ABSTRACT A novel Actor‐Critic Algorithm based Super‐Twisting Sliding Mode Fault‐Tolerant Control (ACSTSM‐FTC) Method for spraying manipulators is proposed in this paper under actuator failures and input constraints. The Euler–Lagrange dynamic model for Attitude Adjustment Mechanism (AAM) has been established. A novel fusion framework has been proposed that integrates the Actor‐Critic algorithm with Radial Basis Function Neural Networks (RBFNN), which can effectively compensate the uncertainty of model parameters and lumped disturbance induced by actuator faults. Based on Actor‐Critic algorithm and super‐twisting sliding mode control method, a nonsingular fault‐tolerant controller is designed, which effectively suppresses the jitter problem of the traditional control method, realizes the accurate tracking of the desired trajectory, and realizes control energy optimization at the same time. Lyapunov analysis proves error uniformly bounded within fixed time. Simulation and experimental results validate the controller's global convergence, accelerated convergence in 1.5 s, and robustness against disturbances across arbitrary initial states.
Zhao et al. (Tue,) studied this question.