FEL UWB Researchers Use AI to Design Advanced Electric Motors

By combining machine learning, simulations, and 3D printing, researchers at FEL UWB are paving the way for more efficient electrical device design. Their latest results were presented at COMPUMAG 2025 in Naples.

Modern design of electromagnetic devices—such as electric motors and transformers—relies heavily on advanced computational methods. At the Faculty of Electrical Engineering of the University of West Bohemia in Pilsen (FEL UWB), a research team led by Pavel Karban has long been developing mathematical models that integrate traditional simulation techniques with artificial intelligence methods. The result is a tool that enables the design of more efficient devices tailored to specific operational conditions.

“Whether optimizing additively manufactured materials or simulating complex devices, robust and accurate models are key,” said Pavel Karban. “Machine learning helps us avoid dead ends and discover new, unconventional solutions,” he added.

A crucial part of their work is topological optimization—finding the most suitable shape and structure of a device to match its function. This approach is applied to 3D-printed magnetic materials, which represent a significant leap over conventional manufacturing. The result is the ability to design structurally unique electrical machines that would be impossible to create with traditional methods.

“There is enormous potential in additive metal manufacturing for producing non-standard electrical machines,” explained researcher Jan Kaska. “Thanks to our cooperation with the Regional Technological Institute of the Faculty of Mechanical Engineering, we have mastered printing with soft magnetic materials. However, high iron losses remain a challenge. Our optimization approach enables the design of new structures that reduce these losses—considering both machine geometry and operational frequencies.”

The research team presented these findings in June at the international COMPUMAG 2025 conference in Naples, Italy. Held since 1976, COMPUMAG is a key global forum for sharing advances in computational electromagnetics. “The participation of FEL UWB’s research team confirms their international relevance and active role in developing new numerical methods and electromagnetic applications in electrical engineering,” concluded Pavel Karban.