Neural-network-based adaptive control for space manipulator system with actuator faults: a fully actuated system approach

In the presence of actuator potential faults, this paper investigates trajectory tracking control for a free-floating space manipulator system (FFSM) subject to nonlinear uncertainties. The study aims to achieve desired tracking performance directly while addressing the challenges posed by system uncertainties and actuator faults. To accomplish this objective, the proposed control strategy relies on the high-order fully actuated (HOFA) system approach, consisting of three main components: a basic part designed to eliminate known nonlinearities and emphasize linear dominant terms; a neural network (NN)-based adaptation technique introduced to estimate the unknown bounds of actuators’ restricted effectiveness loss and lumped uncertainties; and a robustness component incorporated to mitigate the adverse effects of actuator bias faults. Utilizing Lyapunov stability theory, the NN-based adaptive fault-tolerant controller guarantees boundedness of all signals in the robust closed-loop system, while ensuring that position/velocity tracking errors are confined to a small neighborhood of zero relative to a well-designed eigenstructure. Finally, the effectiveness of the proposed control scheme is demonstrated through simulations conducted on a 6-DoF FFSM system.