Discussion on: "Sensor gain fault diagnosis for a class of nonlinear systems"
Résumé
The paper by Y. Wang and D. H. Zhou (European Journal of contol, 2006, vol 12, pp. 523-535) presents an approach of sensor gain fault detection of nonlinear systems. The fisrt step of this approach consists on the location of the sensor gain faults that can be estimated. The proposed algorithm is based on the computation of the failure matrix D such that the zeros of the linear part of the system (A,0,C,D) are stable. The second step is the simultaneous estimation of the state and the fault vectors using a singular systems representation. Finally, the sensor gain faults are estimated using an adaptive estimator. This paper is an interesting contribution to the sensor gain faults diagnosis of nonlinear systems provided that the proposed approach gives both the location and the estimation of the failure. This discussion focuses on the existence conditions of the simultaneous estimation of the state and the fault vector and on the selection of the conditionnally identifiable faults. It is shown first that the detectability of the pair (A,C) is not necessary and only the measurement provided by the conditionally detectable sensors are necessary to estimate the state of the system. An estimation of the fault vector is then given by the output of the observer using the conditionally identifiable sensors. Thereafter, we claim that the simultaneous estimation scheme can be simplified using a low order observer and without using descriptor systems formulation. An nth-order state observer and a qth-order functional observer which estimates the conditionally identifiable faults are then proposed. Finally, an algorithm to determine the matrix D based on a detectability condition is exhibited.
Domaines
Automatique / Robotique
Origine : Fichiers produits par l'(les) auteur(s)
Loading...