A Physics-informed Dynamic Model for Ferrite Magnetics

Abstract: (Sponsored by TC 10) Accurate modeling of magnetic behavior under diverse excitation conditions remains a longstanding challenge. To address this need, the MagNet Challenge 2 was established as a benchmark problem to evaluate advanced modeling approaches for magnetic materials. The GT&GM team addresses the challenge using a physics-informed model, which features a small number of parameters and high interpretability. The magnetic flux density B and field strength H are mapped to displacement and force in the mechanical domain. Correspondingly, the nonlinear aspects of magnetic materials, such as saturation, pinning force, and eddy current, are captured via mechanical components, like springs, sliders, and dampers respectively. There is one more type of components that seems not being reported previously and significant in reducing modeling errors. In the mechanical domain, it can be interpreted as inertia. Two possible physical roots are provided and both of them relate to displacement current. The methods of prediction, parameter identification, and testing are reported. Last but not least, the presented model is vectorial, which makes it adaptable for finite element analysis.