Reinforcement-learning-based controller design for nonaffine nonlinear systems

Abstract

In this paper, we develop an online learning control for a class of unknown nonaffine nonlinear discrete-time systems with unknown bounded disturbances. Under the framework of reinforcement learning, we employ two neural networks (NNs): an action NN is used to generate the control signal, and a critic NN is utilized to estimate the prescribed cost function. By using Lyapunov’s direct method, we prove the stability of the closed-loop system. Moreover, based on the developed adaptive scheme, we show that all signals involved are uniformly ultimately bounded. Finally, we provide an example to demonstrate the effectiveness and applicability of the present approach.