Optimal design for electromagnetic machinery by using 3D finite element method with parallel GA

Abstract

Purpose

The purpose of this paper is to automate the pre‐process that is the modeling and mesh generation of the 3‐dimensional finite element method; and genetic algorithm (GA) is embedded to this program by using Shell script.

Design/methodology/approach

It is operated by Bash (Bourne‐again shell) to implement the GA. In order to adopt the analysis of 3‐D FEM as evaluation in the GA, the pre‐processes such as moving nodes, 3‐D modeling, and mesh generation must be automated. In the “Evaluation”, these pre‐processes are connected and automated by using shell script. The “Evaluation check” is to prevent re‐evaluation. In the population, there are some individuals that are already evaluated in the previous generations because of elitist strategy or result of genetic operations. In order to shorten computing time for the main process, these individuals are skipped in the “Evaluation”.

Findings

It was confirmed that the cogging torque of SPMSM has been decreased by 30 per cent and the improvement about 2.5 times the detent force of the 2 degree of freedoms actuator. Moreover, GA was able to be made parallel by distributing the gene to several PC, and the total calculation time was able to be shortened by below 1/2.

Originality/value

A novel method of the 3‐D FEM with parallel embedded GA has been proposed and demonstrated to be computationally capable.

Citation

Kitagawa, W., Kimura, Y., Shimizu, T. and Takeshita, T. (2012), "Optimal design for electromagnetic machinery by using 3D finite element method with parallel GA", COMPEL - The international journal for computation and mathematics in electrical and electronic engineering, Vol. 31 No. 5, pp. 1386-1395. https://doi.org/10.1108/03321641211247959