Issues of Fault Diagnosis for Dynamic Systems

by Patton, Ron J.; Frank, Paul M.; Clark, Robert N.

• ISBN13: 9783540199687
• ISBN10: 3540199683
• Format: Hardcover
• Copyright: 2000-04-01
• Publisher: Springer Verlag

Summary

There is an increasing demand for dynamic systems to become safer, more reliable and more economical in operation. This requirement extends beyond the normally accepted safety-critical systems e.g., nuclear reactors, aircraft and many chemical processes, to systems such as autonomous vehicles and some process control systems where the system availability is vital. The field of fault diagnosis for dynamic systems (including fault detection and isolation) has become an important topic of research. Many applications of qualitative and quantitative modelling, statistical processing and neural networks are now being planned and developed in complex engineering systems. Issues of Fault Diagnosis for Dynamic Systems has been prepared by experts in fault detection and isolation (FDI) and fault diagnosis with wide ranging experience. Subjects featured include: - Real plant application studies; - Non-linear observer methods; - Robust approaches to FDI; - The use of parity equations; - Statistical process monitoring; - Qualitative modelling for diagnosis; - Parameter estimation approaches to FDI; - Fault diagnosis for descriptor systems; - FDI in inertial navigation; - Stuctured approaches to FDI; - Change detection methods; - Bio-medical studies. Researchers and industrial experts will appreciate the combination of practical issues and mathematical theory with many examples. Control engineers will profit from the application studies.

Table of Contents

List of Contributors	xxiii
Introduction	1	(14)
R. J. Patton
P. M. Frank
R. N. Clark
Background and Definitions	1	(4)
Fault Diagnosis Methodology	5	(1)
Summary of Structure and Content of the Book	6	(9)
Integration of Fault Detection and Diagnosis Methods	15	(36)
Rolf Isermann
Introduction	15	(1)
Fault Detection and Fault Diagnosis	16	(2)
Model-Based Fault Detection Methods	18	(10)
Fault detection with parameter estimation	19	(2)
Fault detection with state-estimation	21	(4)
Fault detection with parity equations	25	(1)
Fault detection with signal models	26	(1)
Change detection and symptom generation	27	(1)
Applicability of Model-Based Fault Detection Methods	28	(4)
Assumptions of model-based fault detection	28	(2)
Fault modelling	30	(1)
Suitability of quantitative model-based fault detection methods	31	(1)
Integration of Different Fault Detection Methods	32	(1)
Fault Diagnosis Methods	33	(18)
Symptom representation	35	(1)
Heuristic knowledge representation	35	(3)
Diagnostic reasoning	38	(3)
Appendix A. Parameter estimation and state estimation for the detection of tool wear and breakage for machine tools	41	(4)
Appendix B: Fault diagnosis of a machine tool feed drive with fuzzy reasoning	45	(6)
Observability and Redundancy Decomposition Application to Diagnosis	51	(36)
Jose Ragot
Didier Maquin
Frederic Kratz
Introduction	51	(2)
Static Redundancy Equations	53	(8)
Presentation	54	(1)
Genration of redundancy equations by direct elimination	55	(1)
Genration of redundancy equations by projection	55	(1)
The constraint static case	56	(2)
A systematic decomposition	58	(2)
Example	60	(1)
Dynamic Redundancy Equations	61	(14)
Presentation	61	(1)
Basis of redundancy generation	62	(1)
Direct redundancy (self-redundancy)	62	(2)
Redundancy between sensors (inter-redundancy)	64	(1)
Direct generation of the redundancy equations	65	(3)
Generation of redundancies by reduction of the state equations	68	(1)
Generation of redundancies by projection	69	(1)
Generation of non-independent redundancy equations	70	(2)
Interconnected systems	72	(1)
Observability decomposition	73	(2)
Analysis of the Residuals	75	(12)
Presentation	76	(3)
Residual criterion analysis	79	(1)
