MagNet AI Challange
MagNet Challange
The MagNet Challenge is a global academic competition jointly initiated by the Princeton University Power Electronics Laboratory, the IEEE Power Electronics Society, and renowned companies such as Google and Enphase.
The aiming of this challenge is to employ neural networks for accurately modeling the power loss in magnetic materials, replacing the traditional approximate experiential modeling methods of GSE and IGSE. This improvement will enable engineers to design motors, inductors, and other power devices in systems like electric vehicles and power grids more efficiently, thereby enhancing performance and saving energy.
Award
1st place in few-shot-learning, 5% accuracy ahead of the 2nd place.
3rd place in model overall performance.
Lead 20 university in world including Tsinghua(清华)
Been invited to APEC 2024 and ECCE Europe 2024 to present.
2 paper research outcome Click to view
Our work is open sourced on GitHub, Please visit
The project is keep running, and our objective is to establish an IEEE industry standard that revolutionizes the method for modeling magnetic materials.