67 / 2023-10-09 20:54:12

Inverse design of electromagnetically induced transparency(EIT) metamaterials based on autoencoder with reconstruction error

Metamaterial,Electromagnetically induced transparency effect,Deep learning,Inverse design

6.AI-driven technology > (2)、Intelligent sensing and data analytics for smart manufacturing

In the paper, we design a new deep learning (DL) network to analysis and design the electromagnetically induced transparency (EIT) metamaterials. The network is divided into two parts: dimensionality reduction reconstruction network and inverse design. The dimensionality reduction reconstruction network mainly deals with high-dimensional input data. In the inverse design, we combine the advantages of convolutional neural network (CNN) and long short term memory (LSTM) networks to predict the EIT metamaterial structure. Finally, the inverse design network has effectively reduced the mean square errors (MSE) on the validation set to 0.0032. Besides, the network can quickly predict the structure parameters within the error of 0.26 µm. By comparing the spectra of real and predicted parameters on the validation set, we are confident that the network will pave a new path for designing EIT metamaterials.