Surrogate Model-Based Optimization Methodology for High Torque and Power Density Permanent Magnet Assisted Synchronous Reluctance Motor

This paper presents a design optimization methodology for a multi-layer permanent magnet assisted synchronous reluctance motor for hybrid and electric vehicles. This methodology is based on a surrogate model generated using 2D finite element simulations. Compared to direct approaches, the proposed methodology allows to reduce computation time by limiting the number of finite element simulations needed to build the surrogate model while maintaining a very good accuracy with a maximum relative error of 1%. The new methodology was used to design a UV-shaped permanent magnet assisted synchronous reluctance motor achieving high torque and power densities (9.5 Nm/kg and 9.7 kW/kg) under constraint of manufacturing process, cooling complexity and available AC current.