High-dimensional optimal design of dual-rotor synchronous reluctance machines based on data-driven torque decomposition

Abstract

The multi-objective optimal design of double-sided stator dual-rotor synchronous reluctance machines (DSS-DRSynRMs) is a challenging high-dimensional problem. The objective of this paper is to present a new optimal design method based on data-driven models and the principle of torque decomposition addressing the aforementioned issue. For this purpose, a 26-parameter optimisation problem is solved by employing the proposed method consisting of three sequential phases. Through the proposed method, the combination of artificial neural network (ANN) and recently introduced waveform targeting surrogate model (WTSM) strategy is investigated to mitigate the computational complexity of the optimisation process. Furthermore, the electromagnetic performance of the final optimal design has been comprehensively analysed showing a significant reduction in torque ripple rate and improved torque density. Moreover, the computational efficiency of the proposed method has been compared to the popular multi-level multi-objective optimisation method. From the discussion, it can be found that the proposed method provides a reduced computation time and wider search space.