GA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

21 21 22 22 25 25 28 28 29 29

vii

viii

Contents

2.6 Review of ML Models for Prediction of Rock Fragmentation . . . . . . 2.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.8 Conclusion and Future Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

30 31 35 36

3 Applications of AI and ML Techniques to Predict Backbreak and Flyrock Distance Resulting from Blasting . . . . . . . . . . . . . . . . . . . . . . 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Measurement of Flyrock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1 Flyrock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2 Backbreak . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Concepts of Some AI Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1 Artificial Neural Network (ANN) . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2 ANFIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.3 Support Vector Machine (SVM) . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.4 ELM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.5 PSO-ELM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 Backbreak Prediction Using AI Techniques . . . . . . . . . . . . . . . . . . . . . 3.5 Flyrock Prediction Using AI Techniques . . . . . . . . . . . . . . . . . . . . . . . . 3.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

41 41 42 42 43 44 44 44 45 45 46 46 47 53 54 54

4 Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Ground Vibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 AOp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Background of Common AI Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.1 Artificial Neural Network (ANN) . . . . . . . . . . . . . . . . . . . . . . . . 4.4.2 Support Vector Machine (SVM) . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.3 Fuzzy Interface System (FIS) . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 Ground Vibration Prediction Using AI Techniques . . . . . . . . . . . . . . . 4.6 AOp Prediction Using AI Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

61 61 62 63 64 64 65 65 66 66 72 73 74

About the Authors

Ramesh M. Bhatawdekar currently works as an Adjunct Professor in the Department of Mining Engineering, Indian Institute of Technology, Kharagpur, India. He also works as the Head of Training and Courses at Geotropik, Department of Civil Engineering, Faculty of Engineering, Universiti Teknologi Malaysia. He has Ph.D. degree in Civil—AI/ML application in Blasting, from Universiti Teknologi Malaysia. His area of research is drilling, rock mechanics, rock mass classification and blasting environmental issues, applying artificial intelligence, and optimization algorithms in geotechnics. He has published 25 papers in well-established ISI and Scopus Journals and presented 40 papers in national and international conferences. Dr. Danial Jahed Armaghani is currently working as a Senior Researcher in the Institute of Architecture and Construction, South Ural State University, Russia. In addition, he is a visiting fellow at the School of Civil and Environmental Engineering, University of Technology Sydney (UTS), Australia. He received his postdoc from Amirkabir University of Technology, Tehran, Iran, and his Ph.D. degree, in Civil Geotechnics, from Universiti Teknologi Malaysia, Malaysia. His areas of research are tunneling, rock mechanics, piling technology, blasting environmental issues, applying artificial intelligence, and optimization algorithms in different areas of civil engineering. He published more than 200 papers in well-established ISI and Scopus journals, national, and international conferences. He is also a recognized reviewer in the areas of rock mechanics and geotechnical engineering. Dr. Aydin Azizi holds a Ph.D. degree in Mechanical Engineering. Certified as an official instructor for the Siemens Mechatronic Certification Program (SMSCP), he currently serves as a Senior Lecturer at the Oxford Brookes University. His current research focuses on investigating and developing novel techniques to model, control, and optimize complex systems. His areas of expertise include Control & Automation, Artificial Intelligence, and Simulation Techniques. He is the recipient of the National Research Award of Oman for his AI-focused research, DELL EMC’s “Envision the Future” completion award in IoT for “Automated Irrigation System”, and ‘Exceptional Talent’ recognition by the British Royal Academy of Engineering. ix

Chapter 1

An Overview of Blasting Operations and Possible Techniques to Solve Environmental Issues of Blasting

1.1 Introduction Albert Nobel invented dynamite during 1867 replacing black powder commonly used for breaking rock. Usage of ammonium nitrate fuel oil (ANFO) started after learning from Oppau explosion in Germany during 1921 and disaster in Texas City during 1947 that explosion can occur with partially mixed/unmixed ammonium and fuel oil. Since 1970, usage of ANFO and later slurry explosives were introduced for blasting. Over the last three decades, usage of TNT-based explosives for blasting has drastically reduced. During the last century, tunnel boring machines, primary and secondary breakers, surface miners were introduced for breaking rock [1–4]. In spite of introduction of new technologies for breaking rock, blasting has remained most popular methodology for breaking rock for surface mining, tunnels, and civil engineering projects. Rock fragmentation, backbreak, and flyrock due to the blasting are interrelated. Bhandari [5] investigated variation of burden and spacing to have an impact on blast fragmentation. During the 1980s–1990s, researchers developed various empirical equations-based blasting design parameters for prediction rock fragmentation and flyrock. There were limitations on modeling methods and physics of blasting. Rock mass properties, explosives properties, blasting geometry, and the detonation timings were further considered for developing complex numerical models. Further input parameters for correlating to blast fragmentation were dynamic rock fracture based on gas expansion with explosives detonation, rock mass characterization, and blastability [6]. Thorne [7] conducted experimental damage model due to the blasting to access rock fragmentation and results were compared with Finite element Pronto Program. Bilgin [8] conducted series of experiment with single hole blasting to determine critical and optimum blasting parameters. The determined optimum and critical burdens were stated in terms of blasthole diameter and bench height. The relationships between burden and angle of breakage, throw, backbreak, and broken material’s volume were established as well. Findings indicated that the maximal volume © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 R. M. Bhatawdekar et al., Environmental Issues of Blasting, SpringerBriefs in Applied Sciences and Technology, https://doi.org/10.1007/978-981-16-8237-7_1

1

2

1 An Overview of Blasting Operations and Possible Techniques …

of broken ore was attained when the burden is approximately half of bench height. Researchers further developed numerical models based on blasted rock swell and muckpile formation, in situ block size distribution, explosives energy distribution [9]. Various researchers have carried out studies for improving blast fragmentation and thus improving downstream operation of crushing and mill adding overall efficiency and profitability [10–14]. Maerz et al. [15] developed technique of comparing blasted broken rock images and comparing with digital images of crushed rock samples. Liu and Kastsabni [16] developed a damaged model for blasting through experimental study on development of the cracks due to blasting. Following are the observations: (a) the rock material fail when applied stress is higher than its static strength; (b) if the rock material is exposed to a stress greater than its static strength, a specific time interval is required for the fracture to occur; (c) the dynamic fracture stress of rock material is greater than its static strength and approximately cube root dependent on the strain rate. With the advancement in power of computing, various researchers have developed computational techniques for prediction of geotechnical issues in mining and civil engineering projects [17]. Researchers have also applied computational techniques for prediction of fragmentation [18, 19]. Flyrock is the major environmental hazard due to the blasting [4, 20, 21]. Prediction of flyrock is done with several computational techniques [22, 23]. Ground vibration and air overpressure (AOp) prediction has been also done with various intelligent techniques [24–29]. In this study, a comprehensive review of the blasting operations including blast design and advance technologies is carried out. In addition, the environmental effects due to blasting such as AOp, ground vibration, and flyrock is studied based on previous literature. Based on this understanding, the feasible techniques to solve environmental issues of blasting will be discussed.

1.2 Blast Design The blast design parameters are based on geometrical dimension, rock mass properties, and site specifics experience of blasting [30]. All these parameters are controllable as they can be decided prior to start of drilling and charging of holes with explosives based on the face condition. During initial stage of mine design, drilling equipment is selected based on production capacity, loading equipment, and bench height [31]. Burden and spacing is correlated with hole diameter and have overall influence on the blast design [32]. With increase in hole diameter, explosive charge concentration per meter increases. Many researchers have found that hole diameter has direct impact on flyrock and ground vibration [33, 34]. Figure 1.1 provides details of blast design parameters for production blast such as burden (B), spacing (S), hole diameter (d), stemming length (ST), bench height (H), sub-drilling (Sd), and hole depth (HD).

1.2 Blast Design

3

Fig. 1.1 Blast design terminology

Table 1.1 Constants for drill hole diameter to burden

Name of researcher

Range of values for constant C1

Jimeno et al. [31]

25–40

Bhandari [32], Hagan [35]

20–35

Dick et al. [36]

20–40

Burden is perpendicular distance between blasting face and hole. Table 1.1 shows relationship between hole diameter and burden proposed by many researchers and represented by Eq. 1.1. B = C1 × d

(1.1)

where B is burden, C1 is constant dependent upon properties of rock mass and explosives, and d is diameter of drill hole. Burden is also expressed in terms of bench height and can be expressed with following Eq. 1.2: B = C2 × H

(1.2)

where H is bench height, C2 is constant which varies from 0.25 to 0.50 for satisfactory blasts [37]. Bench height is based on loading equipment and hole diameter. Fourier and Dohm [38] have defined bench height as vertical distance between toe and crest of bench. Some of the disadvantage of longer bench height are rock fall and flyrock. During blasting operation, when explosives are being converted into gases, stemming acts as

4

1 An Overview of Blasting Operations and Possible Techniques …

buffer to prevent blown out shots. Stemming length is expressed in terms of burden. If stemming length is inadequate, excessive fly rock and AOp is observed [39]. Hole depth is sum of length of subgrade drilling and bench height. Subgrade drilling is required to avoid toe formation. Subgrade drilling may vary from 10 to 20% of bench height. From various research studies, it is observed that instead of a single parameter like hole diameter or burden or spacing or bench height being considered independently, it is in fact the ratios of these parameters that contribute to blast performance. Ratios of burden to blasthole diameter, bench height to burden, spacing to burden, and stemming length to burden have been found to play an important role for blast performance [40]. Spacing is related to burden and can vary from 1 to 1.8 times [41]. Many researchers utilize burden to spacing ratio for evaluation of blast performance. Bench height to burden is called stiffness ratio. Various researchers have used stiffness ratio for the design of blast [24].

1.3 Environmental Effect Due to the Blasting Blasting is accompanied by a number of side effects that have a negative impact on the environment, such as ground vibration, flyrock, AOp, and backbreaks [33, 42]. These effects are more severe if they are carried out in close proximity to residential buildings, factories, offices or if they are improperly designed [43]. Proper designing of a blasting operation plays a crucial role in fragmenting rock for tunneling projects and opencast mines.

1.3.1 AOp or Air Blast It is the air pressure generated due to blast explosions. The shock waves are produced by one to several circumstances as following: a release of inadequately confined gases, release of energy directly from the surface, a shock from a large free face, pulse from stemming column during ejection of stemming, and gas release pulse due to escaping of gases through rock fractures. AOp from blasting comprises wide-ranging frequencies, some of which are detected as noise by people, while the low-frequency component (12

P2

Resistance of fracturing

UTS (Mpa)

1.5

1.5–3

P3

Sturdiness of rock

Density (ρ/(t/m3 ))

2.0

2.0–2.4

2.4–2.75 2.75–3.0

>3.0

P4

Elasticity of rock

E/Gpa

25

25–50

50–100

100–150

>150

P5

Resistance to dynamic loading

P-wave velocity, V p (km/s)

1.5

1.5–2.5

2.5–3.0

3.0–4.0

>4.0

P6

Hardness of rock

SHRN

15

15–30

30–40

40–50

>50

P7

Deformability Poisson ratio

>0.35

0.3–0.35 0.25–0.30 0.20–0.25 3.5

0.25–0.75 0.75–1.5

1.5–2.5

>2.5

1.5–2.5

>2.5

0.1–0.5

0.5–1.5

1.50–2.00 2.00–2.50 2.50–2.75 >2.75

0.35–0.55 0.55–0.75 0.75–0.9

40–60

60–75

75–90

>0.9

>90

0.05–0.15 0.15–0.25 0.25–0.50 >0.50 7.5–15

15–20

20–30

>30

1.6 Blasting Solutions Enabled by the Blastiq™ Platform Blast Information Management BlastIQ™ allows to operate drill and blast data and processes digitally, delivering a protected, centralized online platform for compilation and analysis of information and understandings across the entire blasting operation.

14

1 An Overview of Blasting Operations and Possible Techniques …

1.6.1 Blast Design Optimal blast designs shall be achieved using SHOTPlus™, an innovative 3D blast design and modelling software. The designs could be observed in spatial context, relative to neighboring blasts, and mine landscape. Integration with BlastIQ™ Platform, allows blast loading instructions and rules to be communicated to the field, ensuring the proper explosives are utilized in the right spot and initiated at the correct time, always.

1.6.2 Blast Control BlastIQ™ provides total control throughout the blasting operation including quality control management of blast design execution using enhanced visibility and control of bench operations. It aids in systematized access and execution of design as well as loading guidelines. BlastIQ™ enabled delivery systems accept blast designs, receive, and respond to updated blasthole conditions, load to instruction, and automatically communicate as-loaded data and presents insights and analytics in BlastIQ™, in real time.

1.6.3 BlastIQ™ Advanced Technologies The succeeding generation BlastIQ™ Platform provides a set of improved technologies, allowing enhanced control of blasting results and clearer decision-making. The technologies in the BlastIQ™ Platform as shown in Fig. 1.3 are developed to provide individual operational and economic value. In addition, the advantages are maximized when they are incorporated into a systematized process.

1.7 Conclusion Remarks • Blasting is an essential activity for breaking rock during excavation in any civil or mining engineering projects. • Favorable factors for blasting include desired well-fragmented rock, adequate throw, and appropriate muckpile shape. On the other hand, unfavorable factors due to the blasting include flyrock, ground vibration due to blasting, AOp, dust, and fumes. • During the last century, various empirical techniques were developed for prediction of blast performance based on blast design parameters.

1.7 Conclusion Remarks

15

Fig. 1.3 BlastIQ™ platform

• For prediction of flyrock distance, in addition to empirical equations, semiempirical trajectory physics-based models, and mathematical models shall be utilized. • Index of rock mass blastability depends upon strength, resistance to dynamic loading, resistance to fracturing, elasticity of rock, resistance to breaking, in situ block size, sturdiness of rock, fragility of rock mass, discontinuity plane’s strength, integrity of rock mass, hardness of rock, and deformability. • Fuzzy interface system based on 12 input parameters to predict rock mass blastability is developed. • Blast Information Management BlastIQ™ platform by Orica—one of the major explosives manufacturers provides digital solution on mobile, cloud platform for blast design, blast control for analysis, and maintaining record of each blast which can be used for future reference.

References 1. D.J. Armaghani, E.T. Mohamad, M.S. Narayanasamy, N. Narita, S. Yagiz, Development of hybrid intelligent models for predicting TBM penetration rate in hard rock condition. Tunn. Undergr. Sp. Technol. 63, 29–43 (2017) 2. A. Azizi, D. Jahed Armaghani, A comparative study of artificial intelligence techniques to estimate TBM performance in various weathering zones, in Applications of Artificial Intelligence in Tunnelling and Underground Space Technology, SpringerBriefs in Applied Sciences and Technology (Springer, Singapore, 2021), pp. 55–70. https://doi.org/10.1007/978-981-161034-9_4

16

1 An Overview of Blasting Operations and Possible Techniques …

3. A. Bagchi, R. Gupta, Surface blasting and its impact on environment, in Workshop on Environmental Management of Mining Operations (1990), pp. 262–279 4. E.T. Mohamad, C.S. Yi, B.R. Murlidhar, R. Saad, Effect of geological structure on flyrock prediction in construction blasting. Geotech. Geol. Eng. 36(4), 2217–2235 (2018) 5. S. Bhandari, On the role of stress waves and quasi-static gas pressure in rock fragmentation by blasting. Gasdyn. Explos. React. Syst. 365–383 (1980) 6. D. Grady, M. Kipp, Dynamic rock fragmentation, in Fracture Mechanics of Rock, ed. by B. Atkinson (Academic Press, London, 1987), pp. 429–475 7. B. Thorne, A Damage Model for Rock Fragmentation and Comparison of Calculations with Blasting Experiments in Granite (No. SAND-90–1389), Albuquerque (1990) 8. H. Bilgin, Single hole test blasting at an open pit mine in full scale: a case study. Int. J. Surf. Mining Reclam. Environ. 5(4), 191–194 (1991) 9. J. Latham, A. Munjiza, P. Lu, Rock fragmentation by blasting—a literature study of research in the 1980s and 1990s. Fragblast 3(3), 193–212 (1999) 10. R. Adam, B. Sidall, Blast manipulation or crusher modification, in Proceeding of the Mine-Mill Conference (1988) 11. B. Bulow, P. Smallbone, P. Walker, Blasting for reduced process plant costs at Argyle Diamond Mine, in Proceeding of the Mine-Mill Conference (1988) 12. S. Kanchibotla, S. Morrell, W. Valery, P. O’loughlin, Exploring the effect of blast design on SAG Mill throughput at KCGM, in Proceeding of the Mine-Mill Conference (1998), pp 153–158 13. S. Kanchibotla, W. Valery, S. Morrell, Modeling fines in blast fragmentation and its impact on crushing and grinding, in Explo’99: A Conference on Rock Breaking (1999), pp. 137–144 14. T. Kojovic, S. Kanchibotla, N. Poetschka, J. Chapman, The effect of blast design on the lump: fines ratio at Marandoo iron ore operations, in Proceeding of the Mine-Mill Conference (1998) 15. N. Maerz, J. Franklin, L. Rothenburg, D. Linncoursen, Measurement of rock fragmentation by digital photoanalysis, in 6th International Congress for Rock Mechanics, August 30–September 3 1987 (1987), pp. 687–692 16. L. Liu, P. Katsabanis, Development of a continuum damage model for blasting analysis. Int. J. Rock. Mech. Min. Sci. 34(2), 217–231 (1997) 17. X. Lu, M. Hasanipanah, K. Brindhadevi, H.B. Amnieh, S. Khalafi, ORELM: a novel machine learning approach for prediction of flyrock in mine blasting. Nat. Resour. Res. 29(2), 641–654 (2020) 18. B.R. Murlidhar, D.J. Armaghani, E.T. Mohamad, S. Changthan, Rock fragmentation prediction through a new hybrid model based on imperial competitive algorithm and neural network. Smart Constr. Res. 2(3), 1–12 (2018) 19. P.F. Asl, M. Monjezi, J.K. Hamidi, D.J. Armaghani, Optimization of flyrock and rock fragmentation in the Tajareh limestone mine using metaheuristics method of firefly algorithm. Eng. Comput. 34(2) (2018) 20. Raina et al., Flyrock in bench blasting: a comprehensive review. Bull. Eng. Geol. Env. (2014). https://doi.org/10.1007/s10064-014-0588-6 21. E.T. Mohamad, B.R. Murlidhar, D.J. Armaghani, R. Saad, C.S. Yi, Effect of geological structure and blasting practice in fly rock accident at Johor, Malaysia. J. Teknol. 78, 8–6 (2016) 22. J. Zhou et al., Use of intelligent methods to design effective pattern parameters of mine blasting to minimize flyrock distance. Nat Resour Res (2019). https://doi.org/10.1007/s11053-019-095 19-z 23. B.R. Murlidhar, D. Kumar, D. Jahed Armaghani, E.T. Mohamad, B. Roy, B.T. Pham, A novel intelligent ELM-BBO technique for predicting distance of mine blasting-induced flyrock. Nat. Resour. Res. (2020). https://doi.org/10.1007/s11053-020-09676-6 24. M. Monjezi, H. Khoshalan, A. Varjani, Prediction of flyrock and backbreak in open pit blasting operation: a neuro-genetic approach. Arab. J. Geosci. (2012) 25. B.R. Murlidhar, B.Y. Bejarbaneh, D.J. Armaghani, A.S. Mohammed, E.T. Mohamad, Application of tree-based predictive models to forecast air overpressure induced by mine blasting. Nat. Resour. Res. (2020). https://doi.org/10.1007/s11053-020-09770-9

