桂林电子科技大学学报2024，Vol.44Issue(6)：642-649,8.DOI:10.16725/j.1673-808X.2022320
特征融合卷积神经网络的超表面建模方法
Metasurface modeling method based on feature fusion convolutional neural network
摘要
Abstract
To address the issue of low prediction accuracy resulting from input and output dimension mismatch in deep learning-based metasurface design,a metasurface modeling method based on data fusion convolutional neural networks is proposed.Firstly,the software co-simulation is utilized to model and simulate the structure of the superstrate,obtaining a dataset consisting of one-di-mensional structural parameters,two-dimensional images of the superstrate,and corresponding electromagnetic responses.Then,a CNN is constructed to extract the features of the two-dimensional images of the superstrate,which are combined with the one-di-mensional structural parameters to form a new feature matrix.This feature matrix is then trained with a fully connected layer to pre-dict the S21 parameters of the superstrate with the forward network and to predict the structural parameters of the superstrate with the inverse network.The trained model has good prediction capabilities,with mean squared errors of 3.148× 10-4 and 2.548× 10-3 for the forward and inverse network models,respectively.Further optimization of the network model is achieved through genetic algo-rithms,and transfer learning is utilized to accelerate the training process with new datasets.The results show that the algorithm-opti-mized network effectively avoids a large number of modeling operations and numerical computations,while maintaining high com-putational accuracy.关键词
卷积神经网络(CNN)/超表面/深度学习/特征融合/遗传算法/迁移学习Key words
convolutional neural network(CNN)/metasurface/deep learning/feature fusion/genetic algorithm/transfer learning分类
信息技术与安全科学引用本文复制引用
陈辉,蔡晗,王帅杰..特征融合卷积神经网络的超表面建模方法[J].桂林电子科技大学学报,2024,44(6):642-649,8.基金项目
广西自然科学基金(2021JJA170177) （2021JJA170177）
桂林电子科技大学研究生教育创新计划(2022YCXS027) （2022YCXS027）
