Implementing the Fast Full Wave Electromagnetic Forward Solver Using the Deep Convolutional Encoder-Decoder Architecture

In this paper, a novel deep learning (DL) based solver is proposed for the electromagnetic forward (EMF) process. It is based on the complex-valued deep convolutional neural networks (DConvNets) comprising an encoder network and a corresponding decoder network with pixel-wise regression layer. The e...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on antennas and propagation Vol. 71; no. 1; p. 1
Main Authors: Yao, He Ming, Jiang, Lijun, Ng, Michael
Format: Journal Article
Language:English
Published: New York IEEE 01.01.2023
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Artificial neural networks
Coders
Computational modeling
Convolutional Neural Network
Decoding
Deep Learning
Electromagnetic Forward Process
Electromagnetics
Encoders-Decoders
Mathematical models
Real Time
Real-time systems
Solvers
Three-dimensional displays
Training
Transformations (mathematics)
ISSN:0018-926X, 1558-2221
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, a novel deep learning (DL) based solver is proposed for the electromagnetic forward (EMF) process. It is based on the complex-valued deep convolutional neural networks (DConvNets) comprising an encoder network and a corresponding decoder network with pixel-wise regression layer. The encoder network takes the incident EM wave and the contrast (permittivity) distribution of the object as the input. It channels the processed data into the corresponding decoder network to predict the total EM field due to the scatter of the input incident EM wave. The training of the proposed DConvNets is done using the simple synthetic dataset. Due to its strong approximation capability, the proposed DConvNets can realize the prediction of EM field. Hence, the proposed DL based EMF solver acts as a "inhomogeneous" transformation - the unknown EM field in the objective domain is obtained through the transformation from the information of the incident EM field and the distribution of contrasts (permittivities). Compared with conventional methods, the EMF problem can be solved with higher accuracy and significantly reduced CPU time. Numerical examples have demonstrated the feasibility of this newly proposed approach. This newly proposed deep learning based EMF solver presents a new alternative to electromagnetic computation approaches.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-926X
1558-2221
DOI:10.1109/TAP.2022.3216920