Actuator multiplicative fault estimation in discrete-time LPV systems using switched observers

Abstract

This paper proposes an observer for the joint state and fault estimation devoted to discrete-time linear parameter varying (LPV) systems subject to actuator faults. The major contribution of this work is that the observer is able to estimate multiplicative faults, contrarily to the existing approaches, that consider additive faults. The main characteristic of this observer is that it is scheduled not only by means of the endogenous varying parameters of the faulty model, but also by the input vector. Another contribution of this paper consists in adding a switching component in order to guarantee the feasibility of the conditions for designing the observer gains. It is proved that, as long as the input sequence satisfies some characteristics, the convergence of the observer error dynamics to zero is assured. A numerical example is used to demonstrate the effectiveness of the proposed strategy.