Adaptive fault tolerant optimal control for high-order fully actuated system and its application in spacecraft attitude tracking using reinforcement learning

This study addresses the adaptive optimal tracking problem for high-order fully actuated system (FAS). First, an optimal control strategy based on FAS theory and reinforcement learning (RL) is proposed, effectively solving the tracking problem within the actor-critic framework. Next, within the identifier-actor-critic framework, a sliding mode optimized control scheme for high-order FAS with actuator fault is introduced. Specifically, unlike existing fault-tolerant control results for high-order FAS, this method ensures trajectory tracking and performance optimization even in the presence of actuator faults, while avoiding the need for persistent excitation condition. Considering the strong robustness and rapid response advantages of sliding mode control (SMC), synchronous rapid tracking and optimal control of multiple variables are achieved by establishing a (n-1)-order sliding mode surface. Furthermore, rigorous theoretical analysis proves the boundedness of all signals within the closed-loop system. Finally, the effectiveness of the proposed control approaches are verified by a numerical simulation and a practical example of the spacecraft attitude system.