Imbalances or adjustments using vectors analysis	80	(1)
Generalised likelihood ratio approach	81	(1)
Parity space approach	82	(5)
Fault Detection and Isolation for Linear Systems Using Detection Observers	87	(28)
Chongzhi Fang
Wei Ge
Deyun Xiao
Problem Formulation	87	(1)
System model with fault modes	87	(1)
Fault detection using observers	87	(1)
Detection observers	88	(3)
Definition of detection observers	88	(1)
Fault detectability	89	(2)
Fault Isolation Logic	91	(9)
Fault Isolation problem	91	(1)
Fault Isolation logic	91	(2)
The maximum fault isolation information	93	(4)
Additional fault isolation logic	97	(2)
Solution of fault isolation problem	99	(1)
Detection Observer System Design	100	(7)
Detection signals of a detection observer	100	(1)
Conditions of existence for sensitivity vectors	101	(2)
Solution for detection observer structure conditions	103	(3)
Design algorithm for detection observer with given sensitivity	106	(1)
An Example	107	(8)
Fault Detection and Isolation for Descriptor Systems	115	(30)
Ming Hou
Introduction	115	(1)
Problem Formulation	115	(3)
The Existence of Analytical Redundancy	118	(2)
Fault Detectability	120	(5)
Fault Isolability	125	(2)
FDI Observer Design	127	(3)
Illustrative Example	130	(7)
Concluding Remarks	137	(2)
Appendix: Proofs of Theorems	139	(6)
Robust Observer-Based Fault Diagnosis in Non-Linear Uncertain Systems	145	(44)
Ralf Seliger
Paul M. Frank
Introduction	145	(2)
Problem Statement	147	(2)
Residual Generation by Non-Linear Unknown Input Observers	149	(2)
Process Modelling	151	(5)
Nominal Process Model	152	(1)
Fault and Disturbance Modelling	152	(4)
Disturbance Decoupling	156	(8)
Fault Sensitive Disturbance Decoupling Transformation	156	(1)
Solution Procedures	157	(7)
Design of Non-Linear Unknown Input Fault Detection Observers	164	(11)
Design by Estimation Error Linearisation	165	(5)
Design by Partial Estimation Error Linearisation	170	(2)
Design by Linearisation about the Estimated State	172	(3)
Robust Residual Evaluation by Threshold Selection	175	(4)
A Performance Index	179	(4)
Computational Evaluation of the Performance Index	181	(2)
Application Example	183	(2)
Conclusions	185	(4)
Uncertainty Modelling and Robust Fault Diagnosis for Dynamic Systems	189	(30)
Ron J. Patton
Jie Chen
Introduction	189	(1)
Brief Details of Robust Fault Diagnosis using Eigenstructure Assignment	190	(3)
Direct Determination and Optimisation of Disturbance Distribution Matrix	193	(4)
Noise and additive non-linearity	193	(1)
Model reduction	193	(1)
Parameter perturbations	194	(1)
Low rank approximation of distribution matrix	195	(1)
Bounded uncertainty	196	(1)
Estimation of Disturbance and Disturbance Distribution Matrix	197	(7)
Estimation of disturbance vector using an augmented observer	198	(1)
Estimation of disturbance distribution matrix	199	(1)
Estimation of disturbance vector using de-convolution	200	(4)
Optimisation of Distribution Matrix for Varying Operating Point Cases	204	(1)
Modelling and Fault Detection for Jet Engine System	205	(11)
Application of direct computation and optimisation method	207	(2)
Application of augmented observer method	209	(7)
Conclusion	216	(3)
Reliability Models for Sensor Fault Detection with State-Estimator Schemes	219	(26)
Dirk van Schrick
Peter C. Muller
Introduction	219	(2)
Description of IFD-Estimator Schemes	221	(4)
Basic Theory of Markov Processes	225	(4)
Modelling of the IFD-Estimator Schemes	229	(6)
Comparison of the IFD-Estimator Schemes	235	(8)
Conclusions	243	(2)
A Structural Framework for the Design of FDI System in Large Scale Industrial Plants	245	(40)
M. Staroswiecki
J. P. Cassar
P. Declerck
Introduction	245	(2)
The Model Based Approach for FDI	247	(6)
Residual generation using analytical models	249	(2)
The limitations of analytical models	251	(1)
The structural approach	252	(1)
