Vicenc Puig is Professor of Automatic Control at the Universitat Politècnica de Catalunya (UPC), Spain. He has published more than 80 journal articles and more than 350 articles in international conference/workshop proceedings related to diagnosis and faulttolerant control.Silvio Simani is Professor of Automatic Control in the Engineering Department of Ferrara University, Italy. He has published about 260 journal and conference papers, several book chapters and four monographs on fault diagnosis and sustainable control topics.

• Introduction ixVicenç Puig and Silvio SimaniChapter 1 Mathematical Modeling and Fault Description 1Silvio Simani1.1. Introduction 11.2. Model-based FDI techniques 21.3. Modeling of faulty systems 31.3.1. Fault modeling and description 51.3.2. Mathematical description 61.4. Residual generation 111.5. Residual generation techniques 141.5.1. Residual generation via parameter estimation 151.5.2. Observer-based approaches 181.5.3. Fault detection via parity equations 241.6. Change detection and symptom evaluation 281.7. Residual generation robustness problem 301.7.1. FDI H∞ approach 321.7.2. Active and passive disturbance decoupling 351.8. Fault diagnosis technique integration 361.8.1. Fuzzy logic for residual generation 371.8.2. Neural networks for fault diagnosis 381.8.3. Neuro-fuzzy approaches to FDI 401.8.4. Fault detectability and isolability 421.8.5. NF model structure identification 431.8.6. NF residual generation for FDI 441.9. Conclusion 461.10. References 47Chapter 2 Structural Analysis 57Mattias Krysander and Erik Frisk2.1. Introduction 572.2. Background 582.2.1. Structural models 582.2.2. Dulmage–Mendelsohn decomposition and matchings 602.2.3. Dulmage–Mendelsohn decomposition and simulation 632.3. Fault isolability analysis 642.3.1. Fault detectability analysis 642.3.2. Fault isolability analysis 652.3.3. Canonical isolability decomposition of the overdetermined part 672.4. Testable submodels 692.4.1. Basic definitions 692.4.2. MSO algorithm 712.4.3. Residual generation based on matching 722.5. Sensor placement 742.5.1. The basic sensor placement problem 742.5.2. A structural approach 752.6. Summary and discussion 802.7. References 81Chapter 3 Set-based Fault Detection and Isolation 83Ye Wang and Vicenç Puig3.1. Introduction 833.2. Notations, definitions and properties 843.3. Problem statement 863.3.1. Uncertain discrete-time linear systems 863.3.2. Set-based methods 863.3.3. FDI problem statement 883.4. Proposed techniques 893.4.1. Set-membership approach 893.4.2. Zonotopic observer 903.4.3. Relationship between set-based methods 913.5. Design methods 923.5.1. Robustness conditions 923.5.2. Fault sensitivity condition 963.6. Fault detection and isolation procedures 993.6.1. Fault detection 993.6.2. Fault isolation 1003.7. Application example: quadruple-tank system 1013.7.1. Results with robustness condition 1053.7.2. Results with robustness and fault sensitivity conditions 1053.8. Conclusion 1053.9. References 109Chapter 4 Diagnosis of Stochastic Systems 111Gregory Provan4.1. Introduction 1114.2. Stochastic diagnosis task 1134.2.1. Notation 1134.2.2. Problem formulation 1134.2.3. Representing uncertainty 1154.3. Inference methods for diagnosis task 1164.3.1. Difference with other tasks 1164.4. Model-based approach 1174.4.1. Traditional FDD methods 1174.4.2. Bayesian inversion/filtering 1204.5. Data-driven approaches 1224.5.1. ML methods 1234.5.2. Statistical methods 1244.6. Hybrid approaches: surrogate methods 1254.6.1. Fitting surrogate models via sampling 1254.7. Comparative analysis of approaches 1264.8. Summary and conclusions 1274.9. References 128Chapter 5 Data-Driven Methods for Fault Diagnosis 131Silvio Simani5.1. Introduction 1315.2. Models for linear system fault diagnosis 1335.3. Parameter estimation methods for fault diagnosis 1355.3.1. Data-driven method in ideal conditions 1355.3.2. Data-driven methods in real scenarios 1385.3.3. Algebraic