뉴로퍼지기법에 의한 SRM의 맥동토오크 최소화

A Neuro-Fuzzy Based Torque Ripple Minimization of Switched Reluctance Motors

A neuro-fuzzy based torque profile model of SRM with considerably improved accuracy is obtained using the measured data for training. The inferred torque profiles, which comprise magnetic non-linearities, represent the dynamic model of SRM. Then the reference torque signal with optimized waveform and switching angle are decided to control the torque directly. Hence, the presented scheme controls the torque in an instantaneous basis, allowing powerful torque control with minimum torque ripple even during the transient operation of the motor. Simulation and experimental results demonstrating the effectiveness of the proposed torque control scheme are presented.