An on-line trained adaptive neural controller
Yao Zhang, Pratyush Sen, Grant E. Hearn · IEEE Control Systems · 1995
This article presents a neural network approach for online industrial tracking control applications. In comparison to several existing neural control schemes, the proposed direct neural controller is characterized by the simplicity of its structure and its practical applicability for real-time implementation. In order to enhance the adaptive ability of the neural controller, a set of fuzzy rules is set up for selecting interim training targets. With minor qualitative knowledge about the plant, the scheme is designed for controlling the nonlinear behavior of the plant under conditions of disturbances and noise. Simulations of a ship course-keeping control under random wind forces and measurement noise have been investigated and comparison of performance has been made with a conventional PID controller. Results presented clearly demonstrate the feasibility and adaptive property of the proposed scheme.>