Topology Optimization of the IPMSM for HEVs Using a Normalized Gaussian Network With Divided Region Method for Weighting Coefficients

Abstract

In this paper, topology optimization of the rotor flux-barrier of an interior permanent magnet synchronous motor using a normalized Gaussian network is proposed to improve traction in a hybrid electric vehicle. The flux-barrier that impacts on the electromagnetic performance such as average torque, torque ripple and power is important in the design of electromagnetic motors. The normalized Gaussian network was used to obtain the shape of the optimal flux-barrier, and the weighting coefficients used in this method were optimized using a genetic algorithm. To prevent indiscriminate creation of the flux-barrier, the design region was divided into detailed subregions. The weighting coefficients were generated using different methods in each subregion. In addition, to increase the mechanical stability of the shape by eliminating unstable shapes, the post-processing is performed. The superiority of the performance due to optimizing the topology was confirmed, with the average torque increased by 8.43% and the torque ripple decreased by 20.07%p. The mechanical stability of the optimal shape is verified by conducting stress analysis.

Keywords

• Design optimization
• genetic algorithm
• hybrid electric vehicle
• motor design
• normalized Gaussian network
• shape function