Fault Diagnosis of Hybrid Dynamic and Complex Systems (eBook)

Moamar Sayed-Mouchaweh received his PhD from the University of Reims-France. He was working as Associated Professor in Computer Science, Control and Signal processing at the University of Reims-France in the Research Center in Sciences and Technology of the Information and the Communication. In December 2008, he obtained the Habilitation to Direct Researches (HDR) in Computer science, Control and Signal processing. Since September 2011, he is working as a Full Professor in the High National Engineering School of Mines Department of Computer Science and Automatic Control. He edited the Springer book ‘Learning in Non-Stationary Environments: Methods and Applications‘, in April 2012 and wrote two SpringerBriefs ‘Discrete Event Systems: Diagnosis and Diagnosability’, and ‘Learning from Data Streams in Dynamic Environments’. He was a guest editor of several special issues of international journals. He was IPC Chair of conference Chair of several international workshops and conferences (IEEE International Conference on Machine Learning and Applications IEEE International Conference on Evolving and Adaptive Intelligent Systems). He is working as a member of the Editorial Board of Elsevier Journal “Applied Soft Computing” and Springer Journals “Evolving systems” and “Intelligent Industrial Systems”.

Preface 5
Contents 7
1 Prologue 9
1.1 Hybrid Dynamic Systems: Definition, Classes, and Modeling Tools 9
1.2 Fault Diagnosis of Hybrid Dynamic Systems: Problem Formulation, Methods, and Challenges 11
1.3 Contents of the Book 15
1.3.1 Chapter 2 15
1.3.2 Chapter 3 15
1.3.3 Chapter 4 16
1.3.4 Chapter 5 17
1.3.5 Chapter 6 18
1.3.6 Chapter 7 18
1.3.7 Chapter 8 19
1.3.8 Chapter 9 20
1.3.9 Chapter 10 20
References 21
2 Motor Fault Detection and Diagnosis Based on a Meta-cognitive Random Vector Functional Link Network 23
2.1 Introduction 23
2.1.1 Induction Motor 23
2.1.2 Hybrid Dynamic System 25
2.1.3 Our Approach 26
2.2 Fault Detection and Diagnosis in Induction Motors 29
2.2.1 Fault Detection and Diagnosis Features in an Induction Motor 29
2.2.2 Fault Detection Methods from Single and Multiple Sources 32
2.3 eT2RVFLN Architecture 33
2.3.1 Cognitive Architecture of an eT2RVFLN 33
2.3.2 Meta-cognitive Learning Policy of the eT2RVFLN 36
2.3.2.1 What to Learn 36
2.3.2.2 How to Learn 39
2.3.2.3 When to Learn 44
2.4 Experimental Design 45
2.5 Numerical Results 46
2.6 Conclusion 49
References 50
3 Optimal Adaptive Threshold and Mode Fault Detection for Model-Based Fault Diagnosis of Hybrid Dynamical Systems 53
3.1 Introduction 53
3.2 Bond Graph 57
3.2.1 Hybrid Bond Graph (HBG) Model 60
3.3 Diagnostic HBG Model for Uncertain System 61
3.3.1 Modelling Parameter Uncertainty 62
3.3.2 Modelling Measurement Uncertainty 64
3.3.3 ARR/GARR and Adaptive Threshold 65
3.3.4 Fault Signature Matrix and Coherence Vector 67
3.3.5 Proposed Method for Optimal Threshold and Mode Fault Detection 68
3.4 Case Study: Bench Mark Hybrid Two-Tank System 72
3.4.1 ARRs/GARRs for Hybrid Two-Tank System 74
3.4.2 Optimum Adaptive Threshold for Hybrid Two-Tank System 76
3.4.3 FDI Study for Hybrid Two-Tank System Using Proposed Technique 78
3.5 Conclusions 83
References 84
4 Diagnosing Hybrid Dynamical Systems Using Max-Plus Algebraic Methods 87
4.1 Introduction 87
4.2 Problem Statement 88
4.2.1 Hybrid Systems Model 89
4.2.2 Objective 89
4.2.3 System Architecture 89
4.3 Related Work 90
4.3.1 Algebraic Descriptions of Hybrid Systems 90
4.3.2 Petri Net Models 91
4.3.3 Diagnosing Hybrid Systems 91
4.4 Behaviour Modeling: Switching Max-Plus Linear Systems 91
4.4.1 Max-Plus Algebra 92
4.4.2 Continuous Dynamics: Max-Plus Linear Systems 92
4.4.3 Switching Max-Plus Linear Systems 93
4.4.4 Stochastic SMPL Systems 94
4.4.5 Generality of Approach 94
4.5 Running Example 95
4.5.1 Nominal Model 95
4.5.2 Fault Model 96
4.5.3 Max-Plus Model 97
4.6 Diagnosing Hybrid Systems Using SMPL Automata 98
4.6.1 Observers 98
4.6.2 Isolating Faults 99
4.7 Computational Complexity 100
4.7.1 Fault Detection 100
4.7.2 Fault Isolation 101
4.7.3 Approximation Algorithm 101
4.8 Diagnosis Scenarios 101
4.8.1 Scenario 1: T3 Leak 102
4.8.2 Scenario 2: V3 Blockage 102
4.8.3 Scenario 3: V2 Blockage 103
4.9 Types of Hybrid Systems Covered 103
4.10 Summary 105
References 106
