Combined approach to fault diagnosis based on principal components and influence matrix

This paper presents an on-line combined fault diagnosis approach for diagnosis of multiplicative (parametric) faults in dynamical systems. The fault detection task is performed applying the principal component analysis to the parameters of a discrete-time model. The fault isolation task is based on an influence matrix method, assuming that exist a relationship between the discrete-time model parameters and the physical parameters. A performance analysis regarding faults and disturbances effects is presented, and new contributions regarding the application of this approach to linear and nonlinear systems are presented. The performance of the approach is tested on a simulation model of a DC motor under closed-loop control.