Structural Model And Canonical Decomposition	253	(16)
The structure of the model	253	(3)
Direct redundancy relations (Staroswiecki & Declerck, 1990)	256	(1)
Deduced redundancy relations (Staroswiecki & Declerck, 1990)	257	(9)
Residual generation	266	(1)
Matching under causal constraints	266	(2)
Related studies	268	(1)
Control of the results	268	(1)
Using the Residuals for Fault Detection and Isolation	269	(10)
Boolean signature of a fault	270	(1)
The FDI procedure	271	(6)
Robustness and sensitivity	277	(2)
The Implementation of the FDI Procedure	279	(4)
Decomposition of the FDI procedure	280	(2)
Distribution of the FDI system and smart instruments	282	(1)
Conclusions	283	(2)
Structured Parity Equations in Fault Detection and Isolation	285	(30)
Janos J. Gertler
Introduction	285	(2)
System Description	287	(6)
Nominal (fault-free) system	287	(2)
Additive faults and disturbances	289	(1)
Multiplicative faults and modelling errors	290	(3)
Parity Equations	293	(3)
Parity equations from the input-output model	293	(2)
Parity equations from the state-space model	295	(1)
Structuring the Residuals for Additive Faults and Disturbances	296	(9)
Isolability and disturbance decoupling	297	(1)
Designing the equation structure	298	(3)
Structured parity equations	301	(4)
Structured residuals from the state-equations	305	(1)
Structuring the Residuals for Multiplicative Faults and Modelling Errors	305	(6)
Structuring strategies	306	(1)
Transformation of the residuals	306	(4)
Generating residuals for mixed faults	310	(1)
Conclusion	311	(4)
Input Design for Change Detection	315	(24)
Feza Kerestecioglu
Martin B. Zarrop
Introduction	315	(2)
SPRT and CUSUM Test	317	(5)
SPRT	317	(2)
CUSUM test	319	(1)
Performance measures for the CUSUM test	320	(2)
Input Design Problem	322	(1)
Offline Inputs	323	(7)
Problem refinement	323	(1)
Power constrained inputs	324	(6)
Online Inputs	330	(5)
Problem refinement	330	(2)
A suboptimal solution	332	(3)
Simulation Examples	335	(3)
Conclusions	338	(1)
On-Line Detection and Diagnosis of Sensor and Process Faults in Nuclear Power Plants	339	(40)
Johannes Prock
Introduction: Problem Statement	339	(2)
A Three Level Concept for Fault Detection	341	(24)
1st level: hardware redundancy	342	(2)
2nd level: analytical redundancy	344	(16)
3rd level: system redundancy	360	(5)
Diagnostic Problems	365	(6)
Diagnostic instability	367	(1)
Alarm messages of the three level's modules	368	(1)
The use of an experrt diagnosis system	369	(1)
Hybrid AI-systems ideas	370	(1)
LYDIA - A System for Early Sensor and Process Fault Detection and Diagnosis	371	(5)
A first example	372	(3)
Application to the primary and secondary side of a pressurised water reactor	375	(1)
Conclusions	376	(3)
Application of Sensor Fault Classification Algorithms to a Benson Steam Boiler	379	(34)
Kristian Kroschel
Andreas Wernz
Introduction	379	(3)
Scheme and Model of the Steam Generator	382	(3)
Assumptions about the estimation error signals	384	(1)
Basic Instrument Fault Diagnosis Scheme	385	(6)
Detection of a sensor fault	385	(2)
Estimation and classification of a sensor fault	387	(3)
Experimental results	390	(1)
Suppression of Modelling Errors	391	(6)
De-correlation method	392	(4)
Implementation of the de-correlation filters	396	(1)
Results	397	(1)
Identification of Process Parameters	397	(13)
Bank of state-augmented Kalman filters	400	(1)
Design of the Kalman filters	401	(3)
Computation of the a posteriori probabilities	404	(1)
Figures of merit for detectability and separability	404	(1)
Tuning of the Kalman filters	405	(1)
Results	406	(4)
Conclusions	410	(3)
Detection of Events in Signals VIA The Model-Based Fault Diagnosis Approach: Application to Bio-Electrical Signals	413	(22)
Dominique Sauter
Thierry Cecchin
David Brie
Introduction	413	(1)
Theoretical Background	414	(11)
Problem formulation	414	(2)
Residual generation	416	(3)
Residual evaluation	419	(2)
