RBF neural network application in internal model control of permanent magnet synchronous motor

Abstract

As a significant part of artificial intelligence (AI) techniques, neural network is recently having a great impact on the control of motor. Particularly, it has created a new perspective of decoupling and linearization. With reference to the non-linearization and strong coupling of multivariable permanent magnet synchronous motor (PMSM), this paper presents internal model control (IMC) of PMSM using RBF neural network inverse (RBF-NNI) system. In the proposed control scheme, the RBF-NNI system is introduced to construct a pseudo-linear system with original system, and internal model controller is utilized as a robust controller. Therefore, the new system has advantages of above two methods. The efficiency of the proposed control scheme is evaluated through computer simulation results. By using the proposed control scheme, original system is successfully decoupled, and expresses strong robustness to load torque disturbance, the whole system provides good static and dynamic performance. © 2011 Springer-Verlag.