Skip to main content
APEC 2026 Main banner

Device and Component Modeling

T27.2 - High-Fidelity Modeling of Dynamic Magnetic Components for Power Electronics with Double-Pulse Testing, Automated Circuit Simulation and Deep Learning - WITHDRAWN

Thursday, March 26, 2026
8:40 AM - 9:00 AM CT
Location: 214C
This study proposes a data-driven modeling approach for magnetic components by integrating double pulse testing (DPT) with deep learning. Traditional modeling methods often rely on idealized assumptions, making it difficult to reproduce parasitic parameters in high-frequency regions precisely and requiring considerable effort for model tuning. The proposed method efficiently captures the nonlinear and dynamic behavior of magnetic components using DPT waveforms. It employs convolutional neural networks (CNNs) to regress parasitic parameters from time-series voltage and current waveforms. The results demonstrate the feasibility of accurate parameter estimation using this approach.