References

17

26. E. Tonnizam Mohamad, D. Jahed Armaghani, M. Hasanipanah, B.R. Murlidhar, M.N.A. Alel, Estimation of air-overpressure produced by blasting operation through a neuro-genetic technique. Environ. Earth Sci. 75(2), 1–15 (2016) 27. R. Shirani Faradonbeh et al., Prediction of ground vibration due to quarry blasting based on gene expression programming: a new model for peak particle velocity prediction. Int. J. Environ. Sci. Technol. 13(6) (2016) 28. D.J. Armaghani, M. Hajihassani, E.T. Mohamad, A. Marto, S.A. Noorani, Blasting-induced flyrock and ground vibration prediction through an expert artificial neural network based on particle swarm optimization. Arab. J. Geosci. 7(12), 5383–5396 (2014) 29. Z. He, D.J. Armaghani, M. Masoumnezhad, M. Khandelwal, J. Zhou, B.R. Murlidhar, A combination of expert-based system and advanced decision-tree algorithms to predict air-overpressure resulting from quarry blasting. Nat. Resour. Res. 30(2), 1889–1903 (2021) 30. R. Kumar, D. Choudhury, K. Bhargava, Determination of blast-induced ground vibration equations for rocks using mechanical and geological properties. J. Rock Mech. Geotech. Eng. 8(3) (2016) 31. C. Jimeno, E. Jimeno, F. Carcedo, Drilling and Blasting of Rocks (A. A. Balkema, Rotterdam, 1995) 32. S. Bhandari, Engineering rock blasting operations. A. A. Balkema. 388, 388 (1997) 33. M. Hasanipanah, M. Monjezi, A. Shahnazar, D.J. Armaghani, A. Farazmand, Feasibility of indirect determination of blast induced ground vibration based on support vector machine. Measurement 75, 289–297 (2015) 34. D. Armaghani, M. Hasanipanah, E. Mohamad, A combination of the ICA-ANN model to predict air-overpressure resulting from blasting. Eng. Comput. 32, 155–171 (2016) 35. T. Hagan, The influence of controllable blast parameters on fragmentation and mining costs, in Proceedings of the 1st International Symposium on Rock Fragmentation by Blasting (1983), pp. 31–32 36. R. Dick, L. Fletcher, D. D’Andrea, Explosives and Blasting Procedures Manual (No. 8925) (US Department of the Interior, Bureau of Mines, Washington DC, 1983) 37. G. Adhikari, B. Rajan, H. Venkatesh, A. Thresraj, Blast damage assessment for underground structures, in Proceedings of the National Symposium on Emerging Mining and Ground Control Technologies (1994), pp. 247–255 38. G. Fourie, G. Dohm, Open pit planning and design, in SME Mining Engineering Handbook, ed. by Hartman, vol. 2 (SMME, Colorado, 1992), pp. 1274–1297 39. H. Cevizci, H. Ozkahraman, The effect of blast hole stemming length to rockpile fragmentation at limestone quarries. Int. J. Rock Mech. Min. Sci. 53, 32–35 (2012) 40. S. Sharma, P. Rai, Establishment of blasting design parameters influencing mean fragment size using state-of-art statistical tools and techniques. Measurement 96, 34–51 (2017) 41. G.R. Adhikari, Studies on flyrock at limestone quarries. Rock Mech. Rock Eng. 32(4), 291–301 (1999) 42. B.R. Murlidhar, D.J. Armaghani, E.T. Mohamad, Intelligence prediction of some selected environmental issues of blasting: a review. Open Constr. Build Technol. J. 14(1), 298–308 (2020) 43. S. Chen, Z. Zhang, J. Wu, Human comfort evaluation criteria for blast planning. Environ. Earth. Sci. 74(4), 2919–2923 (2015) 44. M. Hajihassani, D. Jahed Armaghani, H. Sohaei, E. Tonnizam Mohamad, A. Marto, Prediction of airblast-overpressure induced by blasting using a hybrid artificial neural network and particle swarm optimization. Appl. Acoust. 80, 57–67 (2014) 45. E.J. Walter, C.J. Konya, Surface Blast Design (Prentice Hall, Englewood Cliffs, 1990) 46. P.-A. Persson, R. Holmberg, J. Lee, Rock Blasting and Explosives Engineering (CRC Press, Boca Raton, 1993) 47. A. Shahnazar, H. Nikafshan Rad, M. Hasanipanah, M.M. Tahir, D. Jahed Armaghani, M. Ghoroqi, A new developed approach for the prediction of ground vibration using a hybrid PSO-optimized ANFIS-based model. Environ. Earth Sci. 76(15) (2017)

18

1 An Overview of Blasting Operations and Possible Techniques …

48. X.-N. Bui, P. Jaroonpattanapong, H. Nguyen, Q.-H. Tran, N.Q. Long, A novel hybrid model for predicting blast-induced ground vibration based on k-nearest neighbors and particle swarm optimization. Sci. Rep. 9(1), 1–14 (2019) 49. D. Singh, V. Sastry, Influence of structural discontinuity on rock fragmentation by blasting, in Proceedings of the 6th International Symposium on Intense Dynamic Loading and Its Effects, 3–7 June 1986 (1986) 50. H. Verkis, Flyrock: a continuing blast safety threat, in Proceeding of the Thirty-Seventh Annual Conference on Explosives and Blasting Technique, 6–9 February 2011 (2011), pp. 731–739 51. C. Kuzu, A. Fisne, S.G. Ercelebi, Operational and geological parameters in the assessing blast induced airblast-overpressure in quarries. Appl. Acoust. 70(3), 404–411 (2009) 52. D. Siskind, V. Stachura, M. Stagg, J. Kopp, Structure Response and Damage Produced by Airblast from Surface Mining, vol. 8485 (Department of the Interior, Bureau of Mines, Washington DC, US, 1980) 53. C. Wu, H. Hao, Modeling of simultaneous ground shock and airblast pressure on nearby structures from surface explosions. Int. J. Impact Eng. 31(6), 699–717 (2005) 54. R. Rodríguez, J. Toraño, M. Menéndez, Prediction of the airblast wave effects near a tunnel advanced by drilling and blasting. Tunn. Undergr. Sp. Technol. 22(3), 241–251 (2007) 55. P. Segarra, J.F. Domingo, L.M. López, J.A. Sanchidrián, M.F. Ortega, Prediction of near field overpressure from quarry blasting. Appl. Acoust. 71(12), 1169–1176 (2010) 56. W. Hustrulid, Blasting Principles for Open Pit Mining (Balkema, Rotterdam, 1999) 57. J. Heilig, Overpressure Restrictions: Review and Implication on Blast Design, a Report to the Hong Kong Construction Association (2006) 58. G. Adhikari, H. Venkatesh, A. Theresraj, R. Balachander, Measurement and analysis of air overpressure from blasting in surface mines. Visfotak Explos. Saf. Technol. Soc. J. 1(2), 21–26 (2007) 59. G. Adhikari, H. Venkatesh, A. Babu, A. Theresraj, Air overpressure produced by surface mines. Mintech Publ. 16(3 & 4), 9–13 (1995) 60. T. Ongen, D. Karakus, G. Konak, A.H. Onur, Assessment of blast-induced vibration using various estimation models. J. African Earth Sci. 145, 267–273 (2018) 61. M. Monjezi, M. Ahmadi, M. Sheikhan, A. Bahrami, A.R. Salimi, Predicting blast-induced ground vibration using various types of neural networks. Soil Dyn. Earthq. Eng. 30(11), 1233– 1236 (2010) 62. U. Langefors, B. Kihlström, The Modern Technique of Rock Blasting, vol. 405 (Wiley, New York, 1963) 63. B. Davies, I. Farmer, P. Attewell, Ground vibration from shallow sub-surface blasts. Engineer 217(5644), 553–559 (1964) 64. N. Ambraseys, A. Hendron, Dynamic Behaviour of Rock Masses (Wiley, London, 1968) 65. A. Ghosh, J. Daemen, A simple new blast vibration predictor (based on wave propagation laws), in The 24th US Symposium on Rock Mechanics (USRMS), June 1983 (1983) 66. CMRI, Vibration Standards (Dhanbad, 1993) 67. R. Kumar, D. Choudhury, K. Bhargava, Response of shallow foundation in rocks subjected to underground blast loading using FLAC3D. Disaster Adv. 7(2), 64–71 (2014) 68. P. Roy, Putting Ground Vibration Predictions into Practice, vol. 241, no. 2 (Colliery Guard., United Kingdom, 1993), pp. 63–70 69. C. Wu, Y. Lu, H. Hao, W. Lim, Y. Zhou, C. Seah, Characterisation of underground blast-induced ground motions from large-scale field tests. Shock Waves 13(3), 237–252 (2003) 70. H. Nicholls, C. Johnson, W. Duvall, Blasting Vibrations and Their Effects on Structures (No. 656-660) (1971) 71. N. Lundborg, N. Persson, A. Ladegaard-Pedersen, Keeping the lid on flyrock in open-pit blasting. Eng Min J 176, 95–100 (1975) 72. C. McKenzie, Flyrock range and fragment size prediction, in Proceedings of the 35th Annual Conference on Explosives and Blasting Technique, vol. 2 (2009) 73. E. Ghasemi, M. Sari, M. Ataei, Development of an empirical model for predicting the effects of controllable blasting parameters on flyrock distance in surface mines. Int. J. Rock Mech. Min. Sci. 52, 163–170 (2012)

References

19

74. R. Trivedi, T. Singh, A. Raina, Prediction of blast-induced flyrock in Indian limestone mines using neural networks. J. Rock Mech. Geotech. Eng. 6(5), 447–454 (2014) 75. S. Olofsson, Applied Explosives Technology for Construction and Mining (Applex Publisher, Arla, 1990) 76. A. Richards, A. Moore, Flyrock control-by chance or design, in Proceedings of the Annual Conference on Explosives and Blasting Technique, vol. 1 (2004), pp. 335–348 77. T. Little, Flyrock risk, in Proceedings of EXPLO Conference, 3–4 September 2007 (2007), pp. 35–43 78. J.P. Latham, P. Lu, Development of an assessment system for the blastability of rock masses. Int. J. Rock Mech. Min. Sci. 36(1), 41–55 (1999) 79. Y. Azimi, M. Osanloo, M. Aakbarpour-Shirazi, A.A. Bazzazi, Prediction of the blastability designation of rock masses using fuzzy sets. Int. J. Rock Mech. Min. Sci. 47(7), 1126–1140 (2010) 80. J.G. Xue, J. Zhou, X.Z. Shi, H.Y. Wang, H.Y. Hu, Assessment of classification for rock mass blastability based on entropy coefficient of attribute recognition model. J. Cent South Univ. Sci. Technol. 41(1), 251–256 (2010) 81. J. Zhou, X.-B. Li, Integrating unascertained measurement and information entropy theory to assess blastability of rock mass. J. Cent. South Univ. 19, 1953–1960 (2012)

Chapter 2

Review of Empirical and Intelligent Techniques for Evaluating Rock Fragmentation Induced by Blasting

2.1 Introduction Blasting for breaking hard rock is a well-known economical way for civil engineering and mining projects for more than 70 years. Desired good fragmentation is the most important result of any blasting operation which has a direct impact on efficiency and productivity of downstream operation such as loading, transport, crushing, and milling [1]. Several researchers developed prediction of rock fragmentation with empirical equations, which depend upon blast design. Rock mass properties were not considered as a part of blasting in the mentioned equations. Blastability is a unique property of rock mass which is resistance of rock mass in breaking rock and for obtaining desired fragmentation. Lilly [2] introduced concept of blastability index based on material properties, rock mass properties, and local geology. Ghosh [3] developed blastability index for coal bearing area. Further, indexed system for blastability was developed and updated by other researchers considering rock mass and material properties [4–8]. Blastability index was computed based on joint spacing [9–11]. Rock mass classification has been widely used for more than 100 years in numerous applications in rock engineering design such as rippability, slopes, tunnels, foundations, and excavatability. Rock mass classification which simplifies complexity of in situ rock mass is vital for understanding many professionals for working on the same project. With the advancement of technology, blast fragmentation is measured with image analysis as compared to conventional sieve analysis. During the last decade, various researchers have developed machine learning (ML) techniques for prediction blasting performance [12–22]. Additionally, these techniques have been utilized in the area of rock fragmentation induced by blasting [23, 24]. In this study, the used ML models in this area will be described with their advantages and disadvantages.

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 R. M. Bhatawdekar et al., Environmental Issues of Blasting, SpringerBriefs in Applied Sciences and Technology, https://doi.org/10.1007/978-981-16-8237-7_2

21

22

2 Review of Empirical and Intelligent Techniques …

2.2 Rock Fragmentation Rock mass is broken into smaller fragments during blasting operation [25].The blast design, rock mass characteristics, and rapid energy released from blasting influences the fragmentation [26, 27]. Average fragment size of total blasted rock muck pile is termed as mean fragment size (X 50 ). X max is the utmost fragment size, while X 80 is 80% of X max . Empirical equations are developed upon the hypothesis that rock mass characteristics are homogeneous. Yet, the aeromechanical properties vary throughout rock mass and geological parameters are biased in nature. Although quantifiable values are specified for properties of each rock type, they may differ in field. On the other hand, the controllable factors in blasting are maximum charge per delay, explosives, delay demining, loading density of explosives, and blast design. Table 2.1 shows the empirical equations for prediction of blast fragmentation. Blasting engineers conventionally tend to minimize the oversized (+500 mm) rocks during the blasting for improving loading, transportation, and minimizing cost. As per study by Osanloo and Hekmat [34], loading efficiency of shovel depends upon bucket fill factor, percentage oversize fraction, and distribution of fragment size. The oversized rock has adverse influence on the primary crusher. Several researchers [24, 35–38] have confirmed that in situ rock mass property is one of the key contributors to fragmentation and characterizing the blastability is of high significance to blasting operations. If similar blast geometry and equal amount of energy from the explosive material are extended on two rock masses that differ from each other, then they generate completely dissimilar extents of fragmentation. This property of the rock mass is called “blastability”. Size of fragment is known as known as a key feature, which is primarily dependent on the geomechanical characteristics of the host rock mass. Blasting refers to transforming from the state with an in situ block size distribution (ISBD) to the state with a blasted block size distribution (BBSD). Figure 2.1 shows transformation of IBSD-C into either BBSD-1 or BBSD-2.

2.3 Blastability Many researchers have reported that in situ rock properties play an important role in fragmentation and blastability of rock [2, 3, 29, 31]. Rock mass classification for the assessment of blastability in tropically weathered limestones of Sri Lanka, Thailand, and Cambodia was developed [39]. Still empirical blastability rock system is yet to be developed based on current scientific understanding incorporating all important parameters such as geomechanical properties of rock mass, in situ block size and input explosive energy [4]. Some of non-controllable factors affecting blast fragmentation from practical point of view are as follows:

2.3 Blastability

23

Table 2.1 Empirical equation for prediction of rock fragmentation Reference

Empirical equation  0.8 1 Kuznetsov [28] X m = A × QVoT × Q T6

1

Cunningham [29]

Rosin and Rammler [30]

X m = A × K −0.8 × Q e6 ×

Morin and Ficarazzo [32]

Rock mass characteristics and explosive strength accounted. The mean size correlated to the characteristic size of the Rosin–Rammler distribution. The uniformity coefficient is unknown 115 SANFO

19 30

Suitable to calculate mean fragmentation size for a given powder factor. Rock factor is based on rock mass description (massive, jointed, or friable), joint spacing, and density of rock, UCS and Young’s modulus

where: e K = Q Vo

−





X

R = e XC can be rewritten as following: Xc =

Cunningham [31]

Remarks

Xm

Rocks are not uniform in geomechanical properties. Various geological properties also vary within rock mass. This has impact on fragment size distribution, which may not be uniform for different fraction of fragment distribution, e.g. X 50 may not be correlated with X 20 or X m with heterogeneous rock mass

1

0.693 n

 n = 2.2 −

14B D

   1+ BS 0.5  1− 2

W B

 L  H

Input parameters for Monte Carlo simulation based on Kuz-Ram model are drilling accuracy, UCS, JPS, JD, elastic modulus, and dip direction of bench face

Uniformity index is found to be 0.75–1.5 with preference number of 1 Sufficient field information is needed for testing simulation (continued)

24

2 Review of Empirical and Intelligent Techniques …

Table 2.1 (continued) Reference

Empirical equation

Remarks

Gheibie et al. [33]

Modified Kuz-Ram model for single-hole bottom and column when two different types of explosives are used:      √ 1 14B W S n = 2.2 − 1− + D B 2 2B   0.5    BCL − CCL  L  0.1 +   L H

Rocks are not uniform and homogeneous in terms of physio-mechanical properties influencing blastability

X m mean fragment size (cm), UCS uniaxial compressive strength (MPa), A rock factor, V o blast volume (S × B × H), S spacing, B burden, H bench height (m), QT mass of explosive—energy equivalent of TNT explosive charge equivalent of each blast hole, K powder factor in kg of explosives per cum of blasted rock, Qe total explosives charge (kg), S ANFO relative strength of explosive with respect to ANFO, X diameter of fragment (cm), X c characteristic size (cm), n Ronin-Rammler exponent or uniformity index, e base of natural logarithms (2.7183), BCL bottom charge length (m), CCL column charge length (m), AT timing factor, C(A) correction factor for the rock factor, ns uniformity factor governed by the scatter ratio, C(n) correction factor for the uniformity index, D hole diameter (mm), W standard deviation in drilling (m), L total charge length (m), JPS joint plane spacing, JD joint density

Fig. 2.1 The concept of blastability: two different rock masses with the same IBSD but with different blastability are transformed to two different BBSD curves

• Frequency of occurrence of joints and discontinuities: Higher the frequency of joints, easier it would do the blasting; lower the frequency of joints or higher the block size, higher is the strength of the rock mass [40]. • Presence of Cavities: Cavities (also called Vughs) are formed in rock bodies due to the dissolution of parts of rock bodies by ground water. Sulfide ores, limestones, and some iron ore deposit. Large cavities have an adverse effect on the blast performance-non-uniform fragmentation, large size boulders.

2.3 Blastability

25

• Degree of weathering: Rock strength is reduced from fresh, slightly weathered, moderately weathered, highly weathered, and completely weathered rock. Desired fragmentation can be achieved with changing blast design parameters.

2.4 Fragmentation Measurement Although the main purpose of blasting is fragmentation, it is exceptionally challenging to determine degree of fragmentation. Sieving or screening is the technique to assess particle of fragment size distribution which is expensive and time consuming. Hence, the indirect techniques, i.e. image processing programs such as IPACS, GoldSize, FRAGSCAN, CIAS, PowerSieve, TUCIPS, WipFrag, Fragalyst, and SPLIT provide precise blast fragmentation distribution assessment. A few of the systems such as FragScan, Split Desktop, GoldSize, and WipFrag are popular for 2D image processing of size distribution assessment of the blasted rock blocks. Figure 2.2 shows the fragmentation analysis methods. In addition, Table 2.2 shows the summary of various prediction techniques on blast fragmentation. In this table, the used parameters, advantages and disadvantages of the proposed models are discussed.

2.5 Background of ML Models This section presents concepts of several ML models used in solving problem of rock fragmentation induced by blasting. These techniques were selected based on the frequent use by the researchers in the area of blasting and its environmental issues. In the following sub-sections, the background of these models is described.

Fig. 2.2 Prediction of fragmentation analysis methods

Methods adopted

Sobel edge detection technique

Fragalyst 3.0 software

• WipFrag image analysis software • Edge detection variables (EDV)

Reference

Sereshki et al. [41]

Raina et al. [42]

Nanda and Pal [43]

Blast design parameters

Stripping ratio, depth of mine, etc

–

Parameters

Table 2.2 Summary of different prediction techniques on blast fragmentation Disadvantage

WipFrag provides immediate PSD analysis of collected digital images after blast, close up sample of stockpile, laboratory sample, and UAV or drone photographs. cost-effective and precise with auto-scaling capability

Detects the edges between fragments automatically after threshold parameters are properly selected

(continued)

As the digital images taken for analysis does not disclose the fragmentation state behind muckpile surface, the individual analysis of the rock pile samples is considered imprecise

Sometimes detects one rock fragment as two or more rocks. Not able to detect edges perfectly. Unable to detect very small rock fragment

Detects the edges between Unable to detect fine particles fragments through enhancement of rock image in a very short time for rock size distribution and identification of connected areas between the rocks

Advantage

26 2 Review of Empirical and Intelligent Techniques …

Split Desktop consisting of five Blast design parameters, phases: emulite use per rock ton • Scaling image • Segmentation of rock fragments in each image • Issuing permission for editing of desired rock fragments to assure accuracy of results • Analysis of marked rock fragments • Displaying size distribution results in diagrams

Elahi and Hosseini [44]

Parameters

Methods adopted

Reference

Table 2.2 (continued) Up to three scales, bodies can be applied on the image. The image’s resolution is modifiable and use several extensions for digital images including bmp, tif, and jpg. Zones of fine materials and areas which do not require analysis are identified. Possibility of automatic segmentation of fragments and magnifying the image

Advantage

Sample needs to be collected from a reference surface which is time-consuming. High-quality images are must for accurate detection of fragments. In addition, derivation precise 3D information from 2D images is challenging

Disadvantage

2.5 Background of ML Models 27

28

2 Review of Empirical and Intelligent Techniques …

2.5.1 Artificial Neural Network Artificial neural networks (ANNs) are mainly a self-adaptive algorithm and it can capture information from available examples and provide ability to get functional relationships among data, even if relationships are undescribed and tough to determine. ANN is showed as generalization of mathematical simulations of human nervous systems consisting of synapses and neurons in which neurons are one of the core processing entities of neural networks. In simple mathematical model of the ANN, the characteristic of nonlinearity is shown using neurons and characterized by a transfer function, whereas the input signal associated with weight symbolized by the effect of synapses. Consequently, neuron impulse is calculated using the weighted sum of the signal at input layer, which is transformed by the transfer function. Artificial neuron’s learning ability can be achieved by fine-tuning the weights as per the selected learning technique. ANN is consisted of three layers, namely hidden layer, output layer, and input layer. In ANN, input is given at input layer in the form of patterns. Neurons at input layer are in connection with the neurons that are present at hidden layer where real processing is performed by using weighted connections. The neurons at hidden layer are in connection with the neurons present at output layer that generates final output. One of the highly efficient models of ANN is multilayer perceptron (MLP) that can be trained using back propagation (BP) algorithm [39, 45, 46]. BP utilizes mean square error (MSE) and gradient descent methods in order to adjust the weights and biases of the MLP. It is most suitable for solving problems that are complex, highly nonlinear, having many and different variables.

2.5.2 Adaptive Neuro-Fuzzy Inference System (ANFIS) The adaptive neuro-fuzzy inference system (ANFIS) module replicates a fuzzy inference system (FIS). So, ANFIS may be defined as an intelligent neuro-fuzzy process mostly applicable for controlling and modeling different uncertain systems. The membership function parameters in ANFIS are basically tuned using back propagation algorithm or by permutation with a least square-type method. It is basically a multi-layered feed forward neural network and is consist of nodes connected by direct link with no weights associated. These direct links generate a single node output by performing a static node parameterized function with modifiable parameters on its incoming signals. These parameters describe the form of the membership functions and enable it to learn from the modeling data. As it is optimized in such a way that the actual output can be minimized during training session by minimizing the error between actual and target output. Merits of ANFIS are: • ANFIS model may apply learning procedure to calculate the optimum value of the equivalent fuzzy inference system. • It may provide a suitable neural network solution for different problems such as function approximation problem.

