Study of supervisory control for servo system based on RBF neural network

The goal of this work is to present a new control method., which can achieve an accurate and robust position control for servo system. This method combines PD feedback control with RBF neural network feed-forward control. By learning the outputs of PD controller. the RBF network adjusts its weights online. and becomes the main controller gradually. The model of AC servo system and the learning rules of RBF neural network are discussed in detail. Furthermore Simulation analysis have been carried by Matlab, and the experiments have been done based on a DSP platform. The results indicate that the method has a good control effect on both regulating and set-point following, and that is a robust and accurate solution for position-control servo systems with disturbances.