Nonlinear modeling and simulating of switched reluctance motor and its drive

Flux-linkage characteristics and torque characteristics of switched reluctance motor (SRM) are all highly nonlinear function of rotor position and phase current, it is difficult to be expressed accurately by analytic formula. The torque model of SRM with small scale, convergence and accuracy by optimizing network structure and algorithm is developed in this paper by back propagation (BP) neural network with Levenberg-Marquardt algorithm based on measured data, using powerful mapping ability of neural network. Compared with experimental data, accuracy of the torque model of SRM based on BP neural network with Levenberg-Marquardt algorithm is proved. In order to predict switched reluctance drive (SRD) performance, a complete simulation model of double closed ring of speed and current for SRD on MATLAB based on BP neural networks nonlinear model is developed in this paper. Experimental phase current waveform and phase current simulation waveform verify the validity of this SRD simulation model.