Nonlinear sensor fault diagnosis using mixture of probabilistic PCA models

Abstract This paper presents a methodology for sensor fault diagnosis in nonlinear systems using a Mixture of Probabilistic Principal Component Analysis (MPPCA) models. This methodology separates the measurement space into several locally linear regions, each of which is associated with a Probabilistic PCA (PPCA) model. Using the transformation associated with each PPCA model, a parity relation scheme is used to construct a residual vector. Bayesian analysis of the residuals forms the basis for detection and isolation of sensor faults across the entire range of operation of the system. The resulting method is demonstrated in its application to sensor fault diagnosis of a fully instrumented HVAC system. The results show accurate detection of sensor faults under the assumption that a single sensor is faulty.

[1]  Reza Langari,et al.  A hybrid AANN-KPCA approach to sensor data validation , 2007 .

[2]  M. Kramer.  Nonlinear principal component analysis using autoassociative neural networks , 1991 .

[3]  Janos Gertler,et al.  Design of optimal structured residuals from partial principal component models for fault diagnosis in linear systems , 2005 .

[4]  Thomas F. Edgar,et al.  Identification of faulty sensors using principal component analysis , 1996 .

[5]  Reza Langari,et al.  Enhanced auto-associative neural networks for sensor diagnostics (E-AANN) , 2004, 2004 IEEE International Conference on Fuzzy Systems (IEEE Cat. No.04CH37542).

[6]  Robert E. Uhrig,et al.  Use of Autoassociative Neural Networks for Signal Validation , 1998, J. Intell. Robotic Syst..

[7]  Reza Langari,et al.  Isolability of faults in sensor fault diagnosis , 2011 .

[8]  E. C. Malthouse,et al.  Limitations of nonlinear PCA as performed with generic neural networks , 1998, IEEE Trans. Neural Networks.

[9]  Reza Langari,et al.  Sensor fault diagnosis with a probabilistic decision process , 2013 .

[10]  Donald J. Malloy,et al.  Development of Real-Time Engine Diagnostics Tools at the Arnold Engineering Development Center Engine Test Facility , 2005 .

[11]  Weihua Li,et al.  Detection, identification, and reconstruction of faulty sensors with maximized sensitivity , 1999 .

[12]  P. Miller,et al.  Contribution plots: a missing link in multivariate quality control , 1998 .

[13]  Paul M. Frank,et al.  Fault diagnosis in dynamic systems using analytical and knowledge-based redundancy: A survey and some new results , 1990, Autom..

[14]  Y. M. Chen,et al.  Neural networks-based scheme for system failure detection and diagnosis , 2002, Math. Comput. Simul..

[15]  Ten-Huei Guo,et al.  Using Neural Networks for Sensor Validation , 1998 .

[16]  Christopher M. Bishop,et al.  Mixtures of Probabilistic Principal Component Analyzers , 1999, Neural Computation.

[17]  Shi Jian Zhao,et al.  Levenberg-Marquardt algorithm for nonlinear principal component analysis neural network through inputs training , 2004, Fifth World Congress on Intelligent Control and Automation (IEEE Cat. No.04EX788).

[18]  Silvio Simani,et al.  Diagnosis techniques for sensor faults of industrial processes , 2000, IEEE Trans. Control. Syst. Technol..

[19]  Weihua Li,et al.  Isolation enhanced principal component analysis , 1999 .

[20]  Fei Liu,et al.  A New Method for Process Monitoring Based on Mixture Probabilistic Principal Component Analysis Models , 2006, ISNN.

[21]  Michael E. Tipping,et al.  Probabilistic Principal Component Analysis , 1999 .