Fault-Tolerant Control of Steer-by-Wire System Based on Sliding Mode Observer With Neural Network

This paper proposes a novel automatic disturbance rejection fault-tolerant control method based on a sliding mode observer with neural network for steer-by-wire (SbW) systems subjected to internal uncertainty, external disturbance, and steering motor fault. Firstly, to mitigate the adverse effects of non-smooth desired steering angle signals on the transient performance of the control system, a third-order tracking-differentiator is employed to smooth the desired signal. Secondly, a sliding mode observer with neural network is designed, which utilizes an adaptive radial basis function (RBF) neural network to reconstruct the internal and external uncertainties, as well as the fault-induced uncertainty, and employs the terminal sliding mode algorithm to ensure fast convergence of observation errors. The asymptotic stability of the observer is demonstrated by Lyapunov stability theory. Subsequently, a finite-time feedback control law is developed. To compensate for these uncertainties in the closed-loop system, the reconstruction results from the observer are incorporated into the control channel. Based on this framework, stability analysis shows that the tracking errors can reach a time-varying residual set within finite time and subsequently converge asymptotically to the origin as this residual set contracts to zero. Finally, the effectiveness and superiority of the proposed control method are verified through simulations conducted using Matlab/Simulink and CarSim.