Optimal Control Problem via Self-adaptation Sliding Mode Controller with Neural Network

Abstract

This paper proposes the author's new Self-adaptation Sliding Mode Controller which added a neural network (SA-SMC+NN) for the optimal control problem. The controlled system is the linear time invariant system and the system parameter and the disturbance are unknown. As minimizing the quadratic cost function, the neural network gives the coefficient of the switching function of the sliding mode control. According to this proposed technique, we do not have to tune the parameter of the controller when apply SA-SMC+NN. Furthermore, we are able to get the constant feedback gain such as the optimal control regulator for the uncertain system based on the control results by SA-SMC+NN. Differential game is simulated to confirm the effectiveness of the proposed method.