2.5 Background of ML Models

29

• ANFIS is mainly based on clustering of training data of numerical dataset. • It is effectively useful in many scenarios, especially in case of rule-based method control, classification problems, pattern recognition tasks and many more. Meanwhile, the demerits of ANFIS are: • It is mainly designed on input and output of data therefore size of the data is quite crucial. So, in case of very less data availability, generation of dataset is quite expensive.

2.5.3 Support Vector Machine (SVM) During 1992–97, Vladimir Vapnik with his other colleagues developed Support Vector Machine (SVM) models at AT & T Bell Laboratories [47]. It is one of the models of supervised learning that have associated learning algorithms in order to analyze data for regression or classification [48]. In case of classification, first data is plotted as a pint in n-dimensional space (n denotes number of features), where the coordinate of the point is the value of the features. Then a hyper-plane is drawn in such a way that it differentiates the two classes. The advantages of the SVM technique are as follows: • SVMs deliver a decent out-of-sample generalization. Therefore, by selecting a suitable generalization score, they may provide a good solution, even when the training data may contain some bias. • SVMs provide a distinctive solution, as the problem of optimality is convex. It is one of the key benefits of SVM compared to neural networks. As neural network has various solutions related with local minima and because of that it may not be robust over different datasets. SVM may provide better solution in case data that contain non-regularity and may be useful tool for insolvency analysis, basically in case of data having unknown distribution or irregular distribution.

2.5.4 Genetic Algorithm (GA) Genetic Algorithm (GA) is basically a population-based stochastic technique that belongs to family of computational models. It was developed by getting inspiration form the Darwinian Theory of Evolution or by the manner living organisms developed into more successful organisms in nature [49]. The main operation in GA are mainly crossover, mutation, and selection [50]. GA is simulation of the existence of adequate or fitter individual and its genes in which every parameter corresponds a gene and each solution represents a solution. The process on GA take place by modelling evolution that starts with an initial set of hypotheses or results or solutions and

30

2 Review of Empirical and Intelligent Techniques …

creating consecutive “generations” of hypotheses. In case of selection process, in order to improve pitiable solutions all the solutions were selected randomly by using a selection process, for example, roulette wheel selection. This random operation is basically to select the best solutions among all the solution it is because fitness is proportional with the probability. The probability of choosing poor solutions is also there so, in case some good solutions were stuck with other local solution, that may also be obtainable with other solutions. After that operation named as crossover took place in which, chromosomes from the parents exchange randomly. Consequently, the offspring/child shows some characters of the mother chromosomes and some traits of the father chromosomes. In order to change some traits of the offspring, a process named mutation is performed. GA is basically used to optimize different function, instead of that the range of problems where GA is useful are quite wide ranging as it converts a potential solution to a specific problem for example data structure and preserve critical information by applying recombination operators on these structures.

2.6 Review of ML Models for Prediction of Rock Fragmentation ANN technique was introduced by many researchers for prediction of fragmentation [35, 51–53]. ANN resembles network of neurons of human brain. BP method with at least 3 layers was found to be efficient for learning in multi- layer neural networks. BP algorithm consists of forward pass and backward pass. Weight adjustment is done to minimize error. In order to solve problem related to mapping of input–output, feed forward-back propagation neural network was often used. Input dataset is divided into training and testing data and measured and predicted values are compared. FIS technique for prediction of fragmentation was applied by Shams et al. [24] and Monjezi et al. [54] with coefficient of determination (R2 ) of 0.922 and 0.96, respectively. Zadeh [55] was pioneer in introduction of the fuzzy set theory. There is always challenge for solving complex mathematical problems based on either biased or insufficient and unreliable, information. FIS is found sturdy and powerful system connecting three main components database, rule base, and reasoning mechanism through fuzzification connected with membership function. A total of 300 rules were developed based on if then rules, experience of experts, mathematical data, input and output data [54]. SVM-based algorithms have been developed by various researchers for prediction of fragmentation [56–58]. R2 values varied from 0.89 to 0.945. Hasanipanah et al. [59] reported that SVM model showed better results as compared to ANFIS and nonlinear models. Esmaeili et al. [56] mentioned a better performance for SVM as compared to ANFIS (R2 = 0.89, 0.63, respectively). Fang et al. [57] observed SVM model superior to FA-ANN, FA-ANFIS, GPR, and KNN models (R2 = 0.972, 0.967, 0.968, 0.940, 0.963, respectively). Zhang e al. [58] reported that SVM is better than KNN model (R2 = 0.911, 0.892, respectively). Shi et al. [60] showed

2.6 Review of ML Models for Prediction of Rock Fragmentation

31

that SVM model is superior as compared to Kuznetsov, Multivariate Regression Analysis (MVRA) and ANN (R2 = 0.962, 0.614, 0.815 and 0.941, respectively). PSO has many advantages such as understandable easily and user friendly. PSO has fewer parameters and fast in the processing. PSO is valuable and powerful optimization technique to preserve diversification of the swarm. Huang et al. [61] utilized PSO for prediction of fragmentation (R2 = 0.962). ANFIS merge both ANN and FIS. Thus, ANFIS is efficient and competent and capable of managing framework for solving complicated and nonlinear problems. ANFIS model was deployed by several researchers for prediction of fragmentation [26, 62] with R2 in the range of 0.61–0.98. PSO-ANFIS model has been used by many researchers [59, 62] for prediction of fragmentation. Impressive performance is shown by FA algorithm in solving optimization problems as this algorithm belongs swarm intelligence family. FA algorithm is practiced to solve problems with continuous and discrete function. Thus, recently, FA-ANFIS is used recently for prediction of fragmentation by different researchers [57, 63]. Table 2.3 presents some of the important studies in the area of rock fragmentation using ML techniques. In this table, rock type, number of data, input variables, model used, and R2 for different studies are presented. It is shown that there is a wide range for accuracy of the developed models (R2 in the range of 0.6–0.99). In addition, a wide range of variables such as blasting parameters, rock mass properties, and rock material properties has been used by different authors.

2.7 Discussion Statistical methods showed poor efficiency in prediction of fragmentation due to linearity hypothesis [24, 52]. Controllable parameters based on blast design are dependent variables. On the other hand, properties related to rock mass are uncertain and hence considered independent variables. Multiple regression analysis can be carried out to find out relationships with various dependent and indecent variables with prediction of output or fragmentation. Various researchers have applied MVRA Table 2.3 Estimation of rock fragmentation due to blasting using ML techniques Models used

R2

Reference

Rock type

Number of datasets

Input parameters

Bahrami et al. [52]

Iron ore

220

d, B, S, HD, ST, ANN PF, BI, SpD, C, SMR

0.97

Monjezi et al. [54]

Iron ore

–

B, S, PF, SpD, ST, Rd

0.96

FIS

0.80 (continued)

32

2 Review of Empirical and Intelligent Techniques …

Table 2.3 (continued) Reference

Rock type

Number of datasets

Input parameters

Models used

R2

Monjezi et al. [51]

Copper ore

–

B/S, d, ST, C, PF, PLI, DeR, HDe , NR

ANN

0.818 0.985

Hasanipanah et al. [25]

Copper ore

52

C, PF, S/B, ST/ B, H/B, NR , Hi , d, B/d

RES

0.866

Shams et al. [24]

Copper ore

185

B, S, d, SHRN, JD , PF, ST

FIS

0.922

MRA

0.738

Asl et al. [35]

Limestone mine

385

B, S, Sd, HD, ST, GSI, PF, Cde

ANN

0.94

Hasanipanah et al. [59]

Copper mine

58

B, S, ST, PF, C

PSO-ANFIS

0.938

SVM

0.90

ANFIS

0.88

Kulatilake et al. [53]

Copper mine

91

S, B, HD, d, ST, ANN PF, XB , E

–

Esmaeili et al. [56]

Iron ore

NaN

DeNR , PF, ST, d, BI, S/B

SVR

0.89

ANFIS

0.83 0.95

Murlidhar et al. [64]

Limestone

102

B, S, B/d, H/B, ST/B

ANN ANN-ICA

0.94

Sayadi et al. [65]

NaN

103

B, S, HD, SpD

ANN

0.85

Trivedi et al. [66]

NaN

NaN

B, S, ST, PF, UCS, RQD

ANN

0.983

Rad et al. [67]

Iron ore

90

B/S, H/B, SD, LS-SVM ST, Cde, Rd, PF SVR

0.969

B, S, ST, RMR, C, PF

0.985

Huang et al. [61]

NaN

75

CSO PSO

0.945 0.962

Rosales-Huamani et al. [27]

Iron ore

–

B, S, ST, Q, Rd, ANN GOU, PF

0.82

Fang et al. [57]

Limestone

136

B, S, ST, C, PF, H

FA-BGAM

0.980

FA-ANN

0.967

FA-ANFIS

0.968

SVM

0.972

GPR

0.940

KNN

0.963 (continued)

2.7 Discussion

33

Table 2.3 (continued) Reference

Rock type

Number of datasets

Input parameters

Models used

R2

Ghaeini et al. [68]

Copper

36

UCS,JP, RQD, JS, Rd, RF, SC, B, ST, JPO, S/d

MI

0.81

RES

0.75

Mehrdanesh et al. [69]

–

432

H, d, S, B, ST, CART PF, XB , RQD, ANN Vp , UTS, UCS, SHRN, υ, Ch , , Vp, BT, PLI, Rd

0.453

0.897

0.986

Sayevand et al. [70] –

80

B, S, ST, PF, C, RMR

ANN ICA

0.947

Vergara et al. [62]

92

B, S/B, Sd, PF

ANFIS

0.612

ANFIS-PSO

0.851

Zhang e al. [58]

Xie et al. [38]

Zhou et al. [26]

– Limestone

Limestone

Iron ore

136

136

88

B, S, ST, H, PF, C

B, S, ST, H, d, SC

B, S, ST, PF, MC, RMR

ACO-BRT

0.962

PSO-ANFIS

0.929

FA-ANFIS

0.948

FA-ANN

0.952

SVM

0.911

KNN

0.892

PCR

0.926

GP

0.94

FA-GBM

0.996

FA-SVM

0.979

FA-ANN

0.979

FA-GP

0.940

ANFIS-GA

0.989

ANFIS-FA

0.981

ANFIS

0.986

SVR

0.924

ANN

0.948

ANFIS-FA

0.98

Mojtahedi et al. [63]

Shur river dam

72

B, S, ST, PF, C

Mutinda et al. [71]

Limestone

Bisil-6 Simba-6

B, S, ST, H, HD, KCO d, PF, drd , RMR, KCO HF, JPS, RDI, qh , inS, BHP

0.98 0.99

(continued)

34

2 Review of Empirical and Intelligent Techniques …

Table 2.3 (continued) Reference Shi et al. [60]

Rock type

Number of datasets

Input parameters

Models used

102

S/B, H/B, B/d, Kuznetsov ST/B, PF, XB , E MVRA ANN

Shi et al. [72]

20 (Data set A-10, B-10)

J, f, B, Tde , VOD, SC, S

R2 0.614 0.815 0.941

SVR

0.962

MVRA

0.981

ANN

0.729

GA-ANN

0.992

GA-ANN

0.996

B burden, S spacing, d hole diameter (mm), HD hole length (m), H bench height (m), ST stemming length (m), Sd subdrilling, H i hole inclination, B/S burden to spacing ratio, SF stiffness factor (H/B), B/d burden to hole diameter ratio, S/d spacing to hole diameter ratio, SpD specific drilling (m/m3 ), V R volume of rock (m3 ), Q total explosives charge (kg), qh explosive weight (kg/hole), PF powder factor (g/t), SC specific charge (kg/m3 ), C maximum charge per delay (kg), C de charge per delay (kg/ms), TC de total charge per delay (kg/ms), DeR delay between rows (ms), H De holes per delay, DeN R ratio of total delays per number of rows, RWSE relative weight strength of explosives, VOD detonation velocity (m/s), T De delay time interval (MS), N R number of blasting rows, BHP blast hole pattern, inS initiation system Rd rock density (g/cc), RDI rock density influence, BI blastability index (%), RQD rock quality designation (%), GSI geological strength index (%), RMR rock mass rating, SMR slope mass rating, SHRN Schmidt hammer rebound number, PLI point load index (Mpa), UCS uniaxial compressive strength (Mpa), UTS uniaxial tensile strength (Mpa), E elastic modulus (GPa), XB mean block size (m), J rock joints, J n joint set number, J D joint density, JP joint persistency, JPO joint plane orientation, JS joint spacing, JPS joint plane spacing, BT brittleness, υ Poison ratio, C h cohesion, Φ friction angle, V p P wave velocity (m/s), RF rock factor, f Protogyakonov’s coefficient, GOU geotechnical ore units, dr d drilling deviation, HF hardness factor ACO ant colony optimization, ANN artificial neural network, ANFIS adaptive neuro-fuzzy interface system, BGAM boosted generalized additive model, BRT boosted regression tree, CART classification and regression tree, CSO cat swarm optimization, FA firefly algorithm, FIS fuzzy interface system, GA genetic algorithm, GP Gaussian process, GPR Gaussian process regression, ICA imperialism competitive algorithm, KCO Kuznetsov-Cunningham-Ouchterlony, LS least square, KNN k-nearest neighbors, MI mutual information, MRA multiple regression analysis, MVRA multivariate regression analysis, NLMR nonlinear multiple regression, PCR principal component regression, PSO particle swarm optimization, RES rock engineering system, SVM support verctor machine, SVR support vector regression

for prediction of fragmentation. However, there is a limitation in terms of accuracy. Hence, during last decade, better accuracy in the prediction of fragmentation is possible by introduction of ML techniques. Dependent variables are based on blast design parameters, powder factor, maximum charge per delay, or delay interval. Very limited researchers have predicted fragmentation based on only controllable parameters [26, 63, 64]. Fragmentation directly depends upon optimum burden so that desired fragmentation is obtained. Burden plays important role in fragmentation in front row as well burden between rows. Optimum stemming length provides good fragmentation. Hole diameter plays

2.7 Discussion

35

crucial role in fragmentation due to crushing zone around hole and dependency of other blast design parameters Powder factor or specific charge represents total quantum of explosive energy in individual blast and directly related to fragment size. Maximum charge per delay provides maximum release of explosive energy for a given instant. Number of rows per delay, charge per delay are some interrelated parameters having an impact on the fragmentation. Bench height has inverse relationship with fragment size. The recent trend by researchers was to use different ratio of blast design parameters instead of single parameter. Stiffness ratio (H/B), and ratios S/B, H/B, B/d, ST/B, were used by different scholars for prediction of blast fragmentation to minimize number of input parameters (Table 2.3). Most of the researchers have incorporated at least one parameter based on rock mass properties for prediction of fragmentation. Mehrdanesh et al. [69] have incorporated 19 parameters (controllable and independent variables). Practically, it may not be possible to collect all parameters for individual blast. For selecting rock mass parameters, various rock mass classifications for blastablity studies have been carried out which can be introduced at the site. Parameters having maximum variation can be identified and can be introduced for prediction of fragmentation. Different researchers [27, 67] introduced rock density (Rd) as one of the input parameters for rock breakage. In situ block size has a direct correlation with fragmentation produced. Mean block size (X B ) was introduced by some scholars [53, 60, 69]. GSI, RMR, RQD, joint properties have a direct impact on the fragmentation. SHRN, PLI, UCS, UTS, E are properties related to rock strength/material and based on blast design, powder factor, fragmentation was correlated by some of the researchers. In general, ML techniques with their abilities in solving nonlinear and complex problems are a good alternative for statistical and empirical approaches where they have several shortcomings. These techniques can be performed before blasting events to have a close values to our target which is rock fragmentation in this investigation.

2.8 Conclusion and Future Perspective 1.

2.

3.

Prediction of Blast fragmentation is crucial for efficiency and cost of various operations in mining or civil engineering projects during excavation. As compared to statistical or MVRA models, various AI/ML techniques have been developed which have better accuracy. MVRA can be used to benchmark with other models. Hybrid AI/ML models perform better as compared to single models such as ANN, PSO, GA, FA. SVM performed best among single model. ANFIS model was found better as compared ANN or PSO. Further PSO-ANFIS, FAANFIS have been utilized by many researchers. SVM-based hybrid models also performed better as compared to other hybrid models. Various input parameters used by many researches and established hybrid models can be used to predict fragmentation. Further research needs to be done

36

4.

2 Review of Empirical and Intelligent Techniques …

by developing larger database and application of hybrid models in different sites to compare results in consistency of models. Theory-based or physics-based ML is a new direction of these techniques, which can be broadly used in the other fields of engineering like civil and mining. In fact, incorporation of the theory-based physics-based ML models is required to ensure the better generalization of the model and prediction. They will give a better view and understanding to researchers and designers who are not familiar with ML techniques or do not have knowledge of computer science. The mentioned techniques can be applied in the area of rock fragmentation induced by blasting as a new corridor by mining and civil researchers.

References 1. S. Kanchibotla, S. Morrell, W. Valery, P. O’loughlin, Exploring the effect of blast design on SAG Mill throughput at KCGM, in Proceeding of the Mine-Mill Conference (1998), pp. 153–158 2. P.A. Lilly, Empirical method of assessing rock mass blastability, in Symposium Series— Australasian Institute of Mining Metallurgy (January 1986) pp. 89–92 3. A.K. Ghose, Design of drilling and blasting subsystems—a rock mass classification approach, in Mine Planning and Equipment Selection (1988) 4. J.P. Latham, P. Lu, Development of an assessment system for the blastability of rock masses. Int. J. Rock Mech. Min. Sci. 36(1), 41–55 (1999) 5. Y. Azimi, M. Osanloo, M. Aakbarpour-Shirazi, A.A. Bazzazi, Prediction of the blastability designation of rock masses using fuzzy sets. Int. J. Rock Mech. Min. Sci. 47(7), 1126–1140 (2010) 6. D.J. Armaghani, S. Yagiz, E.T. Mohamad, J. Zhou, Prediction of TBM performance in fresh through weathered granite using empirical and statistical approaches. Tunn. Undergr. Sp. Technol. 118, 104183 (2021) 7. D. Jahed Armaghani, A. Azizi, D. Jahed Armaghani, A. Azizi, Developing statistical models for solving tunnel boring machine performance problem. Appl. Artif. Intell. Tunn. Undergr. Sp. Technol. 33–53 (2021) 8. A. Azizi, D. Jahed Armaghani, Empirical, statistical, and intelligent techniques for TBM performance prediction, in Applications of Artificial Intelligence in Tunnelling and Underground Space Technology. SpringerBriefs in Applied Sciences and Technology (Springer, Singapore, 2021), pp. 17–32. https://doi.org/10.1007/978-981-16-1034-9_2 9. M. Chatziangelou, B. Christaras, Blastability index on poor quality rock mass. Int. J. Civ. Eng. 2(5), 9–16 (2013) 10. B. Christaras, M. Chatziangelou, Blastability Quality System (BQS) for using it, in bedrock excavation. Struct. Eng. Mech. 51(5), 823–845 (2014) 11. M. Chatziangelou, B. Christaras, A geological classification of rock mass quality and blast ability for widely spaced formations. J. Geol. Resour. Eng. 4, 160–174 (2016) 12. M. Koopialipoor, B.R. Murlidhar, A. Hedayat, D.J. Armaghani, B. Gordan, E.T. Mohamad, The use of new intelligent techniques in designing retaining walls. Eng. Comput. (2019). https:// doi.org/10.1007/s00366-018-00700-1 13. R. Shirani Faradonbeh et al., Prediction of ground vibration due to quarry blasting based on gene expression programming: a new model for peak particle velocity prediction. Int. J. Environ. Sci. Technol. 13(6) (2016) 14. A. Azizi, D. Jahed Armaghani, A comparative study of artificial intelligence techniques to estimate TBM performance in various weathering zones, in Applications of Artificial Intelligence in Tunnelling and Underground Space Technology. SpringerBriefs in Applied Sciences

References

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28. 29.

30. 31.

32.

