Prediction Network of Metamaterial with Split Ring Resonator Based on Deep Learning.

Area of Science:

• Electromagnetics
• Materials Science
• Computational Physics

Background:

• Metamaterials offer significant advancements in electromagnetics.
• Current on-demand metamaterial structure design is a time-intensive process.
• Deep learning demonstrates strong generalization for data classification and regression tasks.

Purpose of the Study:

• To develop an efficient deep learning model for the on-demand design of metamaterial structures.
• To automate the calculation of structural parameters based on desired optical properties, specifically reflectance.

Main Methods:

• A deep neural network was constructed and trained.
• The network takes desired reflectance as input.
• Structural parameters are automatically calculated and output by the trained network.

Main Results:

• The deep neural network achieved low mean square errors (MSE) of 0.005 on both training and testing datasets.
• The model accurately predicts structural parameters for metamaterial design.
• The trained model effectively guides the design process.

Conclusions:

• Deep learning significantly speeds up the metamaterial design process.
• The developed deep learning approach provides a more accurate and convenient method for on-demand metamaterial design.
• This machine learning strategy enhances the efficiency of creating custom metamaterial structures.