Multi-objective Optimization of Permanent Magnet Adjustable Speed Driver Base on RSM Model and NSGA-II

Abstract

To optimise the structure of a permanent magnet adjustable speed driver (PMASD), a multi-objective optimization design method for increasing the output torque and reducing the eddy current loss was proposed. Firstly, the three-dimensional finite element method model of a PMASD was established, the influence of the main configuration parameters in the PMASD on the output torque and the eddy current loss was analyzed, and the reasonable range of the parameters was determined. When taking the minimal eddy current loss and maximum output torque as the optimal objectives, the secondary response surface numerical model equation was built using the Central Composite Design experimental method and the Response Surface Methodology. Then, while ensuring that the output torque of the PMASD is not less than the rated torque, the NSGA-II was used to perform multi-objective optimization based on the response surface model and the Pareto optimal solution sets for two objectives was obtained. Finally, the minimal eddy current loss model and maximum output torque model were selected to compare with the initial model that was not optimized. The simulation results proved that the output torque of the minimal eddy current loss model increased by 6.54% based on the reduction of eddy current loss by 0.81% and the output torque of the maximum output torque model increased by 24.41% based on an increase in eddy current loss of 10.51%: the performance of both optimization models had been improved significantly. The optimization results show that this method improves the transmission performance of the PMASD.