Neural network real-time IP position controller on-line design for permanent magnetic linear synchronous motor

Abstract

This paper presents a real-time IP position controller realized by a neural network for a permanent magnetic linear synchronous motor (PMLSM) servo system. The proposed neural network, whose weight definitely has material meaning, is simple and can be rapidly adjusted on-line, and the real-time position control for PMLSM is accomplished. In order to improve the robustness and the control precision of a PMLSM drive system, the mover mass, viscous damping factor and disturbance force are estimated by the proposed estimator which is composed of a recursive least-squares estimator (RLSE) and a disturbance force observer A simulation demonstrates that the proposed IP controller makes the system more robust to the uncertain load and the variation of the parameters.