Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

Power Electronics Design Utilizing Machine Learning for Thermals

(This is a translated version of an article posted by Kosuke Ikeda. The original article in Japanese can be found here)

At the 1DCAE & MBD Symposium 2024, held in Hiroshima from November 28 to 29, 2024, a presentation was given on the theme “Power Electronics Design Leveraging Machine Learning and 1D Simulation.”

Symposium｜1DCAE

The electronics industry is becoming increasingly important across various fields, including aircraft, vehicles, and household appliances. However, to design these systems efficiently and compactly, it is essential to address the major challenge of thermal management.

In this presentation, we introduce a new thermal design method for power electronics that leverages machine learning and 1D simulation while taking into account thermofluid characteristics.

As a specific example, we modeled a household washing machine using the following approach:

The motor drive was simulated using the power electronics tool Altair PSIM.
The housing was modeled with the multibody dynamics tool Altair MotionSolve.
The cooling system was analyzed using the 3D thermal-fluid simulation tools Altair SimLab / ElectroFlo.

Furthermore, by applying the model order reduction technology “romAI” based on machine learning, we integrated all these models into the 1D simulation tool Altair Twin Activate for coupled analysis.
This approach enabled precise thermal management simulation of switching devices under real operating conditions, achieving both improved design efficiency and performance optimization.

This method has the potential to greatly contribute to enhancing product reliability and extending lifespan, and it is expected to serve as a new design guideline for the electronics industry with many possible applications.