37

and Technology (Springer, Singapore, 2021), pp. 55–70. https://doi.org/10.1007/978-981-161034-9_4 E. Tonnizam Mohamad, D. Jahed Armaghani, M. Hasanipanah, B.R. Murlidhar, M.N.A. Alel, Estimation of air-overpressure produced by blasting operation through a neuro-genetic technique. Environ. Earth Sci. 75(2), 1–15 (2016) E.T. Mohamad, M. Koopialipoor, B.R. Murlidhar, A. Rashiddel, A. Hedayat, D.J. Armaghani, A new hybrid method for predicting ripping production in different weathering zones through in-situ tests. Measurement (2019). https://doi.org/10.1016/j.measurement.2019.07.054 E.T. Mohamad, D. Li, B.R. Murlidhar, D.J. Armaghani, K.A. Kassim, I. Komoo, The effects of ABC, ICA, and PSO optimization techniques on prediction of ripping production. Eng. Comput. (2019). https://doi.org/10.1007/s00366-019-00770-9 B.R. Murlidhar, D.J. Armaghani, E.T. Mohamad, Intelligence prediction of some selected environmental issues of blasting: a review. Open Constr. Build. Technol. J. 14(1), 298–308 (2020) B.R. Murlidhar, D. Kumar, D. Jahed Armaghani, E.T. Mohamad, B. Roy, B.T. Pham, A novel intelligent ELM-BBO technique for predicting distance of mine blasting-induced flyrock. Nat. Resour. Res. (2020). https://doi.org/10.1007/s11053-020-09676-6 C. Yu et al., Optimal ELM–Harris Hawks optimization and ELM–grasshopper optimization models to forecast peak particle velocity resulting from mine blasting. Nat. Resour. Res. (2021). https://doi.org/10.1007/s11053-021-09826-4 D.J. Armaghani, M. Hajihassani, E.T. Mohamad, A. Marto, S.A. Noorani, Blasting-induced flyrock and ground vibration prediction through an expert artificial neural network based on particle swarm optimization. Arab. J. Geosci. 7(12), 5383–5396 (2014) D.J. Armaghani, E.T. Mohamad, M.S. Narayanasamy, N. Narita, S. Yagiz, Development of hybrid intelligent models for predicting TBM penetration rate in hard rock condition. Tunn. Undergr. Sp. Technol. 63, 29–43 (2017) B.R. Murlidhar, D.J. Armaghani, E.T. Mohamad, S. Changthan, Rock fragmentation prediction through a new hybrid model based on imperial competitive algorithm and neural network. Smart Constr. Res. 2(3), 1–12 (2018) S. Shams, M. Monjezi, V.J. Majd, D.J. Armaghani, Application of fuzzy inference system for prediction of rock fragmentation induced by blasting. Arab. J. Geosci. 8(12), 10819–10832 (2015) M. Hasanipanah, D. Jahed Armaghani, M. Monjezi, S. Shams, Risk assessment and prediction of rock fragmentation produced by blasting operation: a rock engineering system. Environ. Earth Sci. 75(9) (2016) J. Zhou, C. Li, C. Arslan, M. Hasanipanah, H. Amnieh, Performance evaluation of hybrid FFA-ANFIS and GA-ANFIS models to predict particle size distribution of a muck-pile after blasting. Eng. Comput. 37(1), 265–274 (2019) J. Rosales-Huamani, R. Perez-Alvarado, U. Rojas-Villanueva, J. Castillo-Sequera, Design of a predictive model of rock breakage by blasting using artificial neural networks. Symmetry (Basel) 12(9), 1405 (2020) V. Kuznetsov, The mean diameter of the fragments formed by blasting rock. Sov. Min. Sci. 9(2), 144e8 (1973) C. Cunningham, The Kuz-Ram model for production of fragmentation from blasting, in Proceedings of the 1st International Symposium on Rock Fragmentation by Blasting (1983), p. 439e53 P. Rosin, E. Rammler, Laws governing the fineness of powdered coal. J. Inst. Fuel 7, 29–36 (1933) C. Cunningham, Fragmentation estimations and the Kuz-Ram model-four years on, in Proceedings of Second International Symposium on Rock Fragmentation by Blasting (1987), pp. 475–487 M. Morin, F. Ficarazzo, Monte Carlo simulation as a tool to predict blasting fragmentation based on the Kuz-Ram model. Comput. Geosci. 32(3), 352–359 (2006)

38

2 Review of Empirical and Intelligent Techniques …

33. S. Gheibie, H. Aghababaei, S. Hoseinie, Y. Pourrahimian, Modified Kuz—Ram fragmentation model and its use at the Sungun Copper Mine. Int. J. Rock Mech. Min. Sci. 46(6), 967–973 (2009) 34. M. Osanloo, A. Hekmat, Prediction of shovel productivity in the Gol-e-Gohar iron mine. J. Min. Sci. 41(2), 177–184 (2005) 35. P.F. Asl, M. Monjezi, J.K. Hamidi, D.J. Armaghani, Optimization of flyrock and rock fragmentation in the Tajareh limestone mine using metaheuristics method of firefly algorithm. Eng. Comput. 34(2) (2018) 36. O. Akyildiz, T. Hudaverdi, ANFIS modelling for blast fragmentation and blast-induced vibrations considering stiffness ratio. Arab. J. Geosci. 13(21), 1–16 (2020) 37. H. Han, D.J. Armaghani, R. Tarinejad, J. Zhou, M.M. Tahir, Random forest and Bayesian network techniques for probabilistic prediction of flyrock induced by blasting in quarry sites. Nat. Resour. Res. 29, 655–667 (2020). https://doi.org/10.1007/s11053-019-0961138. C. Xie, H. Nguyen, X. Bui, Y. Choi, J. Zhou, T. Nguyen-Trang, Predicting rock size distribution in mine blasting using various novel soft computing models based on meta-heuristics and machine learning algorithms. Geosci. Front. 12(3), 101108 (2021) 39. A. Mahdiyar, D.J. Armaghani, A. Marto, M. Nilashi, S. Ismail, Rock tensile strength prediction using empirical and soft computing approaches. Bull. Eng. Geol. Environ. 78(6), 4519–4531 (2019) 40. D.F. Coates, Rock Mechanics Principles: Energy, Mines and Resources (Mines Branch, Gov. Ottawa, Canada, 1981) 41. F. Sereshki, S.M. Hoseini, M. Ataei, Blast fragmentation analysis using image processing. Int. J. Min. Geo-Engineering 50(2), 211–218 (2016) 42. A.K. Raina, A.K. Chakraborty, P.B. Choudhury, M. Ramulu, V. Udpikar, A. Sinha, Fragalyst 3.0: an indigenous fragmentation assessment tool based on digital image analysis–application and analysis. J. Mines, Met. Fuels 57(3&4), 83–88 (2009) 43. S. Nanda, B.K. Pal, Analysis of blast fragmentation using WipFrag. Int. J. Innov. Sci. Res. Technol. 5(6), 1561–1566 (2020) 44. A. Saber, Effects of window-to-wall ratio on energy consumption: application of numerical and ANN approaches. J. Soft Comput. Civ. Eng. 5(4), 41–56 (2021). https://doi.org/10.22115/ SCCE.2021.281977.1299 45. R. Saisubramanian, V. Murugaiyan, Prediction of compression index of marine clay using artificial neural network and multilinear regression models. J. Soft Comput. Civ. Eng. 5(4), 114–124 (2021). https://doi.org/10.22115/SCCE.2021.287537.1324 46. M. Koopialipoor, E.N. Ghaleini, H. Tootoonchi, D. Jahed Armaghani, M. Haghighi, A. Hedayat, Developing a new intelligent technique to predict overbreak in tunnels using an artificial bee colony-based ANN. Environ. Earth Sci. 78(5), 165 (2019) 47. V. Vapnik, S.E. Golowich, A.J. Smola, Support vector method for function approximation, regression estimation and signal processing, in Advances in Neural Information Processing Systems (1997), pp. 281–287 48. E. Li et al., Developing a hybrid model of salp swarm algorithm-based support vector machine to predict the strength of fiber-reinforced cemented paste backfill. Eng. Comput. (2020). https:// doi.org/10.1007/s00366-020-01014-x 49. M. Khandelwal, D.J. Armaghani, Prediction of drillability of rocks with strength properties using a hybrid GA-ANN technique. Geotech. Geol. Eng. 34(2), 605–620 (2016) 50. E. Momeni, R. Nazir, D.J. Armaghani, H. Maizir, Prediction of pile bearing capacity using a hybrid genetic algorithm-based ANN. Measurement 57, 122–131 (2014) 51. M. Monjezi, H. Amiri, A. Farrokhi, K. Goshtasbi, Prediction of rock fragmentation due to blasting in Sarcheshmeh copper mine using artificial neural networks. Geotech. Geol. Eng. 28(4), 423–430 (2010) 52. A. Bahrami, M. Monjezi, K. Goshtasbi, A. Ghazvinian, Prediction of rock fragmentation due to blasting using artificial neural network. Eng. Comput. 27(2), 177–181 (2011) 53. P. Kulatilake, W. Qiong, T. Hudaverdi, C. Kuzu, Mean particle size prediction in rock blast fragmentation using neural networks. Eng. Geol. 114(3–4), 298–311 (2010)

References

39

54. M. Monjezi, M. Rezaei, A.Y. Varjani, Prediction of rock fragmentation due to blasting in Gol-E-Gohar iron mine using fuzzy logic. Int. J. Rock Mech. Min. Sci. 46(8), 1273–1280 (2009) 55. L.A. Zadeh, Fuzzy sets. Inf. Control 8(3), 338–353 (1965) 56. M. Esmaeili, A. Salimi, C. Drebenstedt, M. Abbaszadeh, A.A. Bazzazi, Application of PCA, SVR, and ANFIS for modeling of rock fragmentation. Arab. J. Geosci. 8(9), 6881–6893 (2015) 57. Q. Fang, H. Nguyen, X.-N. Bui, T. Nguyen-Thoi, J. Zhou, Modeling of rock fragmentation by firefly optimization algorithm and boosted generalized additive model. Neural Comput. Appl. 33(8), 3503–3519 (2020) 58. S. Zhang, X.-N. Bui, N.-T. Trung, H. Nguyen, H.-B. Bui, Prediction of rock size distribution in mine bench blasting using a novel ant colony optimization-based boosted regression tree technique. Nat. Resour. Res. 29(2), 867–886 (2019) 59. M. Hasanipanah, H. Amnieh, H. Arab, M. Zamzam, Feasibility of PSO–ANFIS model to estimate rock fragmentation produced by mine blasting. Neural Comput. Appl. 30(4), 1015– 1024 (2018) 60. X. Shi, Z. Jian, B. Wu, D. Huang, W.E.I. Wei, Support vector machines approach to mean particle size of rock fragmentation due to bench blasting prediction. Trans. Nonferrous Met. Soc. China 22(2), 432–441 (2012) 61. J. Huang, P. Asteris, S. Pasha, A. Mohammed, M. Hasanipanah, A new auto-tuning model for predicting the rock fragmentation: a cat swarm optimization algorithm. Eng. Comput. 1, 1–12 (2020) 62. B. Vergara, M. Torres, V. Aramburu, C. Raymundo, Predictive model of rock fragmentation using the neuro-fuzzy inference system (ANFIS) and Particle swarm optimization (PSO) to estimate fragmentation size in open pit mining, in Advances in Manufacturing, Production Management and Process Control, eds. by S. Trzcielinski, B. Mrugalska, W. Karwowski, E. Rossi, M. Di. Nicolantonio AHFE 2021. Lecture Notes in Networks and Systems, vol. 274 (Springer, Cham, 2021), pp. 124–131 63. S. Mojtahedi, I. Ebtehaj, M. Hasanipanah, H. Bonakdari, H. Amnieh, Proposing a novel hybrid intelligent model for the simulation of particle size distribution resulting from blasting. Eng. Comput. 35(1), 47–56 (2018) 64. B. Murlidhar, D. Armaghani, E. Mohamad, S. Changthan, Rock fragmentation prediction through a new hybrid model based on imperial competitive algorithm and neural network. Smart Constr. Res. 2(1) (2018) 65. A. Sayadi, M. Monjezi, N. Talebi, M. Khandelwal, A comparative study on the application of various artificial neural networks to simultaneous prediction of rock fragmentation and backbreak. J. Rock Mech. Geotech. Eng. 5(4) (2013) 66. R. Trivedi, T. Singh, A. Raina, Prediction of blast-induced flyrock in Indian limestone mines using neural networks. J. Rock Mech. Geotech. Eng. 6(5), 447–454 (2014) 67. H. Rad, M. Hasanipanah, M. Rezaei, A. Eghlim, Developing a least squares support vector machine for estimating the blast-induced flyrock. Eng. Comput. 34(4), 709–717 (2018) 68. N. Ghaeini, M. Mousakhani, H.B. Amnieh, A. Jafari, Prediction of blasting fragmentation using the mutual information and rock engineering system; case study: Meydook copper mine. Int. J. Min. Geo-Eng. 51(1), 23–28 (2017) 69. A. Mehrdanesh, M. Monjezi, A.R. Sayadi, Evaluation of effect of rock mass properties on fragmentation using robust techniques. Eng. Comput. 34(2), 253–260 (2018) 70. K. Sayevand, H. Arab, S.B. Golzar, Development of imperialist competitive algorithm in predicting the particle size distribution after mine blasting. Eng. Comput. 34(2), 329–338 (2017) 71. E. Mutinda, B. Alunda, D. Maina, R. Kasomo, Prediction of rock fragmentation using the Kuznetsov-Cunningham-Ouchterlony model. J. South. African Inst. Min. Metall. 121(3), 107– 112 (2021) 72. X. Shi, D. Huang, J. Zhou, S. Zhang, Combined ANN prediction model for rock fragmentation distribution due to blasting. J. Inf. Comput. Sci. 10(11), 3511–3518 (2013)

Chapter 3

Applications of AI and ML Techniques to Predict Backbreak and Flyrock Distance Resulting from Blasting

3.1 Introduction Optimum fragmentation and minimal environmental impacts such as ground vibration, flyrock, and air overpressure are primary concerns in blasting operation. Flyrock is found to cause more than 40% fatal and 20% serious accidents based on study by Indian Mines [1]. Meanwhile, 50% of blast related accidents were caused by flyrock projections with distance range of 350–900 m and weight up to 500 kg in the United Kingdom during 1980–1985 [2]. Several researchers have found that flyrock causes up to 30% of blasting related accidents [3, 4]. The explosive energy is partially executed in rock fragmentation while the balance causes backbreak and flyrock [5, 6]. Backbreak causes rock beyond the last row of production to be crushed. Therefore, causing difficulty in drilling for the next round and reduce the blast performance [7]. Flyrock is undesirable throw of rock fragments during blasting. This accident is seldom reported and it concerns the prediction regime [8]. It must be noted that documenting flyrock with no accident could enhance the accuracy of current prediction models. Insufficient safeguarding of entry roads to blast area and failure to evacuate people from blast area are among the reasons for flyrock accidents [9]. The failure in evacuation is becoming more due to increase in use of all terrain vehicles (ATV’s). Proper training and education of person in charge for blast area security would lessen the unwanted accidents. Several blasting accidents involving flyrock in Malaysia were investigated by Mohamad et al. [10–12] which showed that excessive powder factor, geological structures, inappropriate blasting practices were the main causes of blasting accident. Various researchers have investigated causes of flyrock and backbreak [1, 13–15]. Factors can be divided into the following: • Geology: Discontinuity in geology and rock structure. • Rock mass properties: Rapid change in rock resistance due to presence of cracks or joints, various degree of rock weathering near an outcrop, faults, and slip planes as well as layers of mud, silt, or soft material in the host rock. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 R. M. Bhatawdekar et al., Environmental Issues of Blasting, SpringerBriefs in Applied Sciences and Technology, https://doi.org/10.1007/978-981-16-8237-7_3

41

42

3 Applications of AI and ML Techniques to Predict …

• Drill and blast design: Deviation of blastholes from planned direction causes a decrease in spacing or burden, excessive powder factor, insufficient stemming, inadequate and inappropriate delay timings. • Impact of previous blast—backbreaks, overhangs, and uneven highwall face. • Failure to identify uncontrollable factors: High concentration of explosive leads to extreme localized energy density caused by migration of explosive charge into mud seams, caverns, fissures, and voids. • Personal factors: Education, related experience, overwork, visual perception, and prior injury history. • Task Factors: Predrill inspection, pre- and post-blast examination, supervise loading, sound warning signal, timely warning and taking shelter, supervise charging and delay sequence, fire the blast, sound all-clear signal, and misfires. • Environmental factors: Uneven ground, drilling noise, foggy or stormy or excessive rain weather condition, lightening, movement of service vehicles, obstruction or visibility, nearby mines or community. • Governance or blast management practices: Lack appropriate training to blasting team. • Lack of technological tools: Every blasting process can be monitored using appropriate technology for blast fragmentation and environmental effect due to the blasting. There are three categories of available approaches for flyrock and backbreak including empirical, statistical-based, and artificial intelligence (AI). The first two categories have several shortcomings including difficulty to calculate, site-specific, and low level of accuracy [12, 16]. However, AI techniques were reported suitable, reliable, and effective by many researchers [17–20]. It is important to note that these techniques have been effectively used in other applications of civil and mining engineering [21–32]. In this study, AI background and techniques in predicting flyrock and backbreak will be discussed in details.

3.2 Measurement of Flyrock 3.2.1 Flyrock Measuring accurate flyrock is crucial for every blast to analyze and such data is useful for future prediction of flyrock. Fragment size more than 5–10 cm is considered dangerous and possible to identify at the site after blasting. Various techniques of measuring flyrock are adopted at each site. • Flyrock is normally measured with tape [33]. • The blast area is cleaned before operation to measure the flyrock distance. In addition, working bench is painted before blasting and two video cameras are set up to observe the projection of flyrock. Upon completion of blasting, the

3.2 Measurement of Flyrock

43

videos are reviewed to determine locations of maximum rock projections. Next, the maximum horizontal distances between landed fragments and free face are determined using a handheld global positioning system (GPS) and considered as flyrock distance [33]. • According to Trivedi et al. [34], only the fragments of more than 10 cm are chosen to determine the maximum throw of flyrock and determined using a handheld global positioning system (GPS). Other parameters such as flyrock launch angle and velocity shall be calculated using video analysis of blasting operation in ‘ProAnalyst 1.5.6.7’ software of XCITEX, USA. The video camera with high resolution (24 g frames/second) shall be utilized to record the event from a risk-free distance. The calibration of instrument shall be made with three red flags separated horizontally and vertically to known distance. In addition, other known parameters such as cut length and bench height shall be calculated and compared to verify instrument calibration.

3.2.2 Backbreak Backbreak is an adverse effect of blasting which results in uneven burden. This may result in flyrock, poor fragmentation in subsequent blast [35–38]. For every blast backbreak is measured. Distance of 3 m parallel to last row points are marked and after blast, backbreak is calculated by deducting distance of crack from marked points in the field Fig. 3.1.

Fig. 3.1 Backbreak measurement process

44

3 Applications of AI and ML Techniques to Predict …

3.3 Concepts of Some AI Models This section presents background and fundamental rules related to some important AI models which have been used in literature by the other researchers. This will give you a better view about these techniques and how they are able to do prediction.

3.3.1 Artificial Neural Network (ANN) Artificial Neural Network is one of the information-processing system and was modeled based on the structure of human brains. It is a computational model that composed of multiple layers and each layer contain multiple neurons or elements. Neurons of every layer were connected with the neurons of next layer through weighted connection that multiplies into the transmitted signal [39, 40]. In ANN, the inner layer(s) (except input and output layer) is known as hidden layer(s). Neurons’ number of the layer depends on the problem and its complexity except at first and last layers of hidden layer where it depends on the quantity of neurons at input and output layers. The weighted connection of links is determined using iteration through sequential calculations and the optimality of the network is determined using dataset that is not used in training of the network [40–47].

3.3.2 ANFIS ANFIS uses membership function to map associated and independent variable in order to get required result through output membership functions. Calculation of ANFIS modal may take place using five different layers named as rule layer, output membership function layer, output layer, input layer, and input membership function layer. In order to perform ANFIS, first dataset is classified into training and testing. After classifying the dataset into two different sets, the process of model generation is initiated. In order to generalize the model, training dataset is used and after generalization to check the validity of the generated model testing dataset is used. After the testing process, error calculation is performed using the same testing dataset. For generalization of ANFIS model, subtractive clustering method is used, which is performed after defining the value of squash factor (SF) and range of influence (ROI). The value of SF and ROI is defined as it directly affect the RMSE of the model because it impacts the number of rules needed for generating ANFIS model. The best value of SF and ROI is selected using the iterative process in order to predict model with least complicacy in terms of rules and RMSE. After getting the best value of SF and ROI, the generated model is tuned in order to reduce the error by performing hybrid algorithm which is a combination of back propagation and least square technique. This hybrid algorithm tunes the ANFIS model to get final output value. During

3.3 Concepts of Some AI Models

45

tuning the model in forward and backward passes, it minimizes the errors related to calculation of weights. In order to get the limit of tuning, i.e. overfitting property which the result of uncommon increase in the error for continuous iteration number, a graph is plotted between error and epochs. The RMSE value of the prediction shows the degree of precision.

3.3.3 Support Vector Machine (SVM) In case of regression, performances of SVM depends on the combination of parameters such as capacity parameter (C), the kernel type (K), εε-insensitive loss function (εε), and their corresponding parameters. The adjustment between minimizing the training error and maximizing the margin is controlled by C parameter. It is because the large value of C overfits the training data where as too small value of C places insufficient stress on fitting of the training data. The value εε totally depends on the noise type present in the dataset, it basically helps to avoid the training dataset in meeting with the boundary condition. Based on the previous researchers, it is found that, in case of SVM, out of all kernel functions that Gaussian radial basis function, has an enhanced efficiency. It may be shown as in Eqs. 3.1 and 3.2. 





k xi , x j = e   k xi , x j = e



−

xi −x j 2



2σ 2

(3.1)

−γ xi −x j 

2



(3.2)

where σ is a constant parameter of the kernel and can either control the amplitude of the Gaussian function or the generalization ability of SVM. In order to evaluate quality of the model RMSE is used.

