T16.7 - Physics-Informed Neural Network Based Modeling of Magnetic Couplers for EV Wireless Charging System

Wednesday, March 25, 2026

11:00 AM - 11:20 AM CT

Location: 217D

Technical Lecture Presenter(s)

Chang Li (he/him/his)

Beijing insitute of techonology

Presentation

Co-Author(s)

JD

Junjun Deng

Beijing Institute of Technology
Beijing Institute of Technology
YZ

Yi Zheng

Beijing insitute of techonology
MD

Mengchen Duan

Beijing insitute of techonology
SW

Shuo Wang

Beijing Institute of Technology

This work proposes a physics-informed neural network (PINN) framework for fast and accurate inductance modeling of magnetic couplers in EV wireless charging systems. Unlike traditional FEM, which is computationally intensive, or data-driven surrogates that ignore physical constraints, the proposed PINN embeds the quasi-static A-form of Maxwell’s equations, boundary/interface conditions, and material effects, with light FEM calibration for scaling. The method models DD-shaped and rectangular coils with ferrite and aluminum shielding and predicts inductive parameters directly from geometry. Experiments at 85 kHz show percent-level accuracy compared with FEM and measurements, while reducing computation time from tens of minutes to only a few minutes per design.