A Fault-Tolerant Model Predictive Controller With Reconfiguration for Optimal Lateral Stability in Over-Actuated Electric Vehicles

This paper presents a fault-tolerant reconfigurable model predictive control method to improve the longitudinal and lateral reference tracking of four-wheel-steering and four-wheel-drive vehicles under concurrent steering actuator faults and fault modes. The proposed method detects, isolates, and then estimates the magnitude of individual faults. Fault magnitude information is then forwarded to the model-predictive controller, which configures its state-space input matrix and constraints based on the information provided to overcome the adverse effects of these faults that may manifest as both actuator loss of effectiveness and jam. It is shown that the integration of the fault estimation and the reconfiguration procedure does not affect the stability of the control system. The performance of the proposed method is first verified using obstacle avoidance maneuvering in a control-oriented, simpler vehicle model. An experiment with a scaled test vehicle is also conducted to prove the real-time applicability of the proposed control algorithm. Our analysis with two representative scenarios shows that the proposed reconfigurable fault-tolerant MPC algorithm can improve the system's performance by a minimum of 30% (and as much as 86%) compared to the conventional and adaptive model predictive controllers designed for vehicle stability.