Ultra-high-dimensional multi-level optimisation strategies for electrical machines

Electrical machine optimisation is normally a high-dimensional non-linear multi-objective optimisation problem. A multi-level optimisation (MO) strategy is currently used to improve efficiency, where sensitivity analysis is required for dividing design parameters into different groups. However, the conventional MO strategy cannot handle ultra-high-dimensional optimisation problems. In this paper, a sensitivity analysis method with variable weighted intervals is proposed to calculate the sensitivity coefficient in the parameter design range. Moreover, three improved multi-level optimisation strategies based on different optimisation algorithms, sequential sensitivity strategies, and machine learning models are proposed, analysed, and compared with the conventional MO strategy. Through a case study of a synchronous reluctance machine, it can be seen that the proposed optimisation strategies can improve the optimisation results and efficiency of ultra-high-dimensional optimisation of electrical machines. Electrical machine optimisation is a complex high-dimensional non-linear multi-objective problem. This paper proposes a new sensitivity analysis method with variable weighted intervals and three improved multi-level optimisation strategies, utilising different optimisation algorithms, sequential sensitivity strategies, and machine learning models. A case study of a synchronous reluctance machine demonstrates that these strategies enhance optimisation results and efficiency for ultra-high-dimensional optimisation problems. image