3.3.4 ELM Extreme learning machine (ELM) was first introduced by Huang et al. [48] in the year of 2006. Its concept is derived from the single hidden layer feed-forward neural network (SLFN). This machine learning is quite easy to implement as it does not require manual assignment of weight and bias and it was developed to overcome the disadvantages of traditional learning techniques based on gradient exchange. ELM consists of three layers, namely input layer, hidden layer, and output layer. It is a learning technique to train ANN which randomly generates bias (b) and input weights (w). Meanwhile, the output weight is altered during training phase in this technique.

46

3 Applications of AI and ML Techniques to Predict …

The benefits of using this model instead of classical singleton model trained by gradient algorithms are: • It is possible to utilize nonlinear and kernel functions in ELM. • ELM does not require momentum and learning rate unlike classical algorithms. • ELM acts faster if only output weights need to be adjusted unlike classical singleton model. • ELM has great generalizability as compared to classical neural networks. As ELM utilize of gradient-based learning algorithms or stochastic selection of learning parameters, it could fall into the local optimal solution. Therefore, it could suffer from poor generalization capability.

3.3.5 PSO-ELM ELM is theoretically able to execute a universal estimation and engage an amount of activation functions [48]. Various researchers have utilized it for estimation due to its fast-learning capacity [49]. Integrating other techniques with ELM could enhance its generalization capacity even better [50]. It also has been optimised by integrating with nature-based algorithms. For an example, Mohapatra et al. [51] developed a hybrid model with ELM and cuckoo search algorithm to categorize medical data. Satapathy et al. [52] employed firefly algorithm for optimization of ELM and used it for photovoltaic interactive microgrid’s stability analysis. Meanwhile, Li et al. [53] utilized whale optimization algorithm to optimize ELM and evaluate aging degree of insulated gate bipolar transistor. Findings shows that optimized ELM provides accurate results than the singleton ELM. The ELM solution models may get trapped in local minima due stochastic initialization of network input weights and hidden biases [54]. Therefore, particle swarm optimization (PSO) shall be utilized for optimization of ELM parameter set (input weights and hidden biases) and acquire ELM of higher learning capacity. Figure 3.2 shows process of PSO in optimizing problems.

3.4 Backbreak Prediction Using AI Techniques Backbreak is an adverse effect in blasting which affects subsequent blast performance such as flyrock, larger size fragmentation. Various studies for prediction of backbreak using AI techniques were reviewed. All researchers used blast design parameters and powder factor as input parameter for prediction of backbreak. For a given rock type, design of optimum spacing and burden is required with a given hole diameter. With excess burden, backbreak is likely to occur. Similarly, with excess spacing side break is likely to occur. With inadequate stemming from last row, backbreak may occur. Sari et al. [38] predicted backbreak with simple regression, multivariable regression analysis (MVRA), and Monte Carlo Simulation. With simple regression of

3.4 Backbreak Prediction Using AI Techniques

47

Fig. 3.2 A simple process of PSO in optimizing problems

backbreak, for individual input parameters burden, spacing, stemming, powder factor, geometric stiffness, coefficient of determination (R2 ) was 0.972, 0.697, 0.957, 0.872, 0.725, respectively. Only spacing showed negative relationship with backbreak. Balance all other parameters showed positive relationship with backbreak. PSO was used to predict backbreak by Hasanipanah et al. [55] and Eskandar et al. [56] with R2 of 0.88 and 0.96, respectively. In addition, PSO (Linear) and PSO (Quadratic) showed R2 of 0.973 and 0.983, respectively in the study conducted by Ghasemi [57]. ANFIS was applied to predict backbreak by different scholars [58–60] with R2 from 0.95 to 0.998. Table 3.1 shows the summary of prediction of backbreak due to the blasting using AI models. Data size, technique, input variables, R2 , and system error of the developed models are presented in this table. More discussion about these techniques will be given later.

3.5 Flyrock Prediction Using AI Techniques As flyrock is the most adverse environmental effect and likely to have maximum impact due to the blasting, prediction of this phenomenon is essential. Input parameters are analyzed based on blast design, rock mass properties, and parameters related to explosives. If front burden (B) or stemming length (ST) is lower, it can cause flyrock. Higher the hole depth (HD), flyrock is higher. With increase in hole diameter, flyrock is increased. Lower the specific drilling, flyrock distance is minimized. Rock density was used as one of the input parameters by many researchers [69–71]. Other rock mass properties by some of the researchers are blastability index (BI), rock mass rating (RMR) and Schmidt hammer rebound number (Rn). Powder factor

48

3 Applications of AI and ML Techniques to Predict …

Table 3.1 Prediction of backbreak due to the blasting using AI models R2

RMSE

42

0.95

0.6

B, S, ST, PF, HD, SD

234

0.987

–

B, S, ST, H, PF, SD

103

0.871

0.22

0.515

0.31

References

Technique

Input

Bazzazi and Esmaeili [58]

ANFIS

SB, H/B, d, ST, PF, RD, UCS, Nr , CLD, CPT

Khandelwal and Monjezi [61]

SVM

Sayadi et al. [62]

BPNN

Data size

RBFNN Esmaeili et al. [59]

ANFIS ANN

SB, H/B, ST, PF, RD, Nr , CLR

Sari et al. [38]

MVRA-LNBB

B, S, ST, PF, K

Ebrahimi et al. [63]

ANN

B, S, ST, HD, PF

Faradonbeh et al. [64]

GP

Ghasemi et al. [60]

RF

42

0.96

0.6

0.92

0.88

175

0.981

0.31

34

0.77

0.53

B, S, ST, PF, H/B

175

0.976

0.327

B, S, ST, PF, K

175

0.971

0.35

ANFIS

Hasanipanah et al. PSO [55]

B, S, ST, PF, RD

Ghasemi [57]

B, S, ST, PF, K

GP

0.998

0.08

76

0.96

12.01

175

0.979

0.314

PSO-linear

0.973

0.352

PSO-quadratic

0.983

0.278

Hasanipanah et al. PSO-ANFIS [65]

B, S, ST, PF

80

0.922

0.13

Eskandar et al. [56]

B, S, ST, PF, RMR

84

0.96

0.081

B, C, PF, SB, ST/B, i, VOD, d, d/B

62

0.934

0.058

0.963

0.041

PSO

Hasanipanah et al. GA [66] ICA Kumar et al. [67]

RF

SB, H/ST, ED, p-wave 140

0.979

0.878

Zhou et al. [68]

ELM

B, S, ST, HD, PF, SD

0.95

0.288

0.965

0.213

GRNN

234

RBF

0.957

0.246

SCA-RF

0.982

0.099 (continued)

(PF) represents explosives energy per tonne of rock blasted. Flyrock has direct correlation with PF. Maximum charge per delay (c) is indicator of maximum explosives energy as a particular instant. Flyrock distance increases with increase in C. ANN as the basic AI technique was used by several researchers to predict flyrpck distance [46, 72, 73]. Number of datasets varied from 65 to 240 and R2 from 0.85 to 0.978 for prediction of flyrock with ANN. SVM model is also used for prediction

3.5 Flyrock Prediction Using AI Techniques

49

Table 3.1 (continued) References

Technique

Input

Data size

HHO-RF

R2

RMSE

0.981

0.106

PLR point load index (Mpa), BPNN back propagation neural network, RBFNN radial basis function neural network, LMR linear multivariate regression analysis, MVRA multivariate regression analysis, PSO particle swarm optimization, ANFIS adaptive neuro-fuzzy inference system, ANN artificial neural network, RT regression tree, NLMR nonlinear multivariate regression analysis, GP genetic programming, LNBB the natural logarithm of backbreak, SCA sine cosine algorithm, HHO Harris hawks optimizer, ELM extreme learning machine, RBF radial basis function, GRNN network and general regression neural network, RF random forest, GA genetic algorithm, ICA imperialist competitive algorithm, SVM support vector machine, B burden, S spacing, ST stemming length, H bench height, PF powder factor/specific charge, HD hole depth, SD specific drilling, SB spacing to burden ratio, H/B stiffness ratio, d hole diameter, ED explosive density, RD rock density, UCS uniaxial compressive strength, N r number of rows, CLR charge last row, CPT charge last row per total charge ratio, K geometric stiffness, H/ST bench height to stemming, t time delay, C maximum instantaneous charge, DO discontinuities orientation to face, PFLT last row powder factor to total powder factor, VOD velocity of detonation, PF powder factor, BD blasthole deviation, RMR rock mass rating, ST/B stemming to burden ratio, i blasthole inclination, B/d burden to hole diameter ratio

of flyrock [61, 73]. Number of datasets varied from 234 to 262 and R2 values varied from 0.91 to 0.993. Various researchers used hybrid models with ANN such as artificial bee colonization (ABC)-ANN, firefly algorithm (FA)-ANN, and PSO-ANN were applied by several scholars [32, 36, 69]. Number of datasets varied from 44 to 262 and R2 varied from 0.841 to 0.98 for prediction of flyrock with these hybrid models. In addition to the mentioned models, many researchers have developed new hybrid models such as recurrent fuzzy neural network (RFNN)-PSO, biogeographybased optimization (BBO)-ELM, PSO-ELM, and RFNN-GA for the same issue [16, 18, 74, 75]. Number of datasets varied from 70 to 262 with a wide variation of R2 . More information regarding the most important AI models in the area of flyrock are presented in Table 3.2. In this table, all predictors are divided into three Table 3.2 Prediction of flyrock due to the blasting using AI models References

Technique

Blast design

Rock mass properties

Parameters related to explosives and others

Data size

R2

Monjezi et al. [76]

ANN

HD, ST, BS, SpD

RD

PF, C, N

192

–

Bharami et al. [46]

ANN

B, S, HD, ST

RMR, BI

PF, C

192

–

Monjezi et al. [77]

ANN

B, d, HD, ST, BS, SpD

BI

PF, C

97

0.9

(continued)

50

3 Applications of AI and ML Techniques to Predict …

Table 3.2 (continued) References

Technique

Rad et al. [70] FIS SM

Blast design

Rock mass properties

Parameters related to explosives and others

Data size

R2

B, S, HD, ST, SpD

RD

PF, C

490

0.984

B, S, d, HD, ST, SpD

–

PF

245

0.85

0.701

Amini et al. [78]

ANN

Monjezi et al. [36]

ANN-GA

B, S, HD, ST, SpD

RMR

PF, C

195

0.98

Monjezi et al. [79]

ANN

B, S, d, HD

–

C

–

0.86

Khandelwal and Monjezi [61]

SVM

B, S, HD, ST, SpD

–

PF

234

0.95

Armaghani et al. [32]

ANN-PSO

B, S, d, HD, ST, SpD

RD

PF, C, N

44

0.93

Ghasemi et al. ANN [80] FIS

B, S, HD, ST

–

PF, C

230

0.94

Marto et al. [69]

ANN-ICA

HD, ST, BS

RD

PF, C

113

0.98

Armaghani et al. [81]

ANN

–

–

PF, C

232

0.92

Faradonbeh et al. [82]

GP

D, HD, BS, St,

–

PF, C

262

0.908

Hasanipanah et al. [55]

PSO

B, S, ST

RD

PF

76

0.96

Faradonbeh et al. [83]

GA

–

PF

76

0.92

FA

B, S, HD, ST

Koopialipoor et al. [84]

ICA-ANN

BS, D, St

–

PF, C

262

0.958

SVM

0.94

0.96

ANFIS

0.98

0.924

GA-ANN

0.932

PSO-ANN Nguyen et al. [85]

ANN

Wu et al. [86]

ICA-Linear

0.959 – B, S, St

W, PF

210

EANN ICA-Power ICA-Quadratic

0.975 0.986

B, S, St

RMR

Weight charge

78

0.954 0.928 0.952 (continued)

3.5 Flyrock Prediction Using AI Techniques

51

Table 3.2 (continued) References

Technique

Blast design

Rock mass properties

Parameters related to explosives and others

Data size

R2

BS, D, St, LD

–

PF, C

262

0.63

S, B, St

RD

ANN Armaghani et al. [19]

PCR SVR

0.841 0.91

MARS Hasanipanah et al. [87]

ANN

0.86 PF

82

ANN-PSO

0.833

ANN-HS

0.872

ANN-ADHS

0.930

Kalaivaani et al. [74]

RFNN-PSO

Lu et al. [88]

ELM

St, B, S

–

C

72

0.933

B, S, St

RD

PF

82

0.958

ANFIS

0.743

ORELM

0.955

ANN Murlidhar et al. [16]

0.832

BBO-ELM PSO-ELM ELM

Rad et al. [89] RFNN-GA

0.912 B/S, hole diameter, St, hole depth

–

B, S, St

–

PF, C

262

0.94 0.93 0.79

C

70

GA-ANN

0.967 0.944

ANN

0.867

Zhou et al. [90]

ANN

HD, B, S, St

–

C, PF

65

0.906

Fattahi and Hasanipanah [71]

ANFIS-GOA

S, B, St

RD

PF

80

0.974

ANFIS-CA

Guo et al. [91] SVRs

0.953 B, S, St

–

W, PF

210

Blasthole length, C, B, S, St

–

PF

240

BS, HD, St,

RD, Rn

SVRs-GLMNET Guo et al. [18] DNN ANN Li et al. [17]

ICA-ANN PSO-ANN

0.972 0.993 0.978 0.8539

C, PF

113

0.960 0.961 (continued)

52

3 Applications of AI and ML Techniques to Predict …

Table 3.2 (continued) References

Technique

Blast design

Rock mass properties

Parameters related to explosives and others

Data size

R2

ABC-ANN

0.967

FA-ANN

0.972

GA-ANN

0.947

Monjezi et al. [92]

GEP

B, S, St

–

PF

152

0.910

Nguyen et al. [93]

WOA-SVM-L

B, St, S

–

Wa, PF

210

0.937

Ye et al. [94] Dehghani et al. [95]

WOA-SVM-P

0.976

WOA-SVM-RBF

0.977

WOA-SVM-HT

0.972

ANFIS

0.965

GBM

0.971

RF

0.972

CART

0.973

ANN

0.971

GP

St, D, BS, HD

–

PF, C

262

0.908

RF GEA

B, S, St

–

CL, PF

318

0.910

0.905

ELM Extreme learning machine, ORLEM outlier robust ELM, ANN artificial neural network, MLR multiple linear regression, RFNN-GA recurrent fuzzy neural network combined with genetic algorithm, GA-ANN genetic algorithm combined with ANN, BBO biogeography-based optimization, DNN deep neural network, LMR linear multivariate regression, GEP gene expression programming, ANFIS-GOA adaptive neuro-fuzzy inference system in combination with grasshopper optimization algorithm, ANFIS-CA adaptive neuro-fuzzy inference system in combination with cultural algorithm, ANN-PSO artificial neural network with particle swarm optimization, ANN-HS artificial neural network coupled with harmony search, ANN-ADHS artificial neural network coupled with advanced dynamical harmony search, GP genetic programming, RF random forest, EANN ensemble of ANN models, ICA-ANN imperialist competitive algorithm ANN, PSO-ANN particle swarm optimization ANN, ABC-ANN artificial bee colonization ANN, FA-ANN firefly algorithm ANN, GA-ANN genetic algorithm ANN, GEA gene expression algorithm, RFNN-PSO recurrent fuzzy neural network PSO, ANFIS adaptive neuro-fuzzy inference system, PCR principal component regression, SVR support vector regression, MARS multivariate adaptive regression splines, NLMR nonlinear multiple regression, CART classification and regression tree, WOA-SVM-L whale optimization algorithm-support vector machine-linear, WOA-SVM-RBF whale optimization algorithmsupport vector machine-radius basis function, WOA-SVM-P whale optimization algorithm-support vector machine-polynomial, WOA-SVM-HT whale optimization algorithm-support vector machinehyperbolic tangent, GBM gradient boosting machine, RF random forest, B burden, S spacing, PF powder factor, St stemming, RD rock density, C maximum charge per delay, D blasthole diameter, BS burden to spacing ratio, HD blasthole depth, W capacity of the explosive charge, CL charge length, LD length of blasthole, Wa amount of explosive used per blast

3.5 Flyrock Prediction Using AI Techniques

53

categories which are blast design, rock mass properties, and explosive properties. We will discuss more about these parameters in the following section.

3.6 Discussion Flyrock is one of the most crucial risk due to the blasting. For minimizing this risk, prediction of flyrock plays an important role. During 1975–1990, flyrock prediction was done by empirical equations dependent upon hole diameter, stemming to burden ratio. Richards and Moore [96] and Little [97] developed equations for flyrock prediction based on scaled burden method for calculation of initial velocity. Equations for various types of flyrock (Face burst, and cratering were based on drill hole angle, burden, spacing, charge per meter, gravitational, and site constants). Pioneer work on blast design parameters and flyrock is even important for input parameters for prediction of flyrock. Powder factor and charge per delay are important explosives related parameters which are essential for flyrock prediction. Some of the researchers have used RMR, BI, RQD, and rock density as input parameters. Still research work needs to be carried out on geological discontinuities, rock mass properties which can play important role as input parameters. Backbreak is negative effect due to the blasting which can have impact of blasts being conducted later on. Hence, it is vital to predict backbreak to minimize this adverse effect. Backbreak is dependent on most of the controllable parameters blast design parameters. Based on the various studies, it can be concluded that burden, bench height, powder factor, maximum charge per delay has direct relationship with backbreak. Stemming length and spacing has inverse relationship with backbreak. However, based on field experimentation and prediction with AI techniques, these parameters can be optimized. Limited researchers have used RMR, UCS as input parameters for backbreak. Initially, researchers utilized simple regression, MVRA, and Monte Carlo simulation to predict backbreak. ANN, LMR, PSO, and other optimization techniques showed good prediction accuracy with backbreak. Still research needs to be carried out based on local geological conditions such as softer rock or variation in strata and prediction of backbreak. After development of various mathematical equations, semi-empirical equations, statistical models were developed for prediction of flyrock. However, all these models were having less accuracy. During the beginning of last decade, ANN was mainly used for prediction of flyrock. ANN can be considered as benchmark compared with other AI techniques in both flyrock and backbreak prediction. SVM, simple optimization algorithms such as PSO, GA, and ICA have shown good prediction accuracy of flyrock and backbreak. Further accuracy is enhanced with hybrid models with ANN or other hybrid models such as RFNN-PSO, BBO-ELM, and PSO-ELM. Various researchers have developed new models such as ICA-Linear, ICA-Power, ICA-Quadratic WOA-SVM-L, WOA-SVM-P, WOA-SVM-RBF, and WOA-SVMHT. Flyrock is likely to occur with single hole condition instead of average parameters utilized for prediction of flyrock. Based on the historical database, further research

54

3 Applications of AI and ML Techniques to Predict …

is required so that AI techniques can provide alternative solutions for charging hole or another correction to blasting engineer at site to minimize risk of flyrock and backbreak after blasting inspection of the blasting face. As a new direction of AI models, physics-based AI can be considered by mining and blast engineers. In this way, computer science knowledge can be mixed with well-known theories from blasting to give a better understanding and view to researchers in the areas of blasting environmental issues.

3.7 Conclusion 1.

2. 3. 4.

5. 6. 7.

8. 9.

Blast design parameters are known as controllable parameters such as burden, spacing, hole diameter, powder factor, and maximum charge per meter play an important role in prediction of flyrock and backbreak. During initial era, prediction of backbreak and flyrock was carried out with empirical equations and statistical models which are less accurate. MVRA technique was used by some of the researchers to benchmark performance of AI techniques. Some of the researchers used RMR, BI, RQD, rock density as rock mass properties for prediction of flyrock. On the other hand, some of the researchers used RMR, UCS for prediction of backbreak. ANN has been widely used by many researchers for prediction of flyrock and backbreak. Hybrid models with ANN, PSO, SVM produce better accuracy for perdition of flyrock and backbreak compared to their base models. Practical application AI techniques for prediction of backbreak or flyrock include to select various input parameters which have contributed at various sites. AI technique can be selected based on which have been deployed by many researchers at different sites. Further research is required for capturing geological data with drones and utilize for design of blast to minimize risk of flyrock and backbreak. Theory-based or physics-based AI is considered as new technologies which are able to combine advantages of both AI and well-known theories in mining and civil engineering to give a better view to researchers.

