Neural Networks for the Output Tracking-Control Problem of Nonlinear Strict-Feedback System

Abstract

This paper focuses on the tracking-control problem of nonlinear strict-feedback system by utilizing neural networks. Combining a novel recurrent neural network and gradient-based neural network, we investigate, develop and design a new controller based on the synthesized neural network model (N-G model) to track the output trajectory performance of the nonlinear strict-feedback system. This presented control scheme could have a good output tracking performance for the nonlinear strict-feedback system. For comparing with the presented N-G model, the classic backstepping design method is also employed to design the control input for the nonlinear strict-feedback control system in this paper. The computer simulation results demonstrate that the controller based on the N-G model could be used to tackle the tracking-control problem with accuracy and effectiveness, together with the faster convergent speed than that based on the backstepping algorithm. Generally speaking, with the appropriate increase of design parameters, the controller based on the N-G model could improve convergence performance for nonlinear strict-feedback system.

Keywords

• Output tracking-control
• nonlinear strict-feedback system
• recurrent neural network
• backstepping algorithm