Optimal Iterative Learning-Based Fault-Tolerant Control for Vehicle-Manipulator Systems with Actuator Failures

Abstract

This study investigates a novel neuroadaptive faulttolerant control (FTC) strategy for vehicle-manipulator systems operating under simultaneous actuator faults and uncertain external disturbances. The proposed framework integrates an optimal iterative control policy with extreme learning machine (ELM)-based neural estimators. By leveraging the state prediction and cost function evaluation derived from state and critic networks, the control law is progressively given through the motion network. A key innovation is the removal of the conventional restriction that only a single actuator fault can occur at any given time. The resulting neuroadaptive FTC ensures that tracking errors converge uniformly to a bounded invariant set around the origin, effectively mitigating the adverse effects of disturbances and actuator anomalies. Comparative simulation studies confirm the superior tracking accuracy and robustness of the proposed control architecture.