Magnetic labyrinth: Artificial Intelligence helped Japanese physicists uncover the secrets of motor overheating

The rapid growth of the electric vehicle market has necessitated increasing the energy efficiency of electric motors. A key technical challenge in this area is magnetic hysteresis, or energy losses in iron. When magnetic fields inside a motor change direction at high frequencies, part of the electrical energy turns into heat, warming the motor core. This process leads to the demagnetization of materials and a decrease in efficiency. This is reported by Ixbt.com reports .

Scientists from the Tokyo University of Science, in collaboration with the universities of Tsukuba, Okayama, and Kyoto, have developed a new theoretical model called eX-GL to solve this problem. The study examined the complex "labyrinth" structure of magnetic domains—microscopic regions within the material. Such structures have the property of undergoing sudden reorganization when the temperature changes, a process that was previously impossible to describe with precise mathematics.

Physicists used Machine Learning and an advanced topological data analysis method—persistent homology—to analyze the process. Artificial Intelligence studied microscopic images of magnetic domains and created a digital map of free energy. This map clearly showed how the microstructure of domains evolves in response to changes in energy states.

As a result of the study, the scientists managed to visualize four main hidden energy barriers that govern the dynamics of magnetic domains. It was found that the complexity of labyrinth domains increases with the total length of domain walls. This discovery will allow for the creation of new magnetic materials for high-efficiency electric motors that lose less energy and do not overheat.