A Computationally Efficient High-Fidelity Multi-Physics Design Optimization of Traction Motors for Drive Cycle Loss Minimization

Abstract

Continuous improvement in performance of interior permanent magnet (IPM) machines is critical for electric vehicle traction applications. However, due to the cross-coupling and saturation effects, a significant amount of time-consuming finite element analysis (FEA) simulations are required to accurately estimate machine performance. Moreover, iterative design optimization will take significantly longer. In this article, an improved rapid performance estimation technique utilizing surrogate models is developed and coupled with a design optimization algorithm. The proposed framework has significantly less computational cost than alternative surrogate-based approaches, and efficiently employs drive cycle loss minimization for a multi-physics, multi-objective traction motor design optimization. Simulation and experimental results suggest the proposed optimization framework yields optimal designs more efficiently than existing methods while maintaining accuracy.