Simultaneous actuator and sensor faults reconstruction for a class of linear systems. Application to a steer-by-wire vehicle system

Abstract

In this article, a fault reconstruction and estimation (FRE) approach is developed for linear systems corrupted simultaneously by sensor and actuator faults. Contrary to classical detectability criteria considered in the literature, the FRE approach that we suggest is designed under weakened assumptions. Indeed, we assume that the faulty system has less outputs than faults and weakened “observer matching” and “minimum phase” assumptions are made for the considered system. First, we apply a filter on the faulty outputs to obtain an appropriate extended state model for which we apply the technique of auxiliary outputs generation and we design a new high gain sliding mode observer whose the role is to reconstruct conjointly sensor faults and auxiliary outputs. Then, we synthesize an unknown input observer to estimate system state and finally we deduce an estimation approach to reconstruct actuator faults based on all estimated signals by the designed cascaded observers. Simulations are performed to illustrate the efficiency of the proposed FRE approach through an example of a vehicle system.