Modeling of switched reluctance motor based on combined clustering RBF network

Abstract

The paper presents combined clustering RBF neural network as a tool to develop the model of the SRM. Combined clustering algorithm is presented here to determine node number of hidden layer and center of RBF neural network. First, the subtractive clustering algorithm is used to find the initial clustering center. FCM(Fuzzy c-means) clustering algorithm is used for further adjustment and effectiveness evaluation. It can generate a good number of clusters according to the influence of each data point in each dimension of the cluster center. Then, optimal data center of radial basis function RBF neural network is achieved. The sampled data set is obtained from the experimental SRM by the finite elements method(FEM). The simulation results show that the model is reasonable and can reflect the electromagnetic characteristics of the motor. The established model is easy to extend, which provides the basis for the analysis and design of SRM control algorithm.