Automated identification of the origin of energy loss in nonoriented electrical steel by feature extended Ginzburg–Landau free energy framework

Abstract

Abstract This study presents the automated identification of the complex magnetization reversal process in nonoriented electrical steel (NOES) using the feature extended Ginzburg–Landau (eX-GL) free energy framework. eX-GL provides a robust connection between microscopic magnetic domains and macroscopic magnetic hysteresis using a data science perspective. This method employs physically meaningful features to analyze the energy landscape, providing insights into the mechanisms behind function. We obtained features representing both the microstructure and energy of the domain wall. The causes of iron loss were traced to the original domain structure, through which we could successfully distinguish and visualize the role of pinning as a promoting and resisting factor. We found that the reversal process was governed not only by general grain boundary pinning but also by segmented magnetic domains within the grain. This method revealed the complex interplay between magnetism and metallography and introduced a new means for transformative material design, bridging structures and functions.