Prescribed Performance Fault-Tolerant Control for Uncertain Nonlinear MIMO System Using Actor–Critic Learning Structure

Xuerao Wang,Qingling Wang,Changyin Sun

DOI: https://doi.org/10.1109/tnnls.2021.3057482

IF: 14.255

2022-01-01

IEEE Transactions on Neural Networks and Learning Systems

Abstract:This article studies the prescribed performance fault-tolerant control problem for a class of uncertain nonlinear multi-input and multioutput systems. A learning-based fault-tolerant controller is proposed to achieve the asymptotic stability, without requiring a priori knowledge of the system dynamics. To deal with the prescribed performance, a new error transformation function is introduced to convert the constrained error dynamics into an equivalent unconstrained one. Under the actor-critic learning structure, a continuous-time long-term performance index is presented to evaluate the current control behavior. Then, a critic network is used to approximate the designed performance index and provide a reinforcement signal to the action network. Based on the robust integral of the sign of error feedback control method, an action network-based controller is developed. It is shown by the Lyapunov approach that the tracking error can converge to zero asymptotically with the prescribed performance guaranteed. Simulation results are provided to validate the feasibility and effectiveness of the proposed control scheme.