Torque Ripple Reduction With Harmonics Superposition and Optimization for Axially-Segmented PM-Assisted Synchronous Reluctance Machines

In conventional permanent magnet-assisted synchronous reluctance machines, the permanent magnet (PM) torque and reluctance torque are coupled with each other due to the rotor structure, and the ripples of these two torque components are superposed, with few design solutions being applicable to reduce the overall torque ripple. In this paper, an axially-segmented permanent magnet-assisted synchronous reluctance machine, which has a reluctance segment and a PM segment on the rotor, is studied. A method of torque ripple reduction using harmonics superposition is proposed, in which the two torque components generated by the two rotor segments are matched in such a way that their main harmonics have appropriate amplitudes and phases to cancel each other. Motor optimizations with surrogate models based on neural network fitting and interpolation are performed to achieve maximum suppression effect of the torque ripple. Finite element analysis and prototype experiments are conducted, confirming the effectiveness of the method.