Comparison of Feedback Controllers for Feedback-Error-Learning Neural Network Control System with Application to a Flexible Micro-Actuator

Abstract

A feedback-error-learning neural network approach to on-line learning control and real time implementation for a flexible micro-actuator is presented. The flexible micro-actuator is made of a bimorphic piezo-electric high-polymer material (Poly Vinylidene Fluoride). The control scheme consists of a feedforward neural network controller and a fixed-gain feedback controller. This neural network controller is trained so as to make the output of the feedback controller zero. In the process, the neural network learns the inverse dynamics of the system. We make some comparisons between using PID and LQG controllers with this neural network controller. Experimental and numerical results for the tracking control of a piezopolymer actuator are presented and they show that the feedback-error-learning neural network is effective in accurately tracking a reference signal.