A generalized least-squares fault detection filter (Q2704618)

Using a generalization of the least-squares derivation of the Kalman filter the authors develop a fault detection and identification algorithm (filter) in order to monitor the target fault and to block the nuisance faults in the case of linear time-invariant and time-varying systems. The approach consists in solving a min-max problem with a generalized least-squares criterion that makes the residual sensitive to the target fault, but insensitive to the nuisance faults. Some approximations to the classical fault detection filter and equivalence to the unknown input observer are discussed. Finally two examples illustrate the obtained results.