CPU-time Reduction for Electric Machine Design by the Response Surface Methodology

This article presents a fundamental study on the application of the response surface methodology to design optimization of electric machines as an effective computational approach. As usual, in designing electric machines, it is difficult to achieve an effective optimal solution by using the FEM within an acceptable CPU-time. In this article, the response surface methodology (RSM) is introduced into the design optimization approach to evaluate the objective physical quantities in much shorter time. In the RSM process, to achieve the high quality optimal solutions, it is very important to make the response surface of high quality approximation in search domains. For that reason, Design of Experiments (DOE) is applied to obtain the appropriate selections of the FEM calculating points. Additionally, the effective iterative approach is developed, which properly narrows the search domains around the solution candidates. The proposed optimization approach results in an overall increase in the optimization speed without degrading the quality of optimal solutions. Some numerical examples that demonstrate the validity of the proposed approach are also presented.