Data-Driven Simulation of Traction Electrical Machines: a Modelling Strategy for Multi-Physics Dataset Generation

The design of traction electrical motors faces increasing challenges in satisfying requests for higher efficiency, speed, torque and cost reduction. Furthermore, the design of new machines must deal with the interaction of multiple physical domains, including electromagnetic, thermal, and structural aspects, leading to high computational costs. The adoption of surrogate data-driven models can significantly accelerate the optimized design of traction electrical machines. To this end, we propose a modelling strategy for the multi-physics dataset generation to build a benchmark for data-driven simulations.