References 1. S. Bhandari, Flyrock during blasting operations—controllable environmental hazard, in 2nd National Seminar on Minerals and Ecology (1994), pp. 279–308 2. S. Bhandari, Engineering Rock Blasting Operations (A.A. Balkema, 1997), p. 388

References

55

3. T.S. Bajpayee, T.R. Rehak, G.L. Mowrey, D.K. Ingram, A summary of fatal accidents due to flyrock and lack of blast area security in surface mining, 1989 to 1999, in Proceeding of the 28th Annual Conference Explosive and Blasting Technique, January 2001 (2002), pp. 105–118 4. T.R. Rehak, T.S. Bajpayee, G.L. Mowrey, D.K. Ingram, Flyrock issues in blasting (2001) 5. T. Singh, V. Singh, An intelligent approach to prediction and control ground vibration in mines. Geotech. Geol. Eng. 23, 249–262 (2005) 6. M. Monjezi, H. Amiri, A. Farrokhi, K. Goshtasbi, Prediction of rock fragmentation due to blasting in Sarcheshmeh copper mine using artificial neural networks. Geotech. Geol. Eng. 28(4), 423–430 (2010) 7. C. Jimeno, E. Jimeno, F. Carcedo, Drilling and Blasting of Rocks (A.A. Balkema, Rotterdam, 1995) 8. P.A. Davies, Risk-based approach to setting of flyrock danger zones for blast sites. Trans. Inst. Min. Metall., 96–100 (1995) 9. V. Kecojevic, M. Radomsky, Flyrock phenomena and area security in blasting-related accidents. Saf. Sci. 43(9), 739–750 (2005) 10. E.T. Mohamad, D.J. Armaghani, M. Hajihassani, K. Faizi, A. Marto, A simulation approach to predict blasting-induced flyrock and size of thrown rocks. Electron. J. Geotech. Eng. 18(B), 365–374 (2013) 11. E.T. Mohamad, B.R. Murlidhar, D.J. Armaghani, R. Saad, C.S. Yi, Effect of geological structure and blasting practice in fly rock accident at Johor, Malaysia. J. Teknol. 78(8–6) (2016) 12. E.T. Mohamad, C.S. Yi, B.R. Murlidhar, R. Saad, Effect of geological structure on flyrock prediction in construction blasting. Geotech. Geol. Eng. 36(4), 2217–2235 (2018) 13. G.R. Adhikari, Studies on flyrock at limestone quarries. Rock Mech. Rock Eng. 32(4), 291–301 (1999) 14. P.P. Roy, Rock Blasting: Effects and Operations (A.A. Balkema Publishers, Leiden, Netherlands, 2005) 15. H.S. Venkatesh, R.M. Bhatawdekar, G.R. Adhikari, A.I. Theresraj, Assessment and mitigation of ground vibrations and flyrock at a limestone quarry, in Proceedings of the Annual Conference on Explosives and Blasting Technique (1999), pp. 145–152 16. B.R. Murlidhar, D. Kumar, D. Jahed Armaghani, E.T. Mohamad, B. Roy, B.T. Pham, A novel intelligent ELM-BBO technique for predicting distance of mine blasting-induced flyrock. Nat. Resour. Res. (2020). https://doi.org/10.1007/s11053-020-09676-6 17. D. Li, M. Koopialipoor, D.J. Armaghani, A combination of fuzzy Delphi method and ANNbased models to investigate factors of flyrock induced by mine blasting. Nat. Resour. Res. (2021). https://doi.org/10.1007/s11053-020-09794-1 18. S. Nazeer, R.K. Dutta, Application of machine learning techniques in predicting the bearing capacity of E-shaped footing on layered sand. J. Soft Comput. Civ. Eng. 5(4), 74–89 (2021). https://doi.org/10.22115/SCCE.2021.303113.1360 19. R. Saisubramanian, V. Murugaiyan, Prediction of compression index of marine clay using artificial neural network and multilinear regression models. J. Soft Comput. Civ. Eng. 5(4), 114–124 (2021). https://doi.org/10.22115/SCCE.2021.287537.1324 20. A. Saber, Effects of window-to-wall ratio on energy consumption: application of numerical and ANN approaches. J. Soft Comput. Civ. Eng. 5(4), 41–56 (2021). https://doi.org/10.22115/ SCCE.2021.281977.1299 21. E. Tonnizam Mohamad, D. Jahed Armaghani, M. Hasanipanah, B.R. Murlidhar, M.N.A. Alel, Estimation of air-overpressure produced by blasting operation through a neuro-genetic technique. Environ. Earth Sci. 75(2), 1–15 (2016) 22. Q. Fang, B.Y. Bejarbaneh, M. Vatandoust, D.J. Armaghani, B.R. Murlidhar, E.T. Mohamad, Strength evaluation of granite block samples with different predictive models. Eng. Comput. (2019). https://doi.org/10.1007/s00366-019-00872-4 23. D.J. Armaghani, E.T. Mohamad, M.S. Narayanasamy, N. Narita, S. Yagiz, Development of hybrid intelligent models for predicting TBM penetration rate in hard rock condition. Tunn. Undergr. Sp. Technol. 63, 29–43 (2017)

56

3 Applications of AI and ML Techniques to Predict …

24. H. Naderpour, A.H. Rafiean, P. Fakharian, Compressive strength prediction of environmentally friendly concrete using artificial neural networks. J. Build. Eng. 16, 213–219 (2018). https:// doi.org/10.1016/j.jobe.2018.01.007 25. E.T. Mohamad, M. Koopialipoor, B.R. Murlidhar, A. Rashiddel, A. Hedayat, D.J. Armaghani, A new hybrid method for predicting ripping production in different weathering zones through in-situ tests. Measurement (2019). https://doi.org/10.1016/j.measurement.2019.07.054 26. R.M. Bhatawdekar, E. Tonnizam Mohamad, T.N. Singh, P. Pathak, D.J. Armaghani, Rock mass classification for the assessment of blastability in tropically weathered limestones, in International Conference on Innovations for Sustainable and Responsible Mining, vol. 109 (2021), pp. 13–44 27. E.T. Mohamad, D. Li, B.R. Murlidhar, D.J. Armaghani, K.A. Kassim, I. Komoo, The effects of ABC, ICA, and PSO optimization techniques on prediction of ripping production. Eng. Comput. (2019). https://doi.org/10.1007/s00366-019-00770-9 28. B.R. Murlidhar, B.Y. Bejarbaneh, D.J. Armaghani, A.S. Mohammed, E.T. Mohamad, Application of tree-based predictive models to forecast air overpressure induced by mine blasting. Nat. Resour. Res. (2020). https://doi.org/10.1007/s11053-020-09770-9 29. D. Jahed Armaghani, A. Azizi, A comparative study of artificial intelligence techniques to estimate TBM performance in various weathering zones, in Applications of Artificial Intelligence in Tunnelling and Underground Space Technology. SpringerBriefs in Applied Sciences and Technology (Springer, Singapore, 2021), pp. 55–70. https://doi.org/10.1007/978-981-16-103 4-9_4 30. D. Jahed Armaghani, A. Azizi, Empirical, statistical, and intelligent techniques for TBM performance prediction, in Applications of Artificial Intelligence in Tunnelling and Underground Space Technology. SpringerBriefs in Applied Sciences and Technology (Springer, Singapore, 2021), pp. 17–32. https://doi.org/10.1007/978-981-16-1034-9_2 31. B.R. Murlidhar, D.J. Armaghani, E.T. Mohamad, Intelligence prediction of some selected environmental issues of blasting: a review. Open Constr. Build. Technol. J. 14(1), 298–308 (2020) 32. E. Ford, K. Maneparambil, N. Neithalath, Machine learning on microstructural chemical maps to classify component phases in cement pastes. J. Soft Comput. Civ. Eng. 5(4), 1–20 (2021). https://doi.org/10.22115/SCCE.2021.302400.1357 33. D.J. Armaghani, A. Mahdiyar, M. Hasanipanah, R.S. Faradonbeh, M. Khandelwal, H.B. Amnieh, Risk assessment and prediction of flyrock distance by combined multiple regression analysis and Monte Carlo simulation of quarry blasting. Rock Mech. Rock Eng. 49(9), 1–11 (2016) 34. R. Trivedi, T.N. Singh, N. Gupta, Prediction of blast-induced flyrock in opencast mines using ANN and ANFIS. Geotech. Geol. Eng. 33(4), 875–891 (2015) 35. M. Monjezi, H. Dehghani, Evaluation of effect of blasting pattern parameters on back break using neural networks. Int. J. Rock Mech. Min. Sci. 45(8), 1446–1453 (2008) 36. M. Monjezi, H. Khoshalan, A. Varjani, Prediction of flyrock and backbreak in open pit blasting operation: a neuro-genetic approach. Arab. J. Geosci. (2012) 37. F. Faramarzi, M. Ebrahimi Farsangi, H. Mansouri, An RES-based model for risk assessment and prediction of backbreak in bench blasting. Rock Mech. Rock Eng. 46(4), 877–887 (2012) 38. M. Sari, E. Ghasemi, M. Ataei, Stochastic modeling approach for the evaluation of backbreak due to blasting operations in open pit mines. Rock Mech. rock Eng. 47(2), 771–783 (2014) 39. A. Abraham, Meta learning evolutionary artificial neural networks. Neurocomputing 56(1–4), 1–38 (2004) 40. M. Monjezi, M. Ahmadi, M. Sheikhan, A. Bahrami, A.R. Salimi, Predicting blast-induced ground vibration using various types of neural networks. Soil Dyn. Earthq. Eng. 30(11), 1233– 1236 (2010) 41. M. Khandelwal, T. Singh, Prediction of blast induced ground vibrations and frequency in opencast mine: a neural network approach. J. Sound Vib. 289(4–5), 711–725 (2006) 42. M. Khandelwal, T. Singh, S. Kumar, Prediction of blast induced ground vibration in opencast mine by artificial neural network. Indian Min. Eng. J. 44, 9–23 (2005)

References

57

43. M. Khandelwal, T. Singh, Prediction of blast-induced ground vibration using artificial neural network. Int. J. Rock Mech. Min. Sci. 46(7), 1214–1222 (2009) 44. M. Khandelwal, D.L. Kumar, M. Yellishetty, Application of soft computing to predict blastinduced ground vibration. Eng. Comput. 27(2), 117–125 (2011) 45. F. Meulenkamp, M. Grima, Application of neural networks for the prediction of the unconfined compressive strength (UCS) from Equotip hardness. Int. J. Rock Mech. Min. Sci. 36(1), 29–39 (1999) 46. A. Bahrami, M. Monjezi, K. Goshtasbi, A. Ghazvinian, Prediction of rock fragmentation due to blasting using artificial neural network. Eng. Comput. 27(2), 177–181 (2011) 47. K. Neaupane, N. Adhikari, Prediction of tunneling-induced ground movement with the multilayer perceptron. Int. J. Tunn. Undergr. Sp. Technol. 21, 151–159 (2006) 48. G.-B. Huang, L. Chen, C.K. Siew, Universal approximation using incremental constructive feedforward networks with random hidden nodes. IEEE Trans. Neural Netw. 17(4), 879–892 (2006) 49. D. Cui, G.-B. Huang, T. Liu, ELM based smile detection using distance vector. Pattern Recognit. 79, 356–369 (2018) 50. H. Zhu, E.C.C. Tsang, J. Zhu, Training an extreme learning machine by localized generalization error model. Soft Comput. 22(11), 3477–3485 (2018) 51. P. Mohapatra, S. Chakravarty, P. Dash, An improved cuckoo search based extreme learning machine for medical data classification. Swarm Evol. Comput. 24, 25–49 (2015) 52. P. Satapathy, S. Dhar, P.K. Dash, An evolutionary online sequential extreme learning machine for maximum power point tracking and control in multi-photovoltaic microgrid system. Renew. Energy Focus 21, 33–53 (2017) 53. L.-L. Li, J. Sun, M.-L. Tseng, Z.-G. Li, Extreme learning machine optimized by whale optimization algorithm using insulated gate bipolar transistor module aging degree evaluation. Expert Syst. Appl. 127, 58–67 (2019) 54. J. Cao, Z. Lin, G.-B. Huang, Self-adaptive evolutionary extreme learning machine. Neural Process. Lett. 36(3), 285–305 (2012) 55. M. Hasanipanah, D.J. Armaghani, H.B. Amnieh, M.Z.A. Majid, M.M.D. Tahir, Application of PSO to develop a powerful equation for prediction of flyrock due to blasting. Neural Comput. Appl. 28(1), 1043–1050 (2017) 56. H. Eskandar, E. Heydari, M. Hasanipanah, M. Jalil Masir, A. Mahmodi Derakhsh, Feasibility of particle swarm optimization and multiple regression for the prediction of an environmental issue of mine blasting. Eng. Comput. (Swansea, Wales) 35(1) (2018) 57. E. Ghasemi, Particle swarm optimization approach for forecasting backbreak induced by bench blasting. Neural Comput. Appl. 28(7), 1855–1862 (2017) 58. A.A. Bazzazi, M. Esmaeili, Prediction of backbreak in open pit blasting by adaptive neuro-fuzzy inference system. Arch. Min. Sci. 57(4), 933–943 (2012) 59. M. Esmaeili, A. Salimi, C. Drebenstedt, M. Abbaszadeh, A.A. Bazzazi, Application of PCA, SVR, and ANFIS for modeling of rock fragmentation. Arab. J. Geosci. 8(9), 6881–6893 (2015) 60. E. Ghasemi, H.B. Amnieh, R. Bagherpour, Assessment of backbreak due to blasting operation in open pit mines: a case study. Environ. Earth Sci. 75(7), 1–11 (2016) 61. M. Khandelwal, M. Monjezi, Prediction of backbreak in open-pit blasting operations using the machine learning method. Rock Mech. rock Eng. 46(2), 389–396 (2013) 62. A. Sayadi, M. Monjezi, N. Talebi, M. Khandelwal, A comparative study on the application of various artificial neural networks to simultaneous prediction of rock fragmentation and backbreak. J. Rock Mech. Geotech. Eng. 5(4) (2013) 63. E. Ebrahimi, M. Monjezi, M.R. Khalesi, D.J. Armaghani, Prediction and optimization of backbreak and rock fragmentation using an artificial neural network and a bee colony algorithm. Bull. Eng. Geol. Environ. 75(1), 27–36 (2016) 64. R.S. Faradonbeh, D.J. Armaghani, M. Monjezi, E.T. Mohamad, Genetic programming and gene expression programming for flyrock assessment due to mine blasting. Int. J. Rock Mech. Min. Sci. 88, 254–264 (2016)

58

3 Applications of AI and ML Techniques to Predict …

65. M. Hasanipanah, A. Shahnazar, H. Arab, S.B. Golzar, M. Amiri, Developing a new hybrid-AI model to predict blast-induced backbreak. Eng. Comput. 33(3) (2017) 66. M. Hasanipanah, H.B. Amnieh, Developing a new uncertain rule-based fuzzy approach for evaluating the blast-induced backbreak. Eng. Comput. (2020) 67. S. Kumar, A. Mishra, B. Choudhary, Prediction of back break in blasting using random decision trees. Eng. Comput. 1, 1–7 (2021) 68. J. Zhou, Y. Dai, M. Khandelwal, M. Monjezi, Z. Yu, Y. Qiu, Performance of hybrid SCARF and HHO-RF models for predicting backbreak in open-pit mine blasting operations. Nat. Resour. Res., 1–19 (2021) 69. A. Marto, M. Hajihassani, D. Jahed Armaghani, E. Tonnizam Mohamad, A.M. Makhtar, A novel approach for blast-induced flyrock prediction based on imperialist competitive algorithm and artificial neural network. Sci. World J. 2014 (2014) 70. H. Rad, M. Hasanipanah, M. Rezaei, A. Eghlim, Developing a least squares support vector machine for estimating the blast-induced flyrock. Eng. Comput. 34(4), 709–717 (2018) 71. H. Fattahi, M. Hasanipanah, An integrated approach of ANFIS-grasshopper optimization algorithm to approximate flyrock distance in mine blasting. Eng. Comput. 1, 1–13 (2021) 72. E.T. Mohamad, D.J. Armaghani, S.A. Noorani, R. Saad, S.V. Alvi, N.K. Abad, Prediction of flyrock in boulder blasting using artificial neural network. Electron. J. Geotech. Eng. 17, 2585–2595 (2012) 73. H. Amini, R. Gholami, M. Monjezi, S. Torabi, Evaluation of flyrock phenomenon due to blasting operation by support vector machine. Neural Comput. Appl. 21(8), 2077–2085 (2012) 74. P. Kalaivaani, T. Akila, M. Tahir, M. Ahmed, A. Surendar, A novel intelligent approach to simulate the blast-induced flyrock based on RFNN combined with PSO. Eng. Comput. 36(2), 435–442 (2019) 75. H. Guo, H. Nguyen, X.-N. Bui, D.J. Armaghani, A new technique to predict fly-rock in bench blasting based on an ensemble of support vector regression and GLMNET. Eng. Comput. 37, 421–435 (2021) 76. M. Monjezi, A. Bahrami, A. Varjani, Simultaneous prediction of fragmentation and flyrock in blasting operation using artificial neural networks. Int. J. Rock Mech. Min. Sci. 47(3), 476–480 (2010) 77. M. Monjezi, A. Bahrami, A.Y. Varjani, A.R. Sayadi, Prediction and controlling of flyrock in blasting operation using artificial neural network. Arab. J. Geosci. 4(3–4), 421–425 (2011) 78. H. Amini, R. Gholami, M. Monjezi, S.R. Torabi, J. Zadhesh, Evaluation of flyrock phenomenon due to blasting operation by support vector machine. Neural Comput. Appl. 21(8), 2077–2085 (2011) 79. M. Monjezi, A. Mehrdanesh, A. Malek, M. Khandelwal, Evaluation of effect of blast design parameters on flyrock using artificial neural networks. Neural Comput. Appl. 23(2), 349–356 (2013) 80. E. Ghasemi, H. Amini, M. Ataei, R. Khalokakaei, Application of artificial intelligence techniques for predicting the flyrock distance caused by blasting operation. Arab. J. Geosci. 7(1), 193–202 (2014) 81. D. Armaghani, E. Mohamad, M. Hajihassani, S. Abad, A. Marto, M. Moghaddam, Evaluation and prediction of flyrock resulting from blasting operations using empirical and computational methods. Eng. Comput. 32(1), 109–121 (2016) 82. R.S. Faradonbeh, D.J. Armaghani, M. Monjezi, Development of a new model for predicting flyrock distance in quarry blasting: a genetic programming technique. Bull. Eng. Geol. Environ. 75(3), 993–1006 (2016) 83. R.S. Faradonbeh, D.J. Armaghani, H.B. Amnieh, E.T. Mohamad, Prediction and minimization of blast-induced flyrock using gene expression programming and firefly algorithm. Neural Comput. Appl., 1–13 (2016) 84. M. Koopialipoor, A. Fallah, D.J. Armaghani, A. Azizi, E.T. Mohamad, Three hybrid intelligent models in estimating flyrock distance resulting from blasting. Eng. Comput. 35(1), 243–256 (2019)

References

59

85. H. Nguyen, X.-N. Bui, T. Nguyen-Thoi, P. Ragam, H. Moayedi, Toward a state-of-the-art of fly-rock prediction technology in open-pit mines using EANNs model. Appl. Sci. 9(21), 4554 (2019) 86. M. Wu, Q. Cai, T. Shang, Assessing the suitability of imperialist competitive algorithm for the predicting aims: an engineering case. Eng. Comput. 35(2), 627–636 (2019) 87. M. Hasanipanah, B. Keshtegar, D.-K. Thai, N.-T. Troung, An ANN-adaptive dynamical harmony search algorithm to approximate the flyrock resulting from blasting. Eng. Comput. (2020) 88. X. Lu, M. Hasanipanah, K. Brindhadevi, H.B. Amnieh, S. Khalafi, ORELM: a novel machine learning approach for prediction of flyrock in mine blasting. Nat. Resour. Res. 29(2), 641–654 (2020) 89. H. Rad, I. Bakhshayeshi, W. Jusoh, M. Tahir, L. Foong, Prediction of flyrock in mine blasting: a new computational intelligence approach. Nat. Resour. Res. 29(2), 609–623 (2019) 90. J. Zhou et al., Use of intelligent methods to design effective pattern parameters of mine blasting to minimize flyrock distance. Nat. Resour. Res. 29(2), 625–639 (2020) 91. H. Guo, H. Nguyen, X.-N. Bui, D.J. Armaghani, A new technique to predict fly-rock in bench blasting based on an ensemble of support vector regression and GLMNET. Eng. Comput. 37(1), 421–435 (2019) 92. M. Monjezi, H. Dehghani, J. Shakeri, A. Mehrdanesh, Optimization of prediction of flyrock using linear multivariate regression (LMR) and gene expression programming (GEP)—Topal Novin mine, Iran. Arab. J. Geosci. 14(15), 1–12 (2021) 93. H. Nguyen, X.N. Bui, Y. Choi, C.W. Lee, D.J. Armaghani, A novel combination of whale optimization algorithm and support vector machine with different kernel functions for prediction of blasting-induced fly-rock in quarry mines. Nat. Resour. Res. (2020) 94. J. Ye, M. Koopialipoor, J. Zhou, D.J. Armaghani, X. He, A novel combination of tree-based modeling and Monte Carlo simulation for assessing risk levels of flyrock induced by mine blasting. Nat. Resour. Res. 30(1), 225–243 (2021) 95. H. Dehghani, M. Pourzafar, M. Zadeh, Prediction and minimization of blast-induced flyrock using gene expression programming and cuckoo optimization algorithm. Environ. Earth Sci. 80(1), 1–17 (2021) 96. A. Richards, A. Moore, Flyrock control-by chance or design, in Proceedings of the Annual Conference on Explosives and Blasting Technique, vol. 1 (2004), pp. 335–348 97. T. Little, Flyrock risk, in Proceedings of EXPLO Conference, 3–4 September 2007 (2007), pp. 35–43

Chapter 4

Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques

4.1 Introduction Ground vibration and air overpressure (AOp) have been environmental concerns due to complaints by surrounding communities near blasting sites. Over the last couple of decades, development of various accessories such as ordinary detonators, millisecond electric detonators, cord relays, sequential blasting machine, Nonel DTH detonators are evolved in controlling these environmental effects. With improvement in accuracy of delay timings of electronic detonators as compared to DTH novel detonators or millisecond electric detonators, better control for ground vibration, and AOp has been reported by researchers [1, 2]. A large quantity of explosives is used for fragmentation in surface mines and civil engineering projects but only a part of explosive energy does useful work. On the other hand, most of the explosive’s energy is converted into ground vibration, AOp, dust, fumes, and flyrock [3]. Vibration due to blasting, which is created as a shock wave in the ground or rock mass and transmitted through the ground. It does not cause permanent deformation [4]. During hard ground excavation involving blasting operations, ground vibration effect on surrounding area, and structures is experienced. Ground vibration is caused by the explosives energy that has not been consumed in fracturing and moving the rock. The impact of ground vibration on the surrounding is critically determined by the properties of the rock mass. Duvall and Petkof [5] developed empirical equations for prediction of ground vibration based on site constants, distance, and maximum charge per delay. Many researchers have developed empirical equations based on these factors applicable for these sites. Air pressure created due to the moment of rock during blasting is known as AOp [6]. Various researchers have developed empirical equations to predict AOp based on maximum charge per delay and distance between monitoring point and blasting site. Prediction of ground vibration and AOp was not accurate enough with empirical equations. During last decades, several researchers have developed artificial intelligence (AI) and machine learning (ML) techniques for prediction of ground vibration © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2021 R. M. Bhatawdekar et al., Environmental Issues of Blasting, SpringerBriefs in Applied Sciences and Technology, https://doi.org/10.1007/978-981-16-8237-7_4

61

62

4 Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques

and AOp. All these prediction techniques have two parameters in common, namely maximum charge per delay and distance. It should be noted that these techniques have been successfully used by many researchers to solve engineering problems [7–14]. In this study, concepts of ground vibration and AOp will be discussed. Then, some of the commonly used AI and ML techniques in these areas will be described with enough detail. A review of these techniques will be given for prediction of ground vibration and AOp. Their effective parameters, accuracy levels, size of data will be discussed. Eventually, the best models will be suggested for both environmental issues of blasting.

