Electromagnetic Inversion Algorithm Based on Convolutional Neural Network

摘要

In order to efficiently reconstruct the electromagnetic parameters of the lossy medium in half-space, a novel method based on convolutional neural network (CNN) was proposed to reconstruct the relative permittivity and position information of the lossy medium. First, a data set was constructed by performing two-dimensional forward simulation on lossy media with different electromagnetic parameters; second, the electric field time domain response sequence was taken as the input to the network, and the corresponding lossy medium position and dielectric constant were used as the output of the neural network to build a neural network model for training; in the end, the trained neural network model was used to invert the lossy medium with unknown location and unknown dielectric constant. The experimental results show that the algorithm has high accuracy in reconstructing the electromagnetic parameters of lossy media in a two-dimensional half-space. The numerical results show the effectiveness and accuracy of the method. Therefore, this study provides an efficient method for real-time inverse scattering research of targets.