Recently, the 10th International Power Electronics and Motion Control Conference (IPEMC 2024-ECCE Asia), one of the important conferences in the field of power electronics and motion control, was held in Chengdu, Sichuan Province. The conference selected 8 excellent papers, 8 excellent posters, and 10 best session organizers. Among them, Hu Tianxiang, a 26’ doctoral student in Electronic Information, Yao Yinan, a 26’ master’s student in Electrical Engineering, Xu Lumeng, a 24’ undergraduate in Computer Engineering, Wang Yiyi and Wang Sichen, 24’ undergraduates in Electrical Engineering, of ZJUI, won the Excellent Poster Award for their paper titled "PINN Based Data Driven Magnetics Loss Modeling" under the guidance of Assoc Professor Li Chushan and Assist Professor Liu Zuozhu.

▲ Yao Yinan received the award on behalf of the research team

The conference received a total of 1249 submissions from 27 countries and regions. After rigorous review by more than 600 experts, 902 full papers were accepted. The conference featured 56 paper presentations, 2 poster sessions, and more than 20 industrial exhibitions, with 335 authors presenting papers and 567 posters displayed (source: IPEMC 2024-ECCE Asia).

Magnetic components are essential in power electronic converters for energy storage, energy conversion, and electrical isolation. Establishing an accurate loss model for magnetic components is crucial for material selection, thermal design, performance prediction, and system optimization. However, magnetic material losses are non-linear and complex, making it challenging to achieve highly accurate characterization through analytical methods. While data-driven approaches can build black-box models to predict losses, deep learning methods require large amounts of data and complex models. The award-winning paper combines the advantages of physical models and data-driven models, integrating physical mechanisms with data methods. The research team established a magnetic loss model based on the Improved Generalized Steinmetz Equation (IGSE), which allows for loss calculation under any conditions. Additionally, they used the core concept of Physics-Informed Neural Networks (PINN), embedding IGSE into the neural network to achieve accurate loss predictions with fewer parameters and less data.

▲ The architecture of PI-GRU+FNN model

▲ PINN largely reduced the demand for the amount of training data

IPEMC 2024-ECCE Asia was hosted by the Chinese Electrotechnical Society (CES) and organized by Southwest Jiaotong University, with co-sponsorship from IEEE Power Electronics Society (PELS), IEEJ Industry Applications Society (IEEJ-IAS), and Korean Institute of Power Electronics (KIPE). The conference aims to provide an interactive platform for experts, scholars, technical personnel, and young students from academia and industry worldwide to discuss and share the latest research results and developments in power electronics and motor motion control (source: IPEMC 2024-ECCE Asia).

The research team is composed of doctoral, master’s, and undergraduate students from ZJUI. This interdisciplinary, cross-grade student team was formed as a result of the international academic competition "Magnet Challenge" initiated by IEEE Power Electronics Society, Princeton University, and companies like Google and Enphase. The competition explores innovative methods combining high-frequency magnetics technology in power electronics with artificial intelligence. In the competition, the research team proposed a PINN model based on the improved Steinmetz equation, earning Honorable Mention in the IEEE International Challenge in Design Methods for Power Electronics. These research results were eventually published at the IPEMC 2024 conference.

This interdisciplinary collaboration and innovative research represent a successful case of academic and professional exchange, bringing new vitality to the development of the power electronics discipline at ZJUI and promoting research innovation. The award at this internationally renowned conference demonstrates the academic quality and innovative spirit of ZJUI students, highlighting ZJUI’s distinctive feature in international student cultivation and interdisciplinary innovation.

▲ Yao Yinan and the post at the exhibition