Nonlinear modelling of switched reluctance motors using artificial intelligence techniques

This paper develops and compares different techniques for the modelling of a switched reluctance motor (SRM) in view of its nonlinear magnetisation characteristics due to the doubly salient structure. A complete range of models based on fuzzy logic, neuro-fuzzy and neural network approach is developed. All models are separately simulated and applied for nonlinear modelling, especially for finding the rotor angle positions at different currents, from a suitable measured data set for an associated SRM. The data comprised flux linkage, current and rotor position. All models are constructed to allow them to be modelled as a function of flux linkage against current with rotor position as an undetermined parameter. The models' evaluation results are compared with the measured values, and the error analyses are given to determine the performance of the developed models. The error analyses have shown great accuracy and successful modelling of SRMs using artificial intelligence techniques.