Fast Full-Wave Electromagnetic Forward Solver Based on Deep Conditional Convolutional Autoencoders

This letter proposes a novel deep learning (DL) based fast solver for the electromagnetic forward (EMF) process. This proposed fast full-wave solver for EMF process is designed based on the deep conditional convolutional autoencoder (DCCAE), consisting of a complex-valued deep convolutional encoder network and its corresponding complex-valued deep convolutional decoder network. The encoder network makes use of the input consisting of the incident electromagnetic (EM) wave and the contrast (permittivities) distribution of the target domain, while the corresponding decoder network predicts the total EM field illuminated by the input incident EM wave. The training of the proposed DCCAE solver for EMF is merely based on the simple synthetic dataset. Thanks to its strong approximation capability, the proposed DCCAE can realize the prediction of the EM field of target domain by using the incident EM field and the distribution of contrasts (permittivities). Therefore, compared with conventional methods, the EMF problem could be solved with higher accuracy and the significant reduced computation time. Numerical examples have illustrated the feasibility of the newly proposed DL-based EMF solver. The newly proposed DL-based EMF solver presents its excellent performance for the real-time online application.

Automated summary

This letter proposes a novel deep learning-based fast solver for the electromagnetic forward process. The solver is based on a deep conditional convolutional autoencoder (DCCAE) consisting of a complex-valued deep convolutional encoder network and its corresponding decoder network. The proposed solver can accurately predict the electromagnetic field of a target domain in real-time applications, significantly reducing computation time compared to conventional methods.