A ROBUST INFLUENCE MATRIX APPROACH TO FAULT-DIAGNOSIS

A robust scheme is proposed to detect faults, isolate them, and estima te their severity, The feature vector, which is a vector formed of the coefficients of the system transfer function, is estimated using a ro bust two-stage identification scheme: first a higher-order model is es timated using a singular value decomposition (SVD)-based batch least-s quares algorithm, and second, a reduced-order model is derived by filt ering-out the noise artifacts, The system is decomposed into functiona l units characterized by physical parameters, The influence of these p hysical parameters on the feature vector is captured in a vector terme d the influence vector. The distance between, and the inner product of the feature vector, and the influence vector are analyzed to diagnose faults, The proposed scheme is evaluated both on a simulated as well as an actual control system.