Topological Design for 3-D Optimization Using the Combination of Multistep Genetic Algorithm with Design Space Reduction and Nonconforming Mesh Connection

Abstract

Topology optimization, which identifies the optimal combination of ON-OFF material, is one of the most attractive tools in the initial conceptual and practical design of electrical machineries for engineers. Heuristic optimization algorithms based on the random search flexibly allow the engineers to define the objective function. However, there are huge iterations of finite element analysis in 3-D topological optimization according to many design variables. It is difficult to show the practical solution without the island and void distributions. This paper presents the 3-D topological optimization method based on multistep utilization of GA. Proposed method can obtain a practical solution, and the convergence is remarkably improved by the design space reduction. Furthermore, the elapsed time for 3-D optimization can be more reduced to a quarter of conventional analysis by the combination of nonconforming mesh connection.