Data-Driven and Model-Based Methods for Fault Detection and Diagnosis

Contents
List of figures
List of tables
About the authors
Acknowledgments
List of acronyms
Nomenclature
	Latin letters
	Greek letters
1 Introduction
	References
2 PCA and PLS-based generalized likelihood ratio for fault detection
	2.1 PCA and PLS-based generalized likelihood ratio for fault detection
		2.1.1 Introduction
		2.1.2 Principal component analysis (PCA)
			2.1.2.1 Modeling using PCA
			2.1.2.2 How many principal components to use?
		2.1.3 Fault detection using PCA method
		2.1.4 Statistical hypothesis testing
			2.1.4.1 Fault detection using hypothesis testing
			2.1.4.2 Generalized likelihood ratio GLRT
		2.1.5 Fault detection using a PCA-based GLRT
		2.1.6 PCA-based GLRT and applications
			2.1.6.1 Ozone monitoring using PCA-based GLRT
			2.1.6.2 Description of the training ozone data
			2.1.6.3 Ozone modeling using PCA
			2.1.6.4 Monitoring the ozone concentrations
			2.1.6.5 Process monitoring of a simulated continuously stirred tank reactor (CSTR)
			2.1.6.6 Modeling the CSTR data using PCA
			2.1.6.7 Simulation results
		2.1.7 Conclusion
	2.2 PLS-based generalized likelihood ratio for fault detection
		2.2.1 Introduction
		2.2.2 Partial Least Square (PLS) method
		2.2.3 PLS-based GLRT for fault detection
		2.2.4 PLS-based GLRT fault detection and applications
			2.2.4.1 Fault detection of continuously stirred tank reactor process
			2.2.4.2 Fault detection of Tennessee Eastman Process
		2.2.5 Conclusions
	References
3 Kernel PCA- and Kernel PLS-based generalized likelihood ratio tests for fault detection
	3.1 Kernel PCA-based generalized likelihood ratio test for fault detection
		3.1.1 Introduction
		3.1.2 Kernel Principal Component Analysis (KPCA) description
		3.1.3 Fault detection using KPCA method
		3.1.4 Enhanced monitoring using kernel GLRT chart
		3.1.5 Kernel GLRT fault detection chart with applications
			3.1.5.1 Application 1: synthetic data
			3.1.5.2 Application 2: nonisothermal CSTR process
		3.1.6 Conclusion
	3.2 Kernel PLS-based generalized likelihood ratio test for fault detection
		3.2.1 Introduction
		3.2.2 Kernel Partial Least Squares (KPLS) method
		3.2.3 KPLS-based GLRT and application to fault detection in CSTR process
		Case 1: faults in the concentration CA
		Case 2: fault in the temperature T
		Case 3: faults in the concentration CA and temperature T
		3.2.4 Conclusion
	References
4 Linear and nonlinear multiscale latent variable-based generalized likelihood ratio for fault detection
	4.1 Linear multiscale latent variable-based generalized likelihood ratio for fault detection
		4.1.1 Introduction
		4.1.2 Multiscale PCA-based GLRT for fault detection
			4.1.2.1 Modeling using multiscale PCA method
			4.1.2.2 Fault detection using GLRT
			4.1.2.3 MSPCA-based MW-GLRT and applications
		4.1.3 Multiscale PLS-based GLRT for fault detection
			4.1.3.1 Multiscale Partial Least Square (MSPLS) method
			4.1.3.2 MSPLS-based GLRT fault detection technique and applications
		4.1.4 Conclusions
	4.2 Multiscale nonlinear latent variable-based generalized likelihood ratio test for fault detection
		4.2.1 Introduction
		4.2.2 Multiscale kernel PCA-based GLRT for fault detection
			4.2.2.1 Multiscale kernel PCA description
			4.2.2.2 Multiscale kernel GLRT fault detection chart with applications
		4.2.3 Multiscale kernel PLS-based GLRT for fault detection
			4.2.3.1 Multiscale Kernel Partial Least Square (KPLS) method
			4.2.3.2 MSKPLS-based GLRT technique and applications
		4.2.4 Conclusion
	References
5 Linear and nonlinear interval latent variable approaches for fault detection
	5.1 Interval latent variable approaches for fault detection
		5.1.1 Introduction
		5.1.2 Interval PCA-based GLRT for fault detection
			5.1.2.1 Interval data description
			5.1.2.2 Principal component analysis for interval-valued data
			5.1.2.3 Interval-valued PCA model identification
			5.1.2.4 Fault detection using complete information PCA-based GLRT
			5.1.2.5 Complete information PCA-based GLRT and applications
			5.1.2.6 Fault detection using midpoints radii PCA-based EWMA
			5.1.2.7 Midpoints radii PCA-based EWMA and applications
		5.1.3 Interval PLS-based GLRT for fault detection
			5.1.3.1 Partial least squares for interval-valued data
			5.1.3.2 Fault detection charts based on interval PLS
			5.1.3.3 Fault detection using interval PLS-based GLRT
			5.1.3.4 Interval PLS-based GLRT and applications
		5.1.4 Conclusion
	5.2 Interval nonlinear latent variable approaches for fault detection
		5.2.1 Introduction
		5.2.2 Interval kernel PCA-based GLRT for fault detection
			5.2.2.1 Kernel PCA for interval-valued data (IKPCA)
			5.2.2.2 Interval KPCA-based fault detection charts
			5.2.2.3 Applications
		5.2.3 Interval kernel PLS-based GLRT for fault detection
			5.2.3.1 Kernel PLS for interval-valued data (IKPLS)
			5.2.3.2 Interval KPLS-based fault detection charts
			5.2.3.3 Interval KPLS-based GLRT and application
		5.2.4 Conclusion
	References
6 Model-based approaches for fault detection
	6.1 Introduction
	6.2 State estimation
		6.2.1 State estimation problem formulation
		6.2.2 State estimation techniques
			6.2.2.1 Extended Kalman filter (EKF)
			6.2.2.2 Unscented Kalman filter (UKF)
			6.2.2.3 Particle filter (PF)
	6.3 Fault detection-based state estimation approaches
		6.3.1 Fault detection using multiscale EWMA chart
			6.3.1.1 EWMA chart
			6.3.1.2 Multiscale EWMA chart
		6.3.2 Application to wastewater treatment plant
			6.3.2.1 State estimation results
			6.3.2.2 Fault detection results
	6.4 Fault detection-based state estimation approach
		6.4.1 Fault detection using optimized weighted SS-DEWMA chart
		6.4.2 Optimized WSS-DEWMA and application to fault detection
			6.4.2.1 Application 1: synthetic example
			6.4.2.2 Application 2: Cad System in E. coli (CSEC)
	6.5 Conclusions
	References
7 Conclusions and perspectives
	7.1 Conclusions
	7.2 Perspectives and research proposals
		7.2.1 Project 1: water distribution networks: modeling, sensor placement, leak and quality monitoring
		7.2.2 Project 2: enhanced operation of wastewater treatment plants
		7.2.3 Project 3: enhanced monitoring of photovoltaic systems
		7.2.4 Project 4: enhanced data validation of an air quality monitoring networks
Appendix
	Applications
		Tennessee Eastman Process (TEP)
		Distillation column
		Air quality monitoring network
		Continuously stirred tank reactor (CSTR)
	References
Index