This study aims to propose a hybrid optimization framework for asymmetric permanent-magnet motors, integrating Boolean geometry projection (BGP)-based parameter optimization and dynamic Gabor filter-based topology optimization. The objective is to improve forward-direction torque and manufacturability by enabling multiple-magnet configurations and highly anisotropic magnetic-core geometries.
Two optimization problems were formulated: (1) the minimization of cogging torque and maximization of forward torque at two operating points and (2) the inclusion of the reverse-torque performance for bidirectional operation. Magnet configurations were modeled using BGP, and magnetic-core topologies were optimized using the dynamic Gabor filter. The results were compared with a conventional hybrid method based on normalized Gaussian network topology optimization.
Both optimized motors exhibited significant reductions in cogging torque and improved forward torque compared with the reference model. The second optimization problem achieved smooth bidirectional torque with reduced ripple, whereas the first showed increased reverse-torque ripple. An analysis of the torque waveforms and flux distributions revealed that asymmetric flux concentration and stable reluctance-torque generation in anisotropic cores produced the performance gains.
The proposed framework uniquely combines BGP and the dynamic Gabor filter to achieve multiple-magnet, manufacturable and anisotropic motor designs, which has not been attained by existing hybrid optimization techniques.
