The Model of Nonlinear Radial Force in Switched Reluctance Motor Based on Radial Basis Function Neuron Network

摘要

Based on radial basis function neuron network (RBFNN), the model of nonlinear radial force in switched reluctance motor(SRM) is constructed in this paper. Training samples for RBFNN are obtained from the calculation results of a prototype SRM(8/6) with finite element method(FEM). Training algorithm is a hybrid method combining nearest neighbor clustering with steepest gradient descent. The simulation comparison results of the RBFNN with the hybrid training algorithm in this paper and the back propagation neuron network (BPNN) with Levenberg-Marquardt training algorithm in MATLAB validates the superiority of the RBFNN.