Frisch scheme 1395.3.4. Dynamic Frisch scheme 1415.3.5. MIMO case Frisch scheme 1455.4. Nonlinear dynamic system identification 1465.4.1. Piecewise affine model 1475.4.2. Hybrid model structure 1485.4.3. Nonlinear system approximation 1495.4.4. Model continuity and domain partitioning 1515.4.5. Local affine model estimation 1545.4.6. Multiple-model estimation 1585.5. Fuzzy data-driven approach to fault diagnosis 1645.5.1. Fuzzy model identification 1655.5.2. Takagi–Sugeno prototypes 1675.5.3. Data-driven fuzzy modeling 1705.5.4. Clustering methods 1705.5.5. Fuzzy c-means clustering algorithms 1725.5.6. Gustafson–Kessel clustering algorithm 1745.5.7. Optimal number of clusters 1765.6. Fuzzy model identification 1765.6.1. Nonlinear model identification 1785.6.2. Product space clustering identification 1815.6.3. Fuzzy clustering model identification 1835.6.4. Antecedent membership function estimation 1835.6.5. Estimating consequent parameters 1855.7. Conclusion 1895.8. References 189Chapter 6 The Artificial Intelligence Approach to Model-based Diagnosis 197Belarmino Pulido, Carlos J. Alonso-González and Anibal Bregon6.1. Introduction 1976.2. Case studies 1996.3. Knowledge-based diagnosis systems 2016.3.1. Diagnosis task and system model 2036.3.2. Diagnosis of physical devices 2066.3.3. Limits of KBS for diagnosis of physical devices 2076.4. Model-based diagnosis 2086.4.1. Formalization of consistency-based diagnosis and its first implementation, GDE 2096.5. CBD for dynamic systems 2176.5.1. Different approaches for CBD of dynamic systems 2196.5.2. PCs for the three-tank system case study 2226.6. Conclusion 2246.7. References 226List of Authors 231Index 233Summary of Volume 2 237

Vicenc Puig is Professor of Automatic Control at the Universitat Politècnica de Catalunya (UPC), Spain. He has published more than 80 journal articles and more than 350 articles in international conference/workshop proceedings related to diagnosis and faulttolerant control.Silvio Simani is Professor of Automatic Control in the Engineering Department of Ferrara University, Italy. He has published about 260 journal and conference papers, several book chapters and four monographs on fault diagnosis and sustainable control topics.

• Introduction ixVicenç Puig and Silvio SimaniChapter 1 Mathematical Modeling and Fault Description 1Silvio Simani1.1. Introduction 11.2. Model-based FDI techniques 21.3. Modeling of faulty systems 31.3.1. Fault modeling and description 51.3.2. Mathematical description 61.4. Residual generation 111.5. Residual generation techniques 141.5.1. Residual generation via parameter estimation 151.5.2. Observer-based approaches 181.5.3. Fault detection via parity equations 241.6. Change detection and symptom evaluation 281.7. Residual generation robustness problem 301.7.1. FDI H∞ approach 321.7.2. Active and passive disturbance decoupling 351.8. Fault diagnosis technique integration 361.8.1. Fuzzy logic for residual generation 371.8.2. Neural networks for fault diagnosis 381.8.3. Neuro-fuzzy approaches to FDI 401.8.4. Fault detectability and isolability 421.8.5. NF model structure identification 431.8.6. NF residual generation for FDI 441.9. Conclusion 461.10. References 47Chapter 2 Structural Analysis 57Mattias Krysander and Erik Frisk2.1. Introduction 572.2. Background 582.2.1. Structural models 582.2.2. Dulmage–Mendelsohn decomposition and matchings 602.2.3. Dulmage–Mendelsohn decomposition and simulation 632.3. Fault isolability analysis 642.3.1. Fault detectability analysis 642.3.2. Fault isolability analysis 652.3.3. Canonical isolability decomposition of the overdetermined part 672.4. Testable submodels 692.4.1. Basic definitions 692.4.2. MSO algorithm 712.4.3. Residual generation based on matching 722.5. Sensor placement 742.5.1. The basic sensor placement problem 742.5.2. A structural approach 752.6. Summary and discussion 802.7. References 81Chapter 3 Set-based Fault Detection and