Optimized Current Waveform for Torque Ripple Mitigation and MTPA Operation of PMSM with Back-EMF Harmonics based on Genetic Algorithm and Artificial Neural Network
Keywords:Artificial neural network (ANN), genetic algorithm (GA), interior permanent magnet synchronous motor (IPMSM), maximum torque per Ampère (MTPA), nonsinusoidal back-EMF, torque ripple
Interior permanent magnet synchronous motors (IPMSMs) with nonsinusoidal flux develop undesirable torque ripple under conventional control strategies, leading to vibration and increase of mechanical stress. To overcome this problem, active torque ripple compensation based on harmonic stator current injection is widely investigated in the literature. However, the combination of the torque ripple minimization strategy with the maximum torque per Ampère (MTPA) operation is a challenge to the field. In this paper a control strategy based on genetic algorithm (GA) and artificial neural network (ANN) is proposed to achieve low torque ripple with MTPA operation of a three-phase IPMSM with significant magnitude of zero sequence back-electromotive force (back-EMF) component. Investigations have found that the consideration of back-EMF zero sequence in torque prodution can increase the torque per Ampère ratio. Further, the results shows that the proposed strategy effectively reduces the torque ripple in real time operation, including steady state and transient torque performances.