Robustness considerations	421	(4)
Bio-Electrical Signal Processing	425	(8)
Separation of fast and slow phases of eye tracking movement	425	(2)
Rejection of artefacts for EEG spectral analysis	427	(3)
Extraction of spindles in sleep EEG for spectral analysis	430	(3)
Conclusions	433	(2)
A Study of Fault-Tolerant Integrated Navigation Systems	435	(26)
Hong-Yue Zhang
Han-Guo Zhang
Jie Chen
Ron J. Patton
Bruce K. Walker
Introduction	435	(2)
Fault Tolerance for a Redundant Sensor Configuration	437	(10)
Redundant sensor systems	437	(2)
Direct comparison of the measurements	439	(2)
Generalised likelihood test (GLT)	441	(6)
Combining Algorithm for Decentralised Estimation	447	(5)
Fault Detection and Isolation for Subsystems	452	(3)
X2 test	452	(2)
Residual test	454	(1)
Application and Simulation Results	455	(5)
Conclusions	460	(1)
A Hierarical Structure for on-Line Process Fault Diagnosis Based on Deep Qualitative Modelling	461	(24)
Jie Zhang
Peter D. Roberts
Introduction	461	(2)
Qualitative Modelling	463	(3)
Qualitative modelling based on confluences	463	(1)
Order of magnitude reasoning	464	(2)
A Hierarchical Structure for On-Line Process Fault Diagnosis	466	(6)
The lower level diagnosis system	466	(2)
The upper level diagnosis system	468	(4)
Application to a Pilot Scale Mixing Process	472	(9)
The mixing process	472	(1)
Qualitative modelling of the mixing process	472	(3)
Fault detection and diagnosis	475	(4)
A case study	479	(1)
Performance of the hierarchical fault diagnosis system	480	(1)
Conclusions	481	(4)
Fault Diagnosis Based on a Predicate Logic Description of Dynamical Systems	485	(32)
Jan Lunze
Frank Schiller
Introduction	485	(5)
The Rationale for Qualitative Diagnosis	485	(3)
Outline of the proposed diagnostic method	488	(2)
Qualitative Modelling	490	(10)
Qualitative vs quantitative modelling	490	(1)
Qualitative description of signals	491	(3)
Description of the perturbed steady state by propositional logic formulae	494	(2)
Description of the perturbed dynamical behaviour by predicate logic formulae	496	(2)
Causality graph and aggregated causality graph	498	(2)
The Diagnostic Problem	500	(1)
Diagnosis Based on a Propositional Logic Description of the System	500	(6)
Solution by a resolution refutation system	501	(1)
Solution by deduction	501	(2)
Reduction of the search space of the deduction system	503	(2)
The structure of the diagnostic algorithm	505	(1)
Solution Based on a Predicate Logic Description	506	(3)
Example - a Tank System	509	(5)
Conclusions	514	(3)
Monitoring and Diagnosis of Fermentation Processes	517	(30)
Kouamana Bousson
Jean-Philippe Steyer
Boutaib Dahhou
Louise Trave-Massuyes
Introduction	517	(2)
Fermentation Processes	519	(5)
General features	519	(4)
Case study	523	(1)
The BIOTECH System	524	(9)
Detection of sensor malfunction	525	(1)
Signal interpretation	525	(2)
Causal qualitative model	527	(3)
Generating influence tables	530	(3)
CA-EN: A Model-Based Reasoning Module	533	(12)
Introduction	535	(1)
Modelling formalism	535	(5)
Prediction and interpretation algorithms	540	(2)
Interpretations and supervisory control	542	(1)
Application to the fed-batch fermentation process	543	(2)
Conclusions	545	(2)
Process Monitoring and Fault Detection Using Multivariate SPC	547	(20)
E.B. Martin
A.J. Morris
Introduction	547	(2)
Process Monitoring and Process Control	549	(1)
Multivariate Statistical Methods	550	(3)
Principal Components Analysis (PCA)	550	(1)
Projection to Latent Structures	551	(2)
Multivariate Statistical Process Control	553	(2)
The Squared Prediction Error Plot	553	(1)
Confidence Bounds	554	(1)
Interpreting the `Out of Control' Signal	555	(1)
Polymer Process Monitoring and Fault Detection	555	(4)
Batch Processes	559	(5)
Multi-way Principal Component Analysis	561	(1)
An example of Process Performance Monitoring of a Batch Process	562	(2)
Conclusions	564	(3)
References	567