Adaptive RBF iterative learning tracking control of multi-degree-of-freedom manipulators

Abstract Targeting on the issue of poor dynamic performance of multi-degree-of-freedom manipulator system under disturbance environment and model parameter uncertainty, this paper designs an adaptive iterative learning tracking control method utilizing a radial basis function (RBF) neural network observer, aiming at realizing stable control and disturbance suppression of closed-loop system. The controller consists of the proportional-differential term based on the backstepping method, the sign function robust term of adaptive sliding mode, and the disturbance observation term of the RBF neural network. The sign function robust term compensates for the repetitive disturbance of the system, and the RBF observer compensates for the non-repetitive disturbance of the system. Under the premise of ensuring the normal functioning of the control system, the sliding mode parameter is updated in the iterative domain, and the RBF network weight parameter is updated in the time domain to enhance the system’s robustness. Simultaneously, the stability of the closed-loop system within the iterative domain and the convergence of the tracking error are demonstrated using the Lyapunov composite energy theory. Finally, the proposed control method’s efficacy and robustness are validated through simulation and experiment.