Active Fault Tolerance for Sensor Failures in Steer-by-Wire Systems via Multi-Model Adaptive Kalman Filter

Abstract

The steer-by-wire (SBW) system is susceptible to parameter perturbations due to external disturbances under complex and variable conditions, leading to unstable performance. To address this challenge, this paper proposes a multi-model adaptive Kalman filter (MMAKF), which can efficiently attenuate the impact of the cornering stiffness variation and significantly enhances the precision of the front wheel angles estimation. On this basis, a fault diagnosis and fault-tolerant control (FDFTC) strategy is also proposed to mitigate the impact of faults. Joint simulations using Carsim and MATLAB/Simulink demonstrate the effectiveness of the proposed FDFTC strategy in detecting and reconstructing various sensor signal faults. Furthermore, the simulations conducted with varying road adhesion coefficients and vehicle speeds prove that the proposed MMAKF offers remarkable robustness and accurate estimation of front wheel angles under challenging conditions. Compared to the traditional Kalman filter (KF), it reduces maximum error by 22.17% and maximum root mean square error (RMSE) by 44.69%. Meanwhile, the validity of the proposed MMAKF algorithm and FDFTC strategy is demonstrated by the HIL platform. These results indicate promising prospects for the proposed MMAKF and FDFTC strategy in SBW fault-tolerant control applications and self-driving technology development.