@inproceedings{cf1a881c089045e1bd2d2d66edabec1c,
title = "An Accurate Physics-Informed Neural Networks for 2-D Time Domain Electromagnetic Modeling",
abstract = "This paper presents an enhanced Physics-Informed Neural Network (PINN) framework for accurately modeling electromagnetic wave propagation in heterogeneous media. The proposed method utilizes smooth sigmoid-based transitions to handle material interfaces, addressing the limitations of traditional PINN approaches that struggle with abrupt changes at material boundaries due to their inherent smoothness constraints conflicting with physical discontinuity requirements. Numerical experiments confirm the framework's improved accuracy for complex dielectric structures and enhanced computational efficiency compared to standard techniques.",
keywords = "Maxwell's equations, deep learning, physics-informed deep model",
author = "Yu, \{Feng Lin\} and Shu Wang and Liang, \{Zi Yang\} and Xin, \{Xi Min\} and Gao, \{Hong Wei\}",
note = "Publisher Copyright: {\textcopyright} 2025 Applied Computational Electromagnetics Society.; 2025 International Applied Computational Electromagnetics Society Symposium, ACES-China 2025 ; Conference date: 08-08-2025 Through 11-08-2025",
year = "2025",
doi = "10.23919/ACES-China66523.2025.11333161",
language = "English",
series = "2025 International Applied Computational Electromagnetics Society Symposium, ACES-China 2025 - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
booktitle = "2025 International Applied Computational Electromagnetics Society Symposium, ACES-China 2025 - Proceedings",
address = "United States",
}