A new method based on machine learning for the FDTD method

In this paper, a novel finite difference time domain method based on neural network is proposed, which uses convolutional neural networks (CNN) and long short-term memory networks (LSTM) to establish a new LSTM-CNN model and provides a new approach for solving complex electromagnetic problems. Long short-term memory networks are used to iterate the time steps in the finite difference time domain method, while convolutional neural networks are used to extract spatial local features in the finite difference time domain method. Compare the new model with previous models and verify its advantages through numerical examples. The results show that the relative error of the new method is smaller.