Fault-tolerant control of steer-by-wire system based on Gaussian process regression

Abstract

To address the issue of steer-by-wire vehicles being unable to steer as expected due to steering motor voltage faults, a steer-by-wire system model based on bond graph theory is established, with the lateral displacement of the rack in the rack-and-pinion steering mechanism as the final output of the model. A sliding mode observer is employed to estimate the system states, and both the observer’s output and sensor measurements are used to perform voltage estimation using a Gaussian Process Regression (GPR) model. The residual between the estimated voltage and the ideal steering voltage is then compared to detect faults and compensate for the faulty voltage. The proposed method is validated through simulations on the CarSim/Simulink co-simulation platform, including voltage estimation, fault diagnosis, vehicle front-wheel steering angle, and vehicle trajectory. The results show that the proposed method can accurately estimate faults and compensate the steer-by-wire system during steering motor faults, thereby improving steering stability and path-tracking performance.