4.2 Ground Vibration When an explosive blast takes place inside a borehole, an intense dynamic gas pressure created by the gases of explosion acts on the wall of borehole and is transmitted to rock mass surrounding the borehole as a strain wave. The strain wave propagates outwards in a radial direction along the rock mass. In the immediate surroundings of the borehole, permanent deformation occurs in the rock mass and failures occur under different mechanisms depending on the distance from the borehole—crushing, compressive failure, radial cracking, tensile failure, and reflection breakage near the free faces. Further, along the radial direction, the stress wave energy is dissipated and reduced to the extent that no further permanent deformations or fractures occur, but the stress waves continue to travel through the rock mass as elastic waves. These elastic waves cause the rock particles to oscillate conforming to viscoelastic behavior of rock. These oscillations are termed as ground vibrations. The waves travel in all directions around the borehole but given the fixed amount of energy so imparted to the rock, it gets dissipated exponentially over the distance from the borehole [15]. There are several inter-dependent factors that affect the intensity and frequency of ground vibration caused by blasting operations [16]. The most important and general factor that affects ground vibration due to blasting is the geo-mechanical condition of the surrounding rock [17]. This factor includes density, rock quality designation (RQD %), geological strength index, rock strength, rock characteristics (rock type, unit weight, layering, slope of layers, rock discontinuities, joints and their orientation, soil rock interface, etc.), and presence of water table. Geological discontinuities have also an important bearing on the blast performance, and hence, they must be carefully taken into account while planning and designing a blast. The presence of discontinuities plays an important role in the transmission of blast vibration [18] and the distance of monitoring equipment affects the measure of ground vibration—longer the distance, lower the measure due to attenuation and dissipation of ground vibration energy.

4.3 AOp

63

4.3 AOp AOp is the air pressure created due to blast explosions. These shock waves are caused by a combination of one to several factors: release of energy directly from the surface, a release of inadequately confined gases, and a shock from a large free face, gas release pulse due to escaping of gases through rock fractures and pulse from stemming column during ejection of stemming. AOp occurs when gas pressure from the explosion is vented into the atmosphere as the rock gets ruptures, by blow-out of stemming, by displacement of rock from the free face, by displacement of rock from the rock around the borehole, and by vibration of the ground in various permutations and combinations [19]. The impact of air blast at a given distance is a function of the distance from the point of blast and the cube root of the mass of explosive [20]. AOp is also known as “Blasting noise”. AOp is a measure of transient pressure changes above or below atmospheric pressure caused by acoustical waves traveling through air [21]. It is measured in Pascals (Pa) using microphones with weighting scales; and converted to sound pressure on a decibels (dB) scale. The major aspects of blast design that will control vibration and AOp are number of blast holes, max charge per delay, stemming length, and delay intervals. Factors which affect Aop are displayed in Fig. 4.1. In bench blasting at opencast mines, Richards [22] has demonstrated that the AOp would propagate at right angle to the bench face. On the other hand, prediction of actual intensity and nature of AOp are the most challenging and difficult. Vital factors influencing AOp include burden, spacing, stemming length, sub-grade drilling, hole diameter, and hole depth which are part of the blast design. For example, if the actual burden is less than the most desired optimum burden, rock is not fully

Fig. 4.1 Factors influencing AOp resulting from blasting

64

4 Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques

fragmented with underutilization of explosives energy resulting highly intense AOp wave associated with noise. With inadequate stemming, the explosives gases escape into the atmosphere without adequate resistance and result AOp. Konya and Walter [19] reported that controlling of AOp is possible to the maximum extent by replacing stemming material and increase in stemming length.

4.4 Background of Common AI Models 4.4.1 Artificial Neural Network (ANN) ANN models are developed based on simplest definition and building blocks equivalent to neurons of human brains, which represent sturdy and strong parallel processors. Multi-layer artificial neural networks are comparable with human brains. Certain coefficient connects between two neurons. During training, network leans through these connection points while information is delivered. The merits of ANN are as following: • Parallel processing: ANN has ability of parallel processing and may perform multiple job at once because of its numerical strength. • After the process of training, ANN may generate the output even with partial data. The generated output may have some performance degradation and this degradation is proportional to missing data. • ANN has ability to learn from the given input/data. Therefore, it produces output from learning only these data. In case, if an event is not shown to ANN during training, it cannot generate tolerance with respect to that event and produces false output when face that event. Meanwhile, the demerits of ANN are as following: • ANN doesn’t have a specific rule to determine the best structure for a particular scenario. It can only be achieved using heat and trial method or by experience. • ANN work as a black box because of those researchers only know about input and output of the model and does not have any clue about “how this output? And why this output?”. • ANN is hardware dependent because based on its structure; it requires different number of processors with property of parallel processing. • ANN can only work with specific type of data like numerical so, in order to apply ANN on any problem/event it has to be converted into that specific type and this conversion may directly influence the performance.

4.4 Background of Common AI Models

65

4.4.2 Support Vector Machine (SVM) Support vector machine (SVM) is developed based on the statistical learning theory using supervised machine. For classification and regression, SVM is found the most of the beneficial. A more detailed description can be found in various literatures [23, 24]. The merits of SVM are as following: • SVM is highly stable, as changes in dataset, may not greatly affect the generated hyperplane/output. • SVM has the ability to handle nonlinearity of the dataset as it uses kernel. • SVM has the ability to solve both regression and classification problem. As support vector regression (SVR) is used for regression and SVM is used for classification. • SVM uses L2 regularization because of that it has better generalization ability that prevent the over-fitting. In addition, performance of SVM is high, in case different classes of dataset is separated with clear margin. The demerits of SVM are as following: • In case training of SVM, if number of features is much greater than number of samples then SVM is prone for over fitting and performance of SVM may differ accordingly. • SVM is sensitive to noise as during training of SVM if training dataset may contain some noise/mislabeled data, the performance of the SVM decreases dramatically. • In case of SVM, in order to handle nonlinearity of the dataset, selection of kernel function is quite complex and tricky as it drastically affects the performance. • SVM requires very high amount of memory as it store all support vectors in the memory only and the size of support vector is directly proportional to the size of dataset. Therefore, SVM is not appropriate for large dataset.

4.4.3 Fuzzy Interface System (FIS) As per recommendation based on the study by Zadeh [25] shows that based on fuzzy set theory, powerful rule-based application, namely FIS is developed during the recent decades. Some of fuzzy systems are based on if-then rules, fuzzifiers, defuzzifier, and fuzzy inference engines and various details are available in the literature [26–29]. The merits of FIS are as following: • FIS is robust, as it does not require any precise input, it can accommodate different types of input that include imprecise, vague, and also distorted data. • FIS doesn’t require a large memory space as it can be coded using very few data. • The structure of FIS is quite simple and can be easily constructed as it has flexible system due to flexible system the used rules can be easily modified. • FIS has the ability to solve and take decision in complex problems even if it contains ambiguous data, as it is similar to human reasoning.

66

4 Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques

The demerits of FIS are as following: • In case of FIS all the rules are based on predefined rules and if the predefined rules may contain some flawed then the generated output is not acceptable. • As FIS is basically works on imprecise inputs and data so, in some cases accuracy of the system may compromised. • FIS is based on human reasoning so there were no any systematic and particular approach of solving any scenario/problem. Due to that, many solutions may arise. Due to these many solutions, there is a chance of ambiguity. • FIS is completely dependent on expertise and knowledge of human. So, it contains inaccuracy and requires large amount of testing, verification, and validation.

4.5 Ground Vibration Prediction Using AI Techniques Table 4.1 shows the parameters used by various researchers for prediction of ground vibration due to blasting. All researchers have used charge relevant factor and distance as input parameters for prediction of ground vibration which show that these parameters are major contributors’ ground vibration. Some of the researchers have used empirical equations for comparing results with AI techniques [30–32]. Number of datasets varies from 68 to 120. Coefficient of determination (R2 ) varies from 0.545 to 0.781. Several researchers have used ANN technique for prediction of ground vibration [32–45]. Number of data sets varies from 20 to 269. R2 varies from 0.781 to 0.98. Few researchers have deployed FIS as an AI technique for prediction of ground vibration. Number of datasets varies from 26 to 120 and R2 varies from 0.92 to 0.95. Many researchers have applied SVM/SVR as AI technique for prediction of ground vibration [27, 35, 39, 43, 46, 47]. Number of datasets varies from 44 to 88. R2 varies from 0.88 to 0.98. Various researchers have applied hybrid AI techniques with ANN with optimization algorithms such as ANN-particle swarm optimization, firefly algorithm-ANN, and genetic algorithm-ANN. Number of datasets varies from 44 to 88. R2 varies from 0.88 to 0.98. In addition, input variables to predict ground vibration are divided into different categories in Table 4.1. This classification can help researchers to select the most effective factors with the highest influence on ground vibration.

4.6 AOp Prediction Using AI Techniques All researchers with empirical or AI techniques necessarily used maximum charge per delay and distance between blast and monitoring station as input parameters in estimating AOp. Table 4.2 shows the parameters used by various researchers for prediction of AOp due to blasting together with their techniques, accuracy level, and size of data. Several researchers used ANN for prediction of AOp [32, 59– 64]. Number of datasets varied from 62 to 166 and R2 varied from 0.84 to 0.961.

4.6 AOp Prediction Using AI Techniques

67

Table 4.1 Input parameters, size of data and AI techniques for prediction of ground vibration References

Technique

Input parameters

R2

Explosives

Other

No. of datasets

E, BI, B, S, Vp HD

C, D, VOD

–

154

0.98

Monjezi et al. [37] ANN

–

BS

C, D, N

–

269

0.95

Fi¸sne et al. [38]

–

–

C, D

–

33

0.92

Monjezi et al. [48] ANN

–

HD

C, D

–

162

0.949

Mohammadnejad et al. [39]

SVM

–

–

C, D

–

37

0.89

Armaghani et al. [16]

ANN-PSO

RD

B, S, HD, SD

C, D, N

–

44

0.94

Ghasemi et al. [49]

FIS

–

B, S, ST

C, D

–

120

0.95

Monjezi et al. [40] ANN

–

–

C, D

–

20

0.93

Hajihassani et al. [50]

ANN-ICA

E, Vp

ST, BS C, D

–

95

0.98

Armaghani et al. [41]

ANN

–

–

–

95

0.856

Ghoraba et al. [42]

ANN

–

HD, ST

Hajihassani et al. [51]

ANN-PSO

RQD

Amiri et al. [43]

ANN

Rock mass Khandelwal and Singh [36]

ANN

FIS

Blast design

ANN

0.85

C, D

0.878 C, D

–

115

0.98

ST, BS, PF, C, D SD

–

88

-

–

–

C, D

–

75

0.82

–

B, S, ST

PF, C, D

–

101

0.885

–

B, S, ST

C, PF

–

72

0.98

–

–

C

D, RD

70

0.988

–

–

C

D

152

0.977

ANN-KNN Bhatawdekar et al. [44]

ANN

Mojtahedi et al. [52]

ANFIS-FA

Azimi et al. [53]

GA-ANN

0.88

LMR LMR

0.804 0.669

ANFIS Bui et al. [27]

PSO-KNN

0.92

SVR

Jiang et al. [54]

0.944

RF

0.952

Empirical

0.579

ANFIS

–

–

C

D

90

0.983

–

–

C, D

–

68

0.955

LMR Nguyen et al. [45] ANN

0.876 (continued)

68

4 Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques

Table 4.1 (continued) References

Technique

Nguyen et al. [55] ANN

Input parameters Rock mass

Blast design

Explosives

Other

No. of datasets

–

–

C, D

–

68

0.886

CART

0.618

Nguyen et al. [29] ABC-SVR-P

0.737 –

B, S

C

D

125

0.98

ICA-GA-RBF

0.987

GA-SVR-RBF

0.991

Empirical

–

–

C

D

181

QRNN

–

B, S, ST, HD, H

C, PF

D

83

ANN

0.937 0.95

–

PSO-XGBoost

B, S, ST

C, PF

D

136

XGBoost

0.965

GBM

0.935

ANN

0.945

ICA-M5Rules

0.965 –

B, S

C

D

125

M5Rules

Huang et al. [28]

0.995 0.971

RF

0.988

SVM

0.983

FA-ANN

–

SD

C, PF

D

88

0.909

ABC-ANN

0.909

ICA-ANN

0.885

PSO-ANN

0.874

GA-ANN Yang et al. [30]

0.988 0.977

SVR Fang et al. [47]

0.952 0.961

RF ICA-XGBoost

0.643 0.924

FCM-QRNN

Ding et al. [35]

0.989

PSO-SVR-P

SVR-RBF Bui et al. [34]

0.981

SVM KNN

Nguyen [31]

R2

Empirical ANFIS ANFIS-PSO ANFIS-GA

0.865 –

B, S, ST

C, PF

D

86

0.874 0.884 0.966 0.979 (continued)

4.6 AOp Prediction Using AI Techniques

69

Table 4.1 (continued) References

Yang et al. [46]

Technique

SVR

Input parameters Rock mass

Blast design

Explosives

Other

No. of datasets

–

B, S, ST

C

D

90

FA-ANN

Zhang et al. [33]

R2

0.969 0.946

FA-SVR

0.992

PSO-SVR

0.979

GA-SVR

0.971

RF

–

ST, HD C, PF

D

102

0.83

CART

0.56

CHAID

0.68

ANN

0.84

SVM

0.85

Zhang et al. [56]

PSO-XG Boost –

B, S

C, PF

D

175

0.968

Zhou et al. [57]

RF

–

ST, HD C, PF

D

102

0.903

–

BS, ST –

D, Vp

–

BN Chen et al. [58]

FA-SVR

0.87 0.984

PSO-SVR

0.977

FA-SVR

0.964

GA-SVR

0.957

PSO-ANN

0.924

FA-ANN

0.925

GA-ANN

0.936

SVM support vector machine, PSO particle swarm optimization, ICA imperialist competitive algorithm, XG Boost extreme gradient boosting, RF random forest, SVR support vector regression, FCM fuzzy c-means clustering algorithm, KNN hierachical K-means clustering and cubist algorithm, B burden, S spacing, ST stemming length, HD hole depth, BS burden spacing ratio, E Young’s modulus, RD rock density, υ Poison’s ratio, BI blastability index, V p p-wave velocity, RQD rock quality designation, C maximum charge per delay, D distance between monitoring point and blasting face, PF powder factor, VOD velocity of detonation, N number of rows, SD subdrilling, LMR linear multiple regression, ANFIS adaptive neuro-fuzzy inference system, CART classification and regression tree, RBF radius basis function, ABC artificial bee colony, QRNN quantile regression neural network, CHAID chi-squared automatic interaction detection, BN Bayesian network

Some of the researchers utilized SVM for prediction of AOp. Number of data sets varied from 62 to 180. There was wide variation in R2 values from 0.13 to 0.855. ANFIS model was used to predict AOp [60, 65] and number datasets and R2 were 128, 62 and 0.855, 0.62 respectively. Various hybrid models with ANN and other optimization algorithms were developed by various researchers such as PSO-ANN,

70

4 Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques

Table 4.2 Input parameters, size of data, and AI techniques for prediction of AOp References

Technique

Input parameters Rock mass

Blast design

Explosives

Other

No. of datasets

R2

Khandelwal and Singh [59]

ANN

–

–

C, D

–

56

0.96

Mohamed [32]

ANN

–

–

C, D

–

162

0.92

Khandelwal and Kankar [69]

SVM

–

–

C, D

–

75

0.855

Mohamad et al. [70]

ANN

–

B, S, d, HD, ST

PF, N

–

38

0.93

Hajihassani et al. [51]

ANN-PSO

RQD

B, S, HD, PF, C, N, D ST

–

62

0.86

Armaghani et al. [60]

ANN-ANFIS

–

ST, B/S

–

128

0.92

Hajihassani et al. [51]

ANN-PSO

RQD

HD, ST, B/S, SB

PF, C, N, D

–

88

0.89

Armaghani et al. [66]

ICA-ANN

–

–

C, D

–

70

0.83

Mohamad et al. [67]

GA-ANN

–

–

C, D

–

76

–

Bui et al. [62]

ANN

–

B, S, HD, PF, D d,

–

113

FIS

0.86

PF, C, D

LMR

0.855

BART

0.961 0.945

BRT

0.898

SVR

0.898

GP

0.949

KNN

0.89

Gao et al. [71]

ANN

Bui et al. [72]

Empirical

Nguyen and Bui [63]

Empirical

RMR

–

C, D

–

85

0.91

–

B, ST

C, PF

D

108

0.838

TLBM-ANN

0.935

GLMNET ANN

0.975 –

B, S, ST

C, PF

D, VOD 114

0.429 0.966 (continued)

4.6 AOp Prediction Using AI Techniques

71

Table 4.2 (continued) References

Technique

Input parameters Rock mass

Blast design

Explosives

Other

No. of datasets

R2

RF

0.939

ANN-RF

0.985

Gao et al. [73]

GMDH–GA

RMR

–

C, PF

D

84

0.988

Murlidhar et al. [74]

GeP

–

ST, JA

C, PF, BI

D

125

0.8621

Nguyen and Bui [61]

M5’

0.7451

MLR

0.7883

Empirical

–

–

C

CART KNN

D, RH, AP, WS, WD, T

121

0.941

ANN

0.957

BRR

0.927

SVR

0.956

GP Nguyen et al. [75]

ANN

0.949 –

BRNN

B, S, ST, H, N

C, PF

B, S, ST, H, N

C, PF

B, S, ST

C, PF

D, RH

146

Empirical

0.816 –

RF

D, RH

146

Temeng et al. [78]

IR

0.871 0.968

GBM Nguyen et al. [77]

0.961 0.936

HYFIS Nguyen et al. [76]

0.466 0.949

0.97 –

D

77

MLP

0.987 0.99

RF

0.978

M5-Rules

0.992

BI-ENN

–

ST, NH

C

D

171

0.824

BPNN

0.8172

SVM

0.8134

GMDH

0.5878

GenP

0.7196

McKenzine

0.7208

Ye et al. [68] ANFIS ANFIS-GA ANFIS-PSO

RQD

B, S, ST, NH, HD

C, PF

D

62

0.873 0.935 0.965 (continued)

72

4 Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques

Table 4.2 (continued) References

Technique

Input parameters Rock mass

Blast design

Explosives

Other

–

–

C, PF

D

No. of datasets

ANFIS-SFS Zhou et al. [79]

FS-FA

R2

0.986 86

0.977

BRR Bayesian ridge regression, GP gaussian process, BART Bayesian additive regression trees, BRT boosted regression trees, GMDH group method of data handling, BRNN Bayesian regularized neural networks, HYFIS hybrid neural fuzzy inference system, GBM gradient boosting machine, M5’ regression tree, GeP genetic programming, SFS stochastic fractal search algorithm, FS fuzzy system, FA firefy algorithm, BI-ENN brain-inspired emotional neural network, BPNN backpropagation neural network, GenP generalized predictor, MF McKenzie formula, NAAS National Association of Australian State, BA bat algorithm, XGBoost extreme gradient boosting tree, GLMNET the lasso and elastic-net regularized generalized linear model, LM Levenberg–Marquardt algorithm, RH relative humidity, AP atmospheric pressure, WS wind speed, WD wind direction, T temperature, RMR rock mass rating, VD vertical monitoring distance from the blast face, JA joint aperture, NH number of holes, HD hole depth, TLBO teaching-learning-based optimization

GA-ANN, and ICA-ANN [51, 66, 67]. Number of datasets varied from 62 to 70 with variation in R2 from 0.7219 to 0.935. Several hybrid models with ANFIS and other optimization algorithms were developed by several researchers such as ANFIS-GA, and ANFIS-PSO [65, 68] with R2 variation from 0.92 to 0.986.