Isolation 83Ye Wang and Vicenç Puig3.1. Introduction 833.2. Notations, definitions and properties 843.3. Problem statement 863.3.1. Uncertain discrete-time linear systems 863.3.2. Set-based methods 863.3.3. FDI problem statement 883.4. Proposed techniques 893.4.1. Set-membership approach 893.4.2. Zonotopic observer 903.4.3. Relationship between set-based methods 913.5. Design methods 923.5.1. Robustness conditions 923.5.2. Fault sensitivity condition 963.6. Fault detection and isolation procedures 993.6.1. Fault detection 993.6.2. Fault isolation 1003.7. Application example: quadruple-tank system 1013.7.1. Results with robustness condition 1053.7.2. Results with robustness and fault sensitivity conditions 1053.8. Conclusion 1053.9. References 109Chapter 4 Diagnosis of Stochastic Systems 111Gregory Provan4.1. Introduction 1114.2. Stochastic diagnosis task 1134.2.1. Notation 1134.2.2. Problem formulation 1134.2.3. Representing uncertainty 1154.3. Inference methods for diagnosis task 1164.3.1. Difference with other tasks 1164.4. Model-based approach 1174.4.1. Traditional FDD methods 1174.4.2. Bayesian inversion/filtering 1204.5. Data-driven approaches 1224.5.1. ML methods 1234.5.2. Statistical methods 1244.6. Hybrid approaches: surrogate methods 1254.6.1. Fitting surrogate models via sampling 1254.7. Comparative analysis of approaches 1264.8. Summary and conclusions 1274.9. References 128Chapter 5 Data-Driven Methods for Fault Diagnosis 131Silvio Simani5.1. Introduction 1315.2. Models for linear system fault diagnosis 1335.3. Parameter estimation methods for fault diagnosis 1355.3.1. Data-driven method in ideal conditions 1355.3.2. Data-driven methods in real scenarios 1385.3.3. Algebraic Frisch scheme 1395.3.4. Dynamic Frisch scheme 1415.3.5. MIMO case Frisch scheme 1455.4. Nonlinear dynamic system identification 1465.4.1. Piecewise affine model 1475.4.2. Hybrid model structure 1485.4.3. Nonlinear system approximation 1495.4.4. Model continuity and domain partitioning 1515.4.5. Local affine model estimation 1545.4.6. Multiple-model estimation 1585.5. Fuzzy data-driven approach to fault diagnosis 1645.5.1. Fuzzy model identification 1655.5.2. Takagi–Sugeno prototypes 1675.5.3. Data-driven fuzzy modeling 1705.5.4. Clustering methods 1705.5.5. Fuzzy c-means clustering algorithms 1725.5.6. Gustafson–Kessel clustering algorithm 1745.5.7. Optimal number of clusters 1765.6. Fuzzy model identification 1765.6.1. Nonlinear model identification 1785.6.2. Product space clustering identification 1815.6.3. Fuzzy clustering model identification 1835.6.4. Antecedent membership function estimation 1835.6.5. Estimating consequent parameters 1855.7. Conclusion 1895.8. References 189Chapter 6 The Artificial Intelligence Approach to Model-based Diagnosis 197Belarmino Pulido, Carlos J. Alonso-González and Anibal Bregon6.1. Introduction 1976.2. Case studies 1996.3. Knowledge-based diagnosis systems 2016.3.1. Diagnosis task and system model 2036.3.2. Diagnosis of physical devices 2066.3.3. Limits of KBS for diagnosis of physical devices 2076.4. Model-based diagnosis 2086.4.1. Formalization of consistency-based diagnosis and its first implementation, GDE 2096.5. CBD for dynamic systems 2176.5.1. Different approaches for CBD of dynamic systems 2196.5.2. PCs for the three-tank system case study 2226.6. Conclusion 2246.7. References 226List of Authors 231Index 233Summary of Volume 2 237

Vicenc Puig is Professor of Automatic Control at the Universitat Politècnica de Catalunya (UPC), Spain. He has published more than 80 journal articles and more than 350 articles in international conference/workshop proceedings related to diagnosis and faulttolerant control.Silvio Simani is Professor of Automatic Control in the Engineering Department of Ferrara University, Italy. He has published about 260 journal and conference papers, several book chapters and four monographs on fault diagnosis and sustainable control topics.

