Yuki Shimizu
Graduate School of Science and Engineering,
Ritsumeikan University
Abstract
Finite element analysis needs to be performed repeatedly to design the optimal motor shape, and therefore, the computation time becomes longer. To solve this problem, “surrogate models” which accurately predict the motor characteristics using machine learning have been attracting attention, and it is possible to perform the design optimization without using finite element analysis. We proposed an automatic design system using a deep generative model to consider various magnet arrangements. This lecture explains the data generation using JMAG, the training method for the system, and the optimization results.
Protected content here, for members only.
You need to sign in as a Regular JMAG Software User (paid user) or JMAG WEB MEMBER (free membership).
By registering as a JMAG WEB MEMBER, you can browse technical materials and other member-only contents for free.
If you are not registered, click the “Create an Account” button.
