From left to right: Yaow-Ming Chen of National Taiwan University and Chief Editor of IEEE Transactions on Power Electronics, Haoran Li of Princeton University, Shukai Wang of Princeton University, Henry Chung of the City University of Hong Kong and past Editor of IEEE Transactions on Power Electronics Letters, and Minjie Chen of Princeton University. Image provided by Minjie Chen.
A team of researchers at Princeton University led by Minjie Chen, associate professor co-appointed in the Department of Electrical and Computer Engineering and the Andlinger Center for Energy and the Environment, has been awarded the First Place Prize Paper Award for 2023 in the IEEE Transactions on Power Electronics (TPEL).
The paper, “How MagNet: Machine Learning Framework for Modeling Power Magnetic Material Characteristics,” introduces machine learning as a tool to model power magnetics, a material which is as complex as a protein, said Chen. Through machine learning modeling, the researchers hope to accelerate the design process for the next generation of power electronics. “This paper changed how people think about the modeling of magnetics,” said Chen.
Over the course of the last four years, Chen had been working on this project with a number of co-authors including Professor of Engineering Charles Sullivan from Dartmouth College, Professor of Electrical and Computer Engineering Niraj Jha from Princeton, and Associate Professor of Statistics and Data Science Yuxin Chen from the University of Pennsylvania. “This is quite an amazing effort,” said Shukai Wang, a PhD candidate in electrical and computer engineering and paper co-author. “This project bridges the gap between hardcore power electronics and machine learning and computer science.”
“This is a good example of cross-disciplinary research which brings our power electronics community more attention with machine learning tools” said Haoran Li, who is also a PhD candidate in electrical and computer engineering and paper co-author. “This work and especially the open-sourced database serves as a foundation to stimulate further ideas in this area.”
The effort included collecting massive amounts of data to train machine learning models and then building a website as an open-source platform for demonstrating the model. To build the website, the team turned to Vineet Bansal, senior research software engineer with the Center for Statistics and Machine Learning. Bansal is named as an author on the paper. “Vineet was the backbone of our website,” said Wang.
Bansal’s work was instrumental to getting the site up and running. “Users from all over the world go to this website,” said Bansal. “You need to take care that it runs fast enough and it's not going to get overloaded with hundreds of people accessing it at the same time.”
Every year, researchers send thousands of articles for selection to be published in IEEE TPEL. Only about 1,000 of those articles are published a year. For 2023, IEEE TPEL selected five of those published papers for their first prize paper award.
“This is a major award,” said Bansal. “It is a huge accomplishment.”