Implementation of Feedback Error Learning for Position Control System of Linear Slider

Abstract

This paper considers implementation of feedback error learning (FEL) for linear slider position control. The effect of modelling error of the slider is compensated by FEL. In algorithm of FEL, parameters of feedforward controller are updated so that output signals of a feedback controller can become small. It leads to improvement of tracking performance. On the other hand, when conventional FEL algorithm is implemented for control of a linear slider, parameters of the feedforward controller tend to converge slowly. In industrial scenes, it is desirable to shorten learning time in order to reduce cost of setting up manufacturing devices. Furthermore, when the learning time is long, guide rails of the slider become worn. It is shown that convergence speed is improved by tuning free parameters of the feedforward controller. Due to further improvement of the convergence, by using prefilter and free parameters of feedforward controller, some parameters of an inverse system are obtained a priori and utilized for learning. Command with mixed frequency is selected in order to identify accurate inverse system of plant. Simulation results are presented to show the effectiveness of proposed method.