On-line modelling of switched reluctance motor for high performance current control

This paper considers the implementation of high performance control for switched reluctance motors (SRMs) and presents a novel approach to the accurate on-line modeling of an SRM. An adaptive B-spline neural network is used to learn the nonlinear flux-linkage, torque, incremental inductance, and back EMF characteristics of an SRM. The training of the B-spline neural network is accomplished on-line and in real-time. The system does not require a priori knowledge of the machine´s electromagnetic characteristics. The potential of the method is demonstrated in simulation and experimentally using a 550 W 8/6 4-phase SRM.