Modelling and Estimation Strategies for Fault Diagnosis of Non-Linear Systems: From Analytical to Soft Computing Approaches Book

Introduction     1
Introductory Background     4
Content     6
Principles of Fault Diagnosis
Analytical Techniques-Based FDI     11
Approaches to Linear Systems     11
Parameter Estimation     11
Parity Relation     13
Observers     14
Approaches to Non-linear Systems     16
Parameter Estimation     17
Parity Relation     17
Observers     19
Robustness Issues     22
Concluding Remarks     29
Soft Computing-Based FDI     31
Neural Networks     32
Basic Structures     32
Neural Networks in FDI     39
Evolutionary Algorithms     41
Introduction to Evolutionary Computation     41
Evolutionary Algorithms in FDI Schemes     44
Concluding Remarks     45
State and Parameter Estimation Strategies
State Estimation Techniques for FDI     49
Extended Unknown Input Observer     49
Convergence Analysis and Design Principles     51
Convergence of the EUIO     52
Illustrative Example     56
Extended Unknown Input Observer Revisited     58
Convergence Analysis and Design Principles     61
Illustrative Examples     66
Design of Observers and Unknown Input Observers for Lipschitz Systems     66
Convergence Analysis     67
Design Procedures     71
Design of an Unknown Input Observer     74
Experimental Results     76
Concluding Remarks     81
Parameter Estimation-Based FDI     85
Experimental Design     86
Impedance Measurement and Diagnosis     88
Problem Formulation     89
Experimental Design     91
Inductive Impedance     95
Towards Robustness     96
Simulation Results     97
Concluding Remarks     100
Soft Computing Strategies
Evolutionary Algorithms     105
Model Design with Genetic Programming     106
Model Structure Determination Using GP     108
State-Space Representation of the System     112
Robustifying the EUIO with Genetic Programming     114
Robustifying the EUIO with the ESSS Algorithm     115
Experimental Result     119
State Estimation and Fault Diagnosis     119
Industrial Application      125
Concluding Remarks     132
Neural Networks     133
Robust Fault Detection with the Multi-layer Perceptron     134
Illustrative Example     139
Algorithms and Properties of D-OED for Neural Networks     141
Locally D-Optimum On-Line Sequential Design     153
Industrial Application     160
Robust Fault Detection with GMDH Neural Networks     163
Model Uncertainty in the GMDH Neural Network     165
Bounded-Error Approach     167
Synthesis of the GMDH Neural Network Via the BEA     172
Robust Fault Detection with the GMDH Model     175
Industrial Application     176
Concluding Remarks     182
Conclusions and Future Research Directions     185
Appendix     191
References     193
Index     205