Nonlinear Modeling of Switched Reluctance Motor Based on GA-BPNN

Abstract

AbstractNonlinear modeling of the flux linkage characteristics is fundamental to the control design and performance evaluation of switched reluctance motor (SRM). Conventional back propagation neural network (BPNN) modeling method has demerits of local minimum which significantly slow the convergence rate. So, genetic algorithm (GA) is introduced to overcome the local minimum of BPNN in this paper. And GA is applied to train the weighs and bias for BPNN. And Based on the flux linkage characteristics obtained from finite element method (FEM) of a 16/12 SRM, a nonlinear flux linkage model using GA-BPNN is set up, and simulation results demonstrate that the model has fast convergence rate, high accuracy, and strong generalization ability.KeywordsNonlinear modelingback propagation neural network (BPNN)genetic algorithm (GA)flux linkage characteristicsswitched reluctance motor (SRM)