ADALINE-Based Friction Identification and Compensation of a Linear Voice-Coil DC Motor

Abstract

This paper presents a nonlinear friction identification and compensation algorithm based on artificial neural networks (ANNs) for a linear positioning system. The objective is to compensate the dead-zone phenomenon caused by static and Coulomb friction and improve the tracking accuracy of the positioning system. An adaptive linear neuron (ADALINE) with a known mass and viscous friction coefficient is used to estimate the Coulomb friction coefficient. A feedforward compensation combined with the Adaline estimated Coulomb friction is implemented. An experimental setup with a linear voice coil DC motor and a high-resolution position encoder is developed to test and validate the identification and compensation algorithms.