Application of Deep Learning in FDTD Method

In this paper, a novel approach based on deep learning (DL) is proposed to predict electromagnetic transmission problem in two-dimensional free space. The Finite-Difference Time-Domain (FDTD) method is employed to establish training data sets for the proposed model based on U-net. We demonstrate the performance of the proposed approach using experimental tests. It is found that the newly proposed approach could predict the electromagnetic field with satisfactory accuracy.