Robust observer design for discrete descriptor-LPV systems with unmeasurable gain scheduling functions
Résumé
This paper presents the design of a robust LPV observer for discrete descriptor-linear parameter varying systems. The gain scheduling functions are considered as depending on the unmeasurable state vector. Designing an effective observer state for this type of systems is more complicated because implies the estimation of the scheduling functions. To reduce the design complexity the original system is transformed into a perturbed system depending on the estimated scheduling functions. The convergence of the estimation error is guaranteed by considering $\mathscr{L}_2$ gain and Lyapunov approach which relies in sufficient conditions given in linear matrix inequalities(LMIs) formulation. Moreover, in fault detection (FD) and fault tolerant control (FTC) framework, the robust discrete LPV observer demonstrates its capacity to detect sensor and actuator faults. Simulation results illustrate effectiveness of the proposed method.