4.7 Discussion Environmental issues such as ground vibration and AOp are challenging which local site management has to deal with due to various complaints by local communities. For the last 60 years, various empirical equations have been developed for prediction of these parameters mainly based on maximum charge delay, distance between blasting site and monitoring station. After empirical equations were developed, various statistical models were applied by researchers, which has disadvantage of lower accuracy. During the last decade, many AI techniques have been proposed for prediction of ground vibration and AOp. Several researchers have utilized ANN for prediction and the same can be considered to benchmark with other techniques to compare the performance. Many researchers applied FIS and SVR/SVM techniques for prediction of ground vibration which have reasonably good accuracy. SVM model showed large variation in the prediction accuracy of AOp as compared to other models. During the last five years, researchers have developed many hybrid models with ANN, ANFIS which have better accuracy compared to the base models (e.g., ANN and ANFIS). Researchers have also predicted AOp using 3 different models such as ANFIS-PNNGA [65]. It is observed that 100 blasting data sets are required for prediction of AOp

4.7 Discussion

73

and ground vibration with AI techniques. Initially various proven AI techniques can be applied for prediction of ground vibration and AOp. Input parameters for prediction of ground vibration and AOp are divided into three groups—blast design parameters, explosives-related parameters, and other parameters. Burden and spacing have to be optimum to minimize ground vibration and AOp. Hole diameter and bench height have direct impact on quantum of explosives and thus there is direct impact on ground vibration and AOp. Stemming length can be used for control of AOp. Maximum charge per delay represents maximum explosives energy for given instant and has direct impact on AOp and PPV. Delay timing and number rows also play important role in AOp and PPV. Powder factor is based on total explosives energy release per unit of quantity of blasted rock. Some of researchers replace this factor by specific charge, i.e. explosives charge per unit volume of rock to be blasted.

4.8 Conclusion 1.

2.

3. 4. 5.

6. 7. 8.

9.

Environmental issues of ground vibration and AOp—actual measurement, monitoring, and prediction have become important not only to meet local regulation but also to avoid or minimize complaints from local communities. With the improvement in data collection technology and recording by seismographs, large database can be developed for a given site which will be useful for future predictions of ground vibration and AOp. During the last decades various AI techniques have been developed are far superior as compared to empirical equations developed during the last century. ANN technique can be considered as benchmark AI techniques for comparing with other AI techniques. SVM/SVR showed good prediction accuracy of ground vibration. On the other hand, researchers reported much lower accuracy and variation of R2 for prediction of AOp. FIS/ANFIS models also showed good prediction for ground vibration and AOp. Hybrid models with ANN or ANFIS with other algorithms showed better accuracy for prediction of ground vibration and AOp. There is need for future research by developing larger database, impact of various rock mass properties, geological structures, joint properties such as spacing, inclination, etc. Theory-guided ML or physics-based ML can be considered as a new direction of research in civil and mining. These techniques with considering physics rules between input and output variables will give a better understanding and view to a civil or mining engineer.

74

4 Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques

References 1. B.R. Murlidhar, D.J. Armaghani, E.T. Mohamad, Intelligence prediction of some selected environmental issues of blasting: a review. Open Constr. Build. Technol. J. 14(1), 298–308 (2020) 2. H. Grobler, Using electronic detonators to improve all-round blasting performances. Fragblast 7(1), 1–12 (2003) 3. M. Khandelwal, T.N. Singh, Prediction of blast induced vibrations and frequency in opencast mine: a neural network approach. J. Sound Vib. 289, 711–772 (2006) 4. M. Khandelwal, T. Singh, Evaluation of blast-induced ground vibration predictors. Soil Dyn. Earthq. Eng. 27(2), 116–125 (2007) 5. W.I. Duvall, B. Petkof, Spherical Propagation of Explosion-Generated Strain Pulses in Rock, no. 5481–5485 (US Department of the Interior, Bureau of Mines, 1959) 6. P.K. Singh, A. Sinha, Rock Fragmentation by Blasting: Fragblast 10 (CRC Press, 2013) 7. E.T. Mohamad, D. Li, B.R. Murlidhar, D.J. Armaghani, K.A. Kassim, I. Komoo, The effects of ABC, ICA, and PSO optimization techniques on prediction of ripping production. Eng. Comput. (2019). https://doi.org/10.1007/s00366-019-00770-9 8. Z. He, D.J. Armaghani, M. Masoumnezhad, M. Khandelwal, J. Zhou, B.R. Murlidhar, A combination of expert-based system and advanced decision-tree algorithms to predict air-overpressure resulting from quarry blasting. Nat. Resour. Res. 30(2), 1889–1903 (2021) 9. B.R. Murlidhar, D.J. Armaghani, E.T. Mohamad, S. Changthan, Rock fragmentation prediction through a new hybrid model based on imperial competitive algorithm and neural network. Smart Constr. Res. 2(3), 1–12 (2018) 10. H. Naderpour, D. Rezazadeh Eidgahee, P. Fakharian, A.H. Rafiean, S.M. Kalantari, A new proposed approach for moment capacity estimation of ferrocement members using Group Method of Data Handling. Eng. Sci. Technol. an. Int. J. 23, 382–391 (2019). https://doi.org/ 10.1016/j.jestch.2019.05.013 11. E. Ford, K. Maneparambil, N. Neithalath, Machine learning on microstructural chemical maps to classify component phases in cement pastes. J. Soft Comput. Civ. Eng. 5(4), 1–20 (2021). https://doi.org/10.22115/SCCE.2021.302400.1357 12. D. Jahed Armaghani, A. Azizi, A comparative study of artificial intelligence techniques to estimate TBM performance in various weathering zones, in Applications of Artificial Intelligence in Tunnelling and Underground Space Technology. SpringerBriefs in Applied Sciences and Technology (Springer, Singapore, 2021), pp. 55–70. https://doi.org/10.1007/978-981-16-103 4-9_4 13. A. Saber, Effects of window-to-wall ratio on energy consumption: application of numerical and ann approaches. J. Soft Comput. Civ. Eng. 5(4), 41–56 (2021). https://doi.org/10.22115/ SCCE.2021.281977.1299 14. A. Khademi, K. Behfarnia, T. Kalman Šipoš, I. Miliˇcevi´c, I. The use of machine learning models in estimating the compressive strength of recycled brick aggregate concrete. Comput. Eng. Phys. Model. 4(4), 1–25 (2021). https://doi.org/10.22115/cepm.2021.297016.1181 15. Y.-B. Yang, H.-H. Hung, A parametric study of wave barriers for reduction of train-induced vibrations. Int. J. Numer. Methods Eng. 40(20), 3729–3747 (1997) 16. D.J. Armaghani, M. Hajihassani, E.T. Mohamad, A. Marto, S.A. Noorani, Blasting-induced flyrock and ground vibration prediction through an expert artificial neural network based on particle swarm optimization. Arab. J. Geosci. 7(12), 5383–5396 (2014) 17. R. Kumar, D. Choudhury, K. Bhargava, Determination of blast-induced ground vibration equations for rocks using mechanical and geological properties. J. Rock Mech. Geotech. Eng. 8(3) (2016) 18. D. Singh, V. Sastry, Influence of structural discontinuity on rock fragmentation by blasting, in Proceedings of the 6th International Symposium on Intense Dynamic Loading and Its Effects 3–7 June (1986) 19. C.J. Konya, E.J. Walter, Surface Blast Design (Prentice Hall, Englewood Cliffs, 1990)

References

75

20. P.-A. Persson, R. Holmberg, J. Lee, Rock Blasting and Explosives Engineering (CRC Press, Boca Raton, 1993) 21. C. Kuzu, A. Fisne, S.G. Ercelebi, Operational and geological parameters in the assessing blast induced airblast-overpressure in quarries. Appl. Acoust. 70(3), 404–411 (2009) 22. A. Richards, Elliptical airblast overpressure model. Min. Technol. 119(4), 205–211 (2010) 23. A.K. Rahul, N. Shivhare, S. Kumar, S.B. Dwivedi, P.K.S. Dikshit, Modelling of daily suspended sediment concentration using FFBPNN and SVM algorithms. J. Soft Comput. Civ. Eng. 5(2), 120–134 (2021). https://doi.org/10.22115/SCCE.2021.283137.1305 24. M.G. Meharie, N. Shaik, Predicting highway construction costs: comparison of the performance of random forest, neural network and support vector machine models. J. Soft Comput. Civ. Eng. 4(2),103–112 (2020). https://doi.org/10.22115/SCCE.2020.226883.1205 25. L.A. Zadeh, Fuzzy sets. Inf. Control 8(3), 338–353 (1965) 26. H. Nguyen, X.-N. Bui, H.-B. Bui, D.T. Cuong, Developing an XGBoost model to predict blastinduced peak particle velocity in an open-pit mine: a case study. Acta Geophys. 67(2), 477–490 (2019) 27. X.-N. Bui, P. Jaroonpattanapong, H. Nguyen, Q.-H. Tran, N.Q. Long, A novel hybrid model for predicting blast-induced ground vibration based on k-nearest neighbors and particle Swarm optimization. Sci. Rep. 9(1), 1–14 (2019) 28. J. Huang, M. Koopialipoor, D.J. Armaghani, A combination of fuzzy Delphi method and hybrid ANN-based systems to forecast ground vibration resulting from blasting. Sci. Rep. 10(1), 1–21 (2020) 29. H. Nguyen, Y. Choi, X.-N. Bui, T. Nguyen-Thoi, Predicting blast-induced ground vibration in open-pit mines using vibration sensors and support vector regression-based optimization algorithms. Sensors 20(1), 132 (2020) 30. H. Yang, M. Hasanipanah, M. Tahir, D.T. Bui, Intelligent prediction of blasting-induced ground vibration using ANFIS optimized by GA and PSO. Nat. Resour. Res. 29(2), 739–750 (2020) 31. H. Nguyen, Support vector regression approach with different kernel functions for predicting blast-induced ground vibration: a case study in an open-pit coal mine of Vietnam. SN Appl. Sci. 1(4), 283 (2019) 32. M.T. Mohamed, Performance of fuzzy logic and artificial neural network in prediction of ground and air vibrations. Int. J. Rock Mech. Min. Sci. 48(5), 845 (2011) 33. H. Zhang, J. Zhou, D.J. Armaghani, M.M. Tahir, B.T. Pham, V. Van Huynh, A combination of feature selection and random forest techniques to solve a problem related to blast-induced ground vibration. Appl. Sci. 10(3), 869 (2020) 34. X.-N. Bui et al., Prediction of blast-induced ground vibration intensity in open-pit mines using unmanned aerial vehicle and a novel intelligence system. Nat. Resour. Res. 29(2), 771–790 (2020) 35. Z. Ding, H. Nguyen, X.-N. Bui, J. Zhou, H. Moayedi, Computational intelligence model for estimating intensity of blast-induced ground vibration in a mine based on imperialist competitive and extreme gradient boosting algorithms. Nat. Resour. Res. 29(2), 751–769 (2020) 36. M. Khandelwal, T. Singh, Prediction of blast-induced ground vibration using artificial neural network. Int. J. Rock Mech. Min. Sci. 46(7), 1214–1222 (2009) 37. M. Monjezi, A. Bahrami, A. Varjani, Simultaneous prediction of fragmentation and flyrock in blasting operation using artificial neural networks. Int. J. Rock Mech. Min. Sci. 47(3), 476–480 (2010) 38. A. Fi¸sne, C. Kuzu, T. Hüdaverdi, Prediction of environmental impacts of quarry blasting operation using fuzzy logic. Environ. Monit. Assess. 174(1–4), 461–470 (2011) 39. M. Mohammadnejad, R. Gholami, A. Ramezanzadeh, M. Jalali, Prediction of blast-induced vibrations in limestone quarries using Support Vector Machine. J. Vib. Control 18(9), 1322– 1329 (2011) 40. M. Monjezi, A. Mehrdanesh, A. Malek, M. Khandelwal, Evaluation of effect of blast design parameters on flyrock using artificial neural networks. Neural Comput. Appl. 23(2), 349–356 (2013)

76

4 Blast-Induced Air and Ground Vibrations: A Review of Soft Computing Techniques

41. D. Jahed Armaghani, M. Hajihassani, A. Marto, R. Shirani Faradonbeh, E.T. Mohamad, Prediction of blast-induced air overpressure: a hybrid AI-based predictive model. Environ. Monit. Assess. 187(11) (2015) 42. S. Ghoraba, M. Monjezi, N. Talebi, M.R. Moghadam, D. Jahed Armaghani, Prediction of ground vibration caused by blasting operations through a neural network approach: a case study of Gol-E-Gohar iron mine. Iran. J. Zhejiang Univ. Sci. A. Doi. 10, 1631 (2015) 43. M. Amiri, H.B. Amnieh, M. Hasanipanah, L.M. Khanli, A new combination of artificial neural network and K-nearest neighbors models to predict blast-induced ground vibration and airoverpressure. Eng. Comput. 32(4), 631–644 (2016) 44. R. Bhatawdekar, P. Sharma, L. Sarma, A. Singh, T. Singh, T. Edy, Prediction of ground vibration and frequency due to blasting, using artificial neural network at a limestone quarry, in Proceedings of National Cement Building Material, Seminar (2017) 45. H. Nguyen, X.-N. Bui, Q.-H. Tran, T.-Q. Le, N.-H. Do, Evaluating and predicting blast-induced ground vibration in open-cast mine using ANN: a case study in Vietnam. SN Appl. Sci. 1(1), 125 (2019) 46. H. Yang, H.N. Rad, M. Hasanipanah, H.B. Amnieh, A. Nekouie, Prediction of vibration velocity generated in mine blasting using support vector regression improved by optimization algorithms. Nat. Resour. Res. 29(2), 807–830 (2020) 47. Q. Fang, H. Nguyen, X.-N. Bui, T. Nguyen-Thoi, Prediction of blast-induced ground vibration in open-pit mines using a new technique based on imperialist competitive algorithm and M5Rules. Nat. Resour. Res. 29(2), 791–806 (2020) 48. M. Monjezi, M. Ghafurikalajahi, A. Bahrami, Prediction of blast-induced ground vibration using artificial neural networks. Tunn. Undergr. Sp. Technol. 26(1), 46–50 (2011) 49. E. Ghasemi, M. Ataei, H. Hashemolhosseini, Development of a fuzzy model for predicting ground vibration caused by rock blasting in surface mining. J. Vib. Control 19(5), 755–770 (2013) 50. M. Hajihassani, D. Jahed Armaghani, A. Marto, E. Tonnizam Mohamad, Ground vibration prediction in quarry blasting through an artificial neural network optimized by imperialist competitive algorithm. Bull. Eng. Geol. Environ. 74, 873–886 (2014) 51. M. Hajihassani, D. Jahed Armaghani, M. Monjezi, E.T. Mohamad, A. Marto, Blast-induced air and ground vibration prediction: a particle swarm optimization-based artificial neural network approach. Environ. Earth Sci. 74(4), 2799–2817 (2015) 52. S. Mojtahedi, I. Ebtehaj, M. Hasanipanah, H. Bonakdari, H. Amnieh, Proposing a novel hybrid intelligent model for the simulation of particle size distribution resulting from blasting. Eng. Comput. 35(1), 47–56 (2018) 53. Y. Azimi, S. H. Khoshrou, M. Osanloo, Prediction of blast induced ground vibration (BIGV) of quarry mining using hybrid genetic algorithm optimized artificial neural network. Measurement 147, 106874 (2019) 54. W. Jiang, C.A. Arslan, M.S. Tehrani, M. Khorami, M. Hasanipanah, Simulating the peak particle velocity in rock blasting projects using a neuro-fuzzy inference system. Eng. Comput. 35(4), 1203–1211 (2019) 55. H. Nguyen, X.-N. Bui, H. Moayedi, A comparison of advanced computational models and experimental techniques in predicting blast-induced ground vibration in open-pit coal mine. Acta Geophys. 67(4), 1025–1037 (2019) 56. X. Zhang et al., Novel soft computing model for predicting blast-induced ground vibration in open-pit mines based on particle swarm optimization and XGBoost. Nat. Resour. Res. 29(2), 711–721 (2020) 57. J. Zhou, P.G. Asteris, D.J. Armaghani, B.T. Pham, Prediction of ground vibration induced by blasting operations through the use of the Bayesian network and random forest models. Soil Dyn. Earthq. Eng. 139(Aug), 106390 (2020) 58. W. Chen, M. Hasanipanah, H. Nikafshan Rad, D. Jahed Armaghani, M. Tahir, A new design of evolutionary hybrid optimization of SVR model in predicting the blast-induced ground vibration. Eng. Comput. 37(2), 1455–1471 (2021)

References

77

59. M. Khandelwal, T. Singh, Prediction of blast induced air overpressure in opencast mine. Noise Vib. Worldw. 36(2), 7–16 (2005) 60. D.J. Armaghani et al., Neuro-fuzzy technique to predict air-overpressure induced by blasting. Arab. J. Geosci. 8(12), 10937–10950 (2015) 61. H. Nguyen, X.-N. Bui, Soft computing models for predicting blast-induced air over-pressure: A novel artificial intelligence approach, Appl. Soft Comput., 106292 (2020) 62. X.-N. Bui, H. Nguyen, H.-A. Le, H.-B. Bui, N.-H. Do, Prediction of blast-induced air overpressure in open-pit mine: assessment of different artificial intelligence techniques. Nat. Resour. Res. 29(2), 571–591 (2020) 63. H. Nguyen, X.-N. Bui, Predicting blast-induced air overpressure: a robust artificial intelligence system based on artificial neural networks and random forest. Nat. Resour. Res. 28(3), 893–907 (2019) 64. X. Zhou, D.J. Armaghani, J. Ye, M. Khari, M.R. Motahari, Hybridization of parametric and non-parametric techniques to predict air over-pressure induced by quarry blasting. Nat. Resour. Res. 30(1), 209–224 (2021) 65. H. Harandizadeh, D.J. Armaghani, Prediction of air-overpressure induced by blasting using an ANFIS-PNN model optimized by GA. Appl. Soft Comput., 106904 (2020) 66. D. Armaghani, M. Hasanipanah, E. Mohamad, A combination of the ICA-ANN model to predict air-overpressure resulting from blasting. Eng. Comput. 32, 155–171 (2016) 67. E. Tonnizam Mohamad, D. Jahed Armaghani, M. Hasanipanah, B.R. Murlidhar, M.N.A. Alel, Estimation of air-overpressure produced by blasting operation through a neuro-genetic technique. Environ. Earth Sci. 75(2), 1–15 (2016) 68. J. Ye, J. Dalle, R. Nezami, M. Hasanipanah, D.J. Armaghani, Stochastic fractal search-tuned ANFIS model to predict blast-induced air overpressure. Eng. Comput. (2020). https://doi.org/ 10.1007/s00366-020-01085-w 69. M. Khandelwal, P. Kankar, Prediction of blast-induced air overpressure using support vector machine. Arab. J. Geosci. 4, 427–433 (2011) 70. E. Tonnizam Mohamad, M. Hajihassani, D. Jahed Armaghani, A. Marto, Simulation of blastinginduced air overpressure by means of artificial neural networks. Int. Rev. Model. Simulations 5(6) (2012) 71. W. Gao, M. Karbasi, A.M. Derakhsh, A. Jalili, Development of a novel soft-computing framework for the simulation aims: a case study. Eng. Comput. 35(1), 315–322 (2018) 72. X.-N. Bui et al., A lasso and elastic-net regularized generalized linear model for predicting blast-induced air over-pressure in open-pit mines. In˙zynieria Mineralna 21 (2019) 73. W. Gao, A.S. Alqahtani, A. Mubarakali, D. Mavaluru, S. khalafi, Developing an innovative soft computing scheme for prediction of air overpressure resulting from mine blasting using GMDH optimized by GA. Eng. Comput. 36(2), 647–654 (2020) 74. B. Murlidhar, B. Bejarbaneh, D. Armaghani, A. Mohammed, E. Mohamad, Application of treebased predictive models to forecast air overpressure induced by mine blasting. Nat. Resour. Res. (2020). https://doi.org/10.1007/s11053-020-09770-9 75. H. Nguyen, X.-N. Bui, H.-B. Bui, N.-L. Mai, A comparative study of artificial neural networks in predicting blast-induced air-blast overpressure at Deo Nai open-pit coal mine, Vietnam. Neural Comput. Appl. 32(8), 3939–3955 (2020) 76. H. Nguyen et al., “A comparative study of empirical and ensemble machine learning algorithms in predicting air over-pressure in open-pit coal mine. Acta Geophys. (2020). https://doi.org/10. 1007/s11600-019-00396-x 77. H. Nguyen et al., A comparative study of different artificial intelligence techniques in predicting blast-induced air over-pressure. 1(2) (Techno-Press, 2020) 78. V.A. Temeng, Y.Y. Ziggah, C.K. Arthur, A novel artificial intelligent model for predicting air overpressure using brain inspired emotional neural network. Int. J. Min. Sci. Technol. (2020). https://doi.org/10.1016/j.ijmst.2020.05.020 79. J. Zhou, A. Nekouie, C.A. Arslan, B.T. Pham, M. Hasanipanah, Novel approach for forecasting the blast-induced AOp using a hybrid fuzzy system and firefly algorithm. Eng. Comput. 36(2), 703–712 (2020)