NN adaptive optimal tracking control for a class of uncertain nonstrict feedback nonlinear systems

Abstract

A novel reinforcement learning-based adaptive optimal controller is proposed to obtain the desired tracking performance for a class of nonlinear nonstrict feedback systems with uncertain dynamics in this paper. The main feature is that the proposed control scheme can handle the control problem that traditional reinforcement learning-based algorithm cannot deal with. To achieve the optimal control of the high-order system, the virtual and the actual control of the system are optimized by using reinforcement learning method. Radial basis function neural networks are employed to approximate the uncertain system dynamics, the optimal cost function and the optimal control law, respectively. According to Lyapunov stability theorem, it is proved that all the error signals in the closed-loop systems are semi-globally uniformly ultimately bounded (SGUUB) while the desired tracking control performance can be obtained. Simulation results are given to illustrate the effectiveness of the proposed algorithm.