Friction-Compensation and Extended State Observer Based Model Predictive Control for PMSM Servo Systems

Abstract

In the conventional design of model predictive control (MPC) for PMSM servo systems, friction is usually treated as a part of the lumped disturbances and is eliminated by disturbance observer or integral control methods. In order to obtain better control performance, a composite MPC method is proposed by introducing friction and disturbance information. Firstly, an approximate friction model is obtained by fitting the Stribeck model linearly and piecewise. Secondly, an extended state observer (ESO) is designed to estimate the lumped disturbances, which are composed of the friction compensation bias and the external load disturbance. Thereafter, a more specific prediction model is established by embedding the results of friction identification and disturbance estimation in the mechanism model. In this way, the influence of friction and disturbance are taken into account in the receding optimization process. Various experimental results demonstrate the feasibility and effectiveness of the proposed method for higher-precision motion trajectory tracking. Â