Actuator and sensor fault detection based on LPV unknown input observer applied to lateral vehicle dynamics

Abstract

Abstract This chapter presents the estimation and detection of actuator and sensor faults via the unknown input observer (UIO) applied to a lateral dynamics model of an automated steering vehicle. The vehicle lateral dynamics have been described by a linear parameter varying (LPV) model taking into account the variations of the longitudinal velocity. It is known that the fault detection based on the observer is more suitable for actuator faults. The sensor faults are transformed into an augmented system with actuator faults. In our case, the sensors are affected by noise. For this reason, a first-order filter is needed to attenuate this noise. This chapter deals with the estimation of actuator and sensor faults by converting sensor faults into actuator faults and designing a UIO to estimate both states and faults. The gains of the observer can be calculated by solving linear matrix inequalities and the convergence of the observer is analyzed. In order to check the proposed method, a vehicle lateral dynamics model with steering angle actuator fault and yaw velocity sensor fault has been tested. Simulation results are presented to demonstrate the effectiveness of the proposed approach, which presents an efficient performance on both state and fault estimation.