• Introduction ixVicenç Puig and Silvio SimaniChapter 1 Mathematical Modeling and Fault Description 1Silvio Simani1.1. Introduction 11.2. Model-based FDI techniques 21.3. Modeling of faulty systems 31.3.1. Fault modeling and description 51.3.2. Mathematical description 61.4. Residual generation 111.5. Residual generation techniques 141.5.1. Residual generation via parameter estimation 151.5.2. Observer-based approaches 181.5.3. Fault detection via parity equations 241.6. Change detection and symptom evaluation 281.7. Residual generation robustness problem 301.7.1. FDI H∞ approach 321.7.2. Active and passive disturbance decoupling 351.8. Fault diagnosis technique integration 361.8.1. Fuzzy logic for residual generation 371.8.2. Neural networks for fault diagnosis 381.8.3. Neuro-fuzzy approaches to FDI 401.8.4. Fault detectability and isolability 421.8.5. NF model structure identification 431.8.6. NF residual generation for FDI 441.9. Conclusion 461.10. References 47Chapter 2 Structural Analysis 57Mattias Krysander and Erik Frisk2.1. Introduction 572.2. Background 582.2.1. Structural models 582.2.2. Dulmage–Mendelsohn decomposition and matchings 602.2.3. Dulmage–Mendelsohn decomposition and simulation 632.3. Fault isolability analysis 642.3.1. Fault detectability analysis 642.3.2. Fault isolability analysis 652.3.3. Canonical isolability decomposition of the overdetermined part 672.4. Testable submodels 692.4.1. Basic definitions 692.4.2. MSO algorithm 712.4.3. Residual generation based on matching 722.5. Sensor placement 742.5.1. The basic sensor placement problem 742.5.2. A structural approach 752.6. Summary and discussion 802.7. References 81Chapter 3 Set-based Fault Detection and Isolation 83Ye Wang and Vicenç Puig3.1. Introduction 833.2. Notations, definitions and properties 843.3. Problem statement 863.3.1. Uncertain discrete-time linear systems 863.3.2. Set-based methods 863.3.3. FDI problem statement 883.4. Proposed techniques 893.4.1. Set-membership approach 893.4.2. Zonotopic observer 903.4.3. Relationship between set-based methods 913.5. Design methods 923.5.1. Robustness conditions 923.5.2. Fault sensitivity condition 963.6. Fault detection and isolation procedures 993.6.1. Fault detection 993.6.2. Fault isolation 1003.7. Application example: quadruple-tank system 1013.7.1. Results with robustness condition 1053.7.2. Results with robustness and fault sensitivity conditions 1053.8. Conclusion 1053.9. References 109Chapter 4 Diagnosis of Stochastic Systems 111Gregory Provan4.1. Introduction 1114.2. Stochastic diagnosis task 1134.2.1. Notation 1134.2.2. Problem formulation 1134.2.3. Representing uncertainty 1154.3. Inference methods for diagnosis task 1164.3.1. Difference with other tasks 1164.4. Model-based approach 1174.4.1. Traditional FDD methods 1174.4.2. Bayesian inversion/filtering 1204.5. Data-driven approaches 1224.5.1. ML methods 1234.5.2. Statistical methods 1244.6. Hybrid approaches: surrogate methods 1254.6.1. Fitting surrogate models via sampling 1254.7. Comparative analysis of approaches 1264.8. Summary and conclusions 1274.9. References 128Chapter 5 Data-Driven Methods for Fault Diagnosis 131Silvio Simani5.1. Introduction 1315.2. Models for linear system fault diagnosis 1335.3. Parameter estimation methods for fault diagnosis 1355.3.1. Data-driven method in ideal conditions 1355.3.2. Data-driven methods in real scenarios 1385.3.3. Algebraic Frisch scheme 1395.3.4. Dynamic Frisch scheme 1415.3.5. MIMO case Frisch scheme 1455.4. Nonlinear dynamic system identification 1465.4.1. Piecewise affine model 1475.4.2. Hybrid model structure 1485.4.3. Nonlinear system approximation 1495.4.4. Model continuity and domain partitioning 1515.4.5. Local affine model estimation 1545.4.6. Multiple-model estimation 1585.5. Fuzzy data-driven approach to fault diagnosis 1645.5.1. Fuzzy model identification 1655.5.2. Takagi–Sugeno prototypes 1675.5.3. Data-driven fuzzy modeling 1705.5.4. Clustering methods 1705.5.5. Fuzzy c-means clustering algorithms 1725.5.6. Gustafson–Kessel clustering algorithm 1745.5.7. Optimal number of clusters 1765.6. Fuzzy model identification 1765.6.1. Nonlinear model identification 1785.6.2. Product space clustering identification 1815.6.3. Fuzzy clustering model identification 1835.6.4. Antecedent membership function estimation 1835.6.5. Estimating consequent parameters 1855.7. Conclusion 1895.8. References 189Chapter 6 The Artificial Intelligence Approach to Model-based Diagnosis 197Belarmino Pulido, Carlos J. Alonso-González and Anibal Bregon6.1. Introduction 1976.2. Case studies 1996.3. Knowledge-based diagnosis systems 2016.3.1. Diagnosis task and system model 2036.3.2. Diagnosis of physical devices 2066.3.3. Limits of KBS for diagnosis of physical devices 2076.4. Model-based diagnosis 2086.4.1. Formalization of consistency-based diagnosis and its first implementation, GDE 2096.5. CBD for dynamic systems 2176.5.1. Different approaches for CBD of dynamic systems 2196.5.2. PCs for the three-tank system case study 2226.6. Conclusion 2246.7. References 226List of Authors 231Index 233Summary of Volume 2 237