Due to the complex rotor design of reluctance synchronous machines, finite element analysis is crucial for accurately calculating performance objectives such as mean torque and torque ripple. This requires numerous simulation steps, leading to high computational demands. An efficient optimization algorithm is therefore essential. This paper introduces a novel framework for single-objective machine design optimization using Gaussian Process Regression (GPR) and Bayesian Optimization (BO). Various kernel functions and hyperparameters are assessed to evaluate regression accuracy for objectives like mean torque, torque ripple, and power factor. Bayesian Optimization demonstrates faster and superior results compared to traditional methods like genetic or particle swarm algorithms, especially for designs with 18 variables.

Find the paper here: Link