Nonlinear modeling of switched reluctance motor using different methods

Accurate modeling of flux-linkage characteristics is the basis of design and control of switched reluctance motor (SRM). The flux-linkage characteristic of SRM is a function of both the excitation current and rotor position. But due to the highly nonlinear characteristics of SRM, modeling is cumbersome. In this paper, three effective algorithms for modeling of SRM are investigated, which are based on subregional mathematical model, neural network and adaptive neural fuzzy inference systems(ANFIS) respectively. The performance of each method is validated and compared via matlab tools. Simulation results show that all these models can accurately and successfully modeling of SRM, while RBF neural network is the best. Moreover, a new SRM dynamic simulation model based on RBF is developed for verification. Simulation and experimental results validated the accuracy of RBF modeling algorithm.