Fuzzy Adaptive Tracking of Constrained Nonlinear Systems With Event-Sampling Reinforcement Learning

This article is concerned with the optimized tracking control problem of high-order strict-feedback nonlinear systems with full-state constraints via dynamical event-triggered strategy. An adaptive optimized control algorithm is proposed using the backstepping technique and reinforcement learning method. The algorithm incorporates an identifier–critic–actor architecture and features a flexible design of dynamic event-triggered mechanism (ETM) to decrease signal transmission. The ETM is adaptively adjusted by the estimated fuzzy adaptive weights and design parameters, making it more efficient in terms of communication resources. Moreover, to ensure the system states do not exceed their constrained sets while achieving the optimization objective, a novel performance index function based on the barrier Lyapunov function is constructed. Finally, it can be demonstrated that the system output accurately tracks the desired reference signal, while ensuring that all closed-loop signals remain bounded. Simulation example confirms the validity of the developed design methodology.