An optimum robust approach to statistical failure detection and identification

The failure detection and identification (FDI) problem for noise-corrupted linear time invariant systems is considered. The authors apply to this problem an optimum minmax robust likelihood ratio testing approach which is known optimal in the Gaussian case. The originality of this approach is that detection probability and false alarm probability in the presence of noise are considered when referring to optimality and robustness. A new implementation of this approach for recursive FDI in noisy systems reduces the isolation of the failure of interest by rejecting alternative failures, and performing a descriptor Kalman filter to account for the presence of noise and to produce the desired likelihood ratio