Topology Optimization for a Spoke-Type Permanent Magnet Synchronous Motor Based on a Siamese Convolutional Network

Combining stochastic algorithm-based topology optimization methods with deep neural networks holds great promise for enhancing the performance of electric machines. In this paper, we address the optimization of rotor structure with flux barriers in a spoke-type permanent magnet synchronous motor (PMSM) and introduce a siamese convolutional network based topology optimization method (SCNTO). After training the Siamese convolutional network using a cross-sectional image dataset, predetermined conditions are designed to filter a subset for further assessment using topology optimization. Experiments demonstrate that the proposed method achieves a smooth rotor shape within specified ranges of average torque and torque ripple, while minimizing iron loss with reduced computational costs.