Adaptive Model Predictive Fault-Tolerant Control for Four-Wheel Independent Steering Vehicles with Sensitivity Estimation

This paper presents an adaptive model predictive fault-tolerant control (FTC) algorithm based on sensitivity estimation and exponential forgetting-based recursive least squares (RLS) for four-wheel independent steering vehicles. The model predictive control algorithm was designed according to physical constraints for four-wheel independent steering control with adaptive integral action. To improve the control performance in transient and steady-state regions, sensitivity-based adaptive rules for the weighting factor of the model predictive controller and integral gain were developed using the gradient descent method. The sensitivity was defined by a virtual relationship function and was estimated using RLS with a forgetting factor. Additionally, a FTC strategy with the equality constraint was proposed for enhancing the yaw-rate tracking control performance despite the existence of faults in the steering system. The proposed fault-tolerant steering control algorithm was developed in a MATLAB/Simulink environment, and its performance was evaluated via co-simulation in the MATLAB/Simulink and CarMaker software programs under various evaluation scenarios.