PMLSM controller design based on self-constructing feedback fuzzy neural network

Abstract

Self-constructing feedback fuzzy neural network controller (SCFFNNC) is designed to aim at the parameter uncertainties for permanent magnet linear synchronous motor (PMLSM) since they cause negative influence to the dynamic performance of PMLSM servo system. The thought about self-constructing, in which the number of neurons for the whole structure can be able to increase on line according to the variety of error, is introduced based on combining the non-linear identification of fuzzy control with self-learning of neural network. It can reserve the self-learning abilities, improve the real-time performance for neural network and then enhance the dynamic performance of PMLSM. The simulation results show that the servo system for PMLSM based on SCFFNNC can realize quick response, high precision and strong robustness for the parameter uncertainties of PMLSM.