A methodology for block-oriented industrial nonlinear system by nonlinear separation control with neural learning

Abstract

This paper presents a general methodology for designing controller for block-oriented industrial nonlinear system by nonlinear separation control with neural learning. In this study, through rough approximation of inverse input output nonlinear statics and accurate compensation of nonlinear dynamics with rigid definition of neural network as well as learning from actual system, control performances can be improved. Based on this method, high-precision contour control of industrial articulated robot arm and outlet working fluid heat rate control of evaporator in energy conversion plant were realized. The experiment and simulation verified the significant potential of the proposed method to industrial nonlinear systems.