Physics-Informed Deep Neural Networks for Transient Electromagnetic Analysis
In this paper, we propose a deep neural network based model to predict the time evolution of field values in transient electrodynamics. The key component of our model is a recurrent neural network, which learns representations of long-term spatial-temporal dependencies in the sequence of its input d...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2020-01-01
|
| Series: | IEEE Open Journal of Antennas and Propagation |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9158400/ |
| id |
doaj-836a34a2ce3f4f0abdc48e6039609142 |
|---|---|
| record_format |
Article |
| spelling |
doaj-836a34a2ce3f4f0abdc48e60396091422021-03-29T18:56:13ZengIEEEIEEE Open Journal of Antennas and Propagation2637-64312020-01-01140441210.1109/OJAP.2020.30138309158400Physics-Informed Deep Neural Networks for Transient Electromagnetic AnalysisOameed Noakoasteen0https://orcid.org/0000-0001-6655-3378Shu Wang1https://orcid.org/0000-0003-0602-708XZhen Peng2https://orcid.org/0000-0001-9907-017XChristos Christodoulou3Department of Electrical and Computer Engineering, University of New Mexico, Albuquerque, NM, USADepartment of Electrical and Computer Engineering, University of New Mexico, Albuquerque, NM, USADepartment of Electrical and Computer Engineering, University of Illinois at Urbana–Champaign, Urbana, IL, USADepartment of Electrical and Computer Engineering, University of New Mexico, Albuquerque, NM, USAIn this paper, we propose a deep neural network based model to predict the time evolution of field values in transient electrodynamics. The key component of our model is a recurrent neural network, which learns representations of long-term spatial-temporal dependencies in the sequence of its input data. We develop an encoder-recurrent-decoder architecture, which is trained with finite difference time domain simulations of plane wave scattering from distributed, perfect electric conducting objects. We demonstrate that, the trained network can emulate a transient electrodynamics problem with more than 17 times speed-up in simulation time compared to traditional finite difference time domain solvers.https://ieeexplore.ieee.org/document/9158400/Computer visionelectromagneticsfinite difference methodsmachine learningrecurrent neural networksunsupervised learning |
| collection |
DOAJ |
| language |
English |
| format |
Article |
| sources |
DOAJ |
| author |
Oameed Noakoasteen Shu Wang Zhen Peng Christos Christodoulou |
| spellingShingle |
Oameed Noakoasteen Shu Wang Zhen Peng Christos Christodoulou Physics-Informed Deep Neural Networks for Transient Electromagnetic Analysis IEEE Open Journal of Antennas and Propagation Computer vision electromagnetics finite difference methods machine learning recurrent neural networks unsupervised learning |
| author_facet |
Oameed Noakoasteen Shu Wang Zhen Peng Christos Christodoulou |
| author_sort |
Oameed Noakoasteen |
| title |
Physics-Informed Deep Neural Networks for Transient Electromagnetic Analysis |
| title_short |
Physics-Informed Deep Neural Networks for Transient Electromagnetic Analysis |
| title_full |
Physics-Informed Deep Neural Networks for Transient Electromagnetic Analysis |
| title_fullStr |
Physics-Informed Deep Neural Networks for Transient Electromagnetic Analysis |
| title_full_unstemmed |
Physics-Informed Deep Neural Networks for Transient Electromagnetic Analysis |
| title_sort |
physics-informed deep neural networks for transient electromagnetic analysis |
| publisher |
IEEE |
| series |
IEEE Open Journal of Antennas and Propagation |
| issn |
2637-6431 |
| publishDate |
2020-01-01 |
| description |
In this paper, we propose a deep neural network based model to predict the time evolution of field values in transient electrodynamics. The key component of our model is a recurrent neural network, which learns representations of long-term spatial-temporal dependencies in the sequence of its input data. We develop an encoder-recurrent-decoder architecture, which is trained with finite difference time domain simulations of plane wave scattering from distributed, perfect electric conducting objects. We demonstrate that, the trained network can emulate a transient electrodynamics problem with more than 17 times speed-up in simulation time compared to traditional finite difference time domain solvers. |
| topic |
Computer vision electromagnetics finite difference methods machine learning recurrent neural networks unsupervised learning |
| url |
https://ieeexplore.ieee.org/document/9158400/ |
| work_keys_str_mv |
AT oameednoakoasteen physicsinformeddeepneuralnetworksfortransientelectromagneticanalysis AT shuwang physicsinformeddeepneuralnetworksfortransientelectromagneticanalysis AT zhenpeng physicsinformeddeepneuralnetworksfortransientelectromagneticanalysis AT christoschristodoulou physicsinformeddeepneuralnetworksfortransientelectromagneticanalysis |
| _version_ |
1724196280070045696 |