5 Monitoring of Hybrid Dynamic Systems: Application to Chemical Process 108
5.1 Introduction 108
5.2 Residual Generation by the Extended Kalman Filter 109
5.2.1 State Estimator: Extended Kalman Filter 110
5.2.2 Residual Generation 110
5.3 Residual Estimation: Signature Generation 112
5.4 Determination of the Incidence Matrix 113
5.5 Fault Isolation 114
5.5.1 Principle 115
5.5.2 Distances 117
5.5.3 Decision Making 120
5.6 Monitoring of a Complex Chemical Process 121
5.6.1 Simulation of the Reference Model 122
5.6.2 Detection 123
5.6.3 Diagnosis 125
5.7 Conclusion 126
References 127
6 Hybrid Bond-Graph Possible Conflicts for Hybrid Systems Fault Diagnosis 129
6.1 Introduction 129
6.2 Characterizing PCs in the BG Modelling Framework 131
6.2.1 Hybrid Bond-Graphs for Hybrid Systems Modelling 131
6.2.2 Possible Conflicts from BG Models 132
6.3 Characterizing HBG-PCs 136
6.4 SHBG-PCs Motivating Example 138
6.5 Computing Structural HBG-PCs 140
6.5.1 The Algorithms 140
6.5.2 SHBG-PCs Found in the Motivating Example 142
6.6 Common Framework for Discrete and Parametric Faults 143
6.6.1 Assumptions 143
6.6.2 Fault Signature Matrices for Fault Isolation 144
6.6.3 The Diagnosis Framework 146
6.6.4 Complexity of the Approach 148
6.7 Case Study 149
6.7.1 Results for the Case Study 150
6.7.1.1 Discrete Fault in SW1 151
6.7.1.2 Parametric Fault in R01 154
6.8 Conclusions 154
References 156
7 Hybrid System Model Based Fault Diagnosis of Automotive Engines 159
7.1 Introduction 159
7.2 HNS Modeling of an SI Engine 161
7.2.1 Model Equations 165
7.2.2 Model Parameters and Tuning 167
7.2.3 Fault Modelling 168
7.3 Engine State Estimation 168
7.3.1 The Extended Kalman Filter 170
7.3.2 Estimators with Adaptive Q and/or R 171
7.4 Residual Prediction and Joint Estimation 172
7.4.1 Residual Prediction 173
7.4.2 Fault Detection Based on Generalized Likelihood Ratio Test (GLRT) on Predicted Residuals 174
7.4.3 Fault Isolation 175
7.5 Results 176
7.5.1 Estimation Results 176
7.5.2 Fault Diagnosis Results 178
7.6 Conclusions 181
References 182
8 Diagnosis of Hybrid Systems Using StructuralModel Decomposition 185
8.1 Introduction 185
8.2 Hybrid Systems Modeling 188
8.2.1 Compositional Modeling 188
8.2.2 Fault Modeling 191
8.2.3 Causality 192
8.2.4 Structural Model Decomposition 194
8.3 Problem Formulation 197
8.3.1 Architecture 199
8.4 Qualitative Fault Isolation for Hybrid Systems 199
8.4.1 Fault Signatures 200
8.4.2 Hybrid Systems Diagnosis 201
8.4.3 Scalability 203
8.5 Case Study 203
8.5.1 System Modeling 204
8.5.2 Structural Model Decomposition 207
8.5.3 Diagnosability 208
8.5.4 Results 209
8.6 Conclusions 211
References 211
9 Diagnosis of Hybrid Systems Using Hybrid Particle Petri Nets: Theory and Application on a Planetary Rover 214
9.1 Introduction 214
9.2 Related Work 216
9.3 Health Monitoring Methodology for Hybrid Systems 218
9.4 Hybrid System Modeling 221
9.4.1 Hybrid Particle Petri Nets 221
9.4.2 Illustration Example 224
9.4.3 Marking Evolution Rules in HPPN for Diagnosis 225
9.5 Hybrid System Diagnosis 227
9.5.1 Uncertainty 228
9.5.2 Diagnoser Generation 230
9.5.3 Diagnoser Process 234
9.6 Case Study 236
9.6.1 Rover Description 236
9.6.2 Rover Modeling 239
9.6.3 Simulation Results 240
9.6.3.1 Scenario 1 240
9.6.3.2 Scenario 2 241
9.6.3.3 Performance Analysis, Comparison with Other Approaches 242
9.7 Conclusion 243
References 244
10 Diagnosis of Hybrid Dynamic Systems Based on the Behavior Automaton Abstraction 247
10.1 Introduction 247
10.2 Hybrid System Diagnosis Methodology Overview 249
10.2.1 Diagnosis Architecture 249
10.2.2 Historical Review 251
10.3 Hybrid System Diagnosis Framework 253
10.3.1 The Hybrid Automaton Model 253
10.3.2 Consistency Indicators 256
10.3.3 Mode Discernibility Analysis 257
10.3.3.1 Case 1 258
10.3.3.2 Case 2 259
10.3.3.3 Case 3 259
10.3.4 The Behaviour Automaton Abstraction 260
10.3.5 The Hybrid Diagnoser 262
10.3.6 Mode Tracking Logic 262
10.4 Incremental Hybrid System Diagnosis 263
10.4.1 Incremental Diagnosis Architecture 263
10.4.2 Incremental Hybrid System Diagnosis Framework 264
10.5 Application Case Study 266
10.5.1 Barcelona Sewer Network 266
10.5.2 Hybrid Modeling 267
10.5.3 Hybrid System Diagnosis 269
10.5.3.1 Design 269
10.5.4 Incremental Hybrid System Diagnosis 274
10.5.5 Results 274
10.6 Conclusions 279
References 280
Index 283