Multi-level robust surrogate-based optimization applied to design of electrical machines

This paper introduces a multi-level robust surrogate-based optimization (MRSBO) applied to design of electrical machines. Manufacturing tolerances, material properties, operational and environmental conditions may generate uncertainties that can significantly affect machine performance. The framework developed includes the finite element method (FEM), statistical uncertainty quantification (SUQ) and optimization algorithms. The switched reluctance motor (SRM) is chosen to test the effectiveness of the approach and results suggest that the method is viable for robust design of electrical machines.