Title :
Failure detection and isolation: a new paradigm
Author :
Doraiswami, R. ; Diduch, C.P. ; Kuehner, J.
Author_Institution :
Dept. of Electr. & Comput. Eng., New Brunswick Univ., Fredericton, NB, Canada
Abstract :
A new approach to the failure detection and isolation problem or FDI is presented that combines parametric and nonparametric methods. The system is modeled as an interconnection of components that are subject to parametric failures. The FDI scheme is implemented in three stages. A diagnostic model is first developed to characterize the evolution of a feature vector and influence matrix as parameters in each component are subject to failures. Next a residual is generated using a parity equation and it is shown that the residual is a linear function of the change in the diagnostic parameters. In the third stage, isolation of faults is accomplished by finding the maximum correlation between the residual and estimates of the residual generated using a number of failure hypotheses. The development of the diagnostic model, a technique for identifying the diagnostic model using input-output measurements and its application in FDI is a significant contribution of the paper
Keywords :
Bayes methods; fault diagnosis; identification; least squares approximations; pattern classification; transfer functions; white noise; FDI; diagnostic model; diagnostic parameters; failure detection and isolation; failure hypotheses; feature vector; influence matrix; input-output measurements; nonparametric methods; parametric failures; parametric methods; parity equation; residual generated; Attenuation measurement; Equations; Fault detection; Fault diagnosis; Filters; Least squares methods; Noise measurement; Solid modeling; Transfer functions; Vectors;
Conference_Titel :
American Control Conference, 2001. Proceedings of the 2001
Conference_Location :
Arlington, VA
Print_ISBN :
0-7803-6495-3
DOI :
10.1109/ACC.2001.945589
