Modeling of switched reluctance motor based on π-σ neural network

Flux linkage of switch reluctance motor is in nonlinear function of both rotor position and phase current. Establishing this nonlinear mapping is the base to compute the mathematical equations of switch reluctance motor accurately. In this paper, the pi-sigma neural network is employed to develop the nonlinear model of switch reluctance motor. By taking advantage of the benefit of neural network and Takagi-Sugeno type fuzzy logic inference, the pi-sigma neural networks has a simple structure, less training epoch, fast computational speed and a property of robustness. Compared with the training data and generalization test data, the output data of the developed model are in good agreement with those data. The simulated current wave is also in good agreement with the measured current wave. This proves that the model developed in this paper has high accuracy, strong generalization ability, fast computational speed and characteristic of robustness.