News Releases 2018Meiden developed a design assistance program for EV motors that harnesses an AI-based topology optimization method in collaboration with Hokkaido University

Meidensha Corporation (Meiden), in collaboration with Hokkaido University, has developed a design assistance program for electric vehicle (EV) motors that harnesses an artificial intelligence-based topology optimization method. The new program can automatically calculate an optimal rotor shape for EV motors to ensure high energy-saving capability and efficiency.

Meiden and Hokkaido University started their joint research in 2012 and completed the design assistance program recently.

The program is for PM motors, which have permanent magnets embedded inside the rotor. Conventionally, designers determine the shape of the magnets and air gaps based on their experience, simulation results or experimental data. The design assistance program, on the other hand, automatically enables designers to find an optimal rotor shape with the assistance of AI.

In this project, researchers established the topology optimization method using a normalized Gaussian network (NGnet) and confirmed its effectiveness in designing EV motors. Existing optimization methods have been unsuitable for practical use because many air gaps emerge in the material used. The new method does not separate the material apart so much, creating fewer air gaps. In addition, the program factors in energy loss and strength of the motor as well as torque (a force that produces rotation) and torque ripple (a periodic increase or decrease in output torque), which are essential characteristics of a motor, to automatically find an optimal shape.

Furthermore, Meiden and Hokkaido University have developed a visualization system to easily revise the complicated optimal design to a simpler, practical design for production. It shows designers which parts cannot be amended because revisions would negatively affect the motor’s performance, and which parts can be altered because doing so would not have such an effect. Previously, designers had difficulty in assessing how their design revisions would affect a motor’s performance.

This program uses AI so it is possible to design shapes that no human being could think of, possibly leading to the design of motors with better performance. It can also help supply EV motors that better cater to client needs, such as a more comfortable ride in the vehicle, noise reduction and better energy-saving capabilities, because it enables designers to create an optimal design by inputting targets and conditions for desired motor performance.

Meiden will proactively use this new program to design and develop motors for EVs and other purposes for its clients.

Note) The topology optimization method searches for the optimal allocation of a material in a given space to ensure high motor performance.

*Shapes shown in this press release were used in research. They are not actual product shapes.

Press contacts:

General Affairs & Public Relations Division
Tel: +81-3-6420-8100