2020 | vol. 68 | nr. 2 | art. 3

Multi-Objective Optimization Design of Permanent Magnet Spherical Motor based on PSO-BP Neural Network

Xiwen Guo, Ronglin Zhang, Qunjing Wang, Yan Wen, Nengwei Gong
Abstract
Aimed at the problems of low output torque of permanent magnet spherical motor (PMSM), an effective multi-objective optimization design method to Improve torque characteristics and reduce limited material volume, improve the motor load characteristics, reduce design costs is present. At first, based on finite element analysis (FEA) data, a particle swarm optimization-optimized BP neural network (PSO-BP) modelling method is proposed, which solves the modelling by drawing the nonlinear and complex relationship between structural parameters and torque problem. Then, an adaptive grid multi-objective particle swarm optimization (AGA-MOPSO) algorithm is proposed to search for the best structural parameters, and a Pareto optimization frontier map are obtained. Finally, according to the FEA simulation analysis, the best solution is selected from the Pareto optimization results. The simulation results show that compared with the original motor, the torque has increased by approximately 43.52% and the volume has decreased by approximately 7.8%, which proves the correctness and feasibility of this method.
Keywords: permanent magnet spherical motor (PMSM), multi-objective design optimization, finite element analysis (FEA), particle swarm optimization-optimized BP (PSO-BP), adaptive grid multi-objective particle swarm optimization algorithm (AGA-MOPSO)
To cite this article: Xiwen Guo, Ronglin Zhang, Qunjing Wang, Yan Wen, Nengwei Gong, “Multi-objective optimization design of permanent magnet spherical motor based on PSO-BP neural network”, in Electrotehnica, Electronica, Automatica (EEA), 2019, vol. 67, no. 2, pp. 23-31, ISSN 1582-5175.

 

Download