Machine learning approach to transform scattering parameters to complex permittivities
Machine learning approach to transform scattering parameters to complex permittivities
复制标题
DOI：
10.1080/08327823.2021.1993046
复制
发表时间：
2021-10
影响因子：
1.5
通讯作者：
Robert Tempke;Liam A Thomas;Christina Wildfire;D. Shekhawat;T. Musho
中科院分区：
文献类型：
--
作者：
Robert Tempke;Liam A Thomas;Christina Wildfire;D. Shekhawat;T. Musho
摘要：本文研究了利用人工神经网络来确定由VNA散射参数测量得到的复杂介电材料特性。该研究利用有限元方法综合生成数据来训练神经网络。神经网络使用监督学习方法进行训练，并使用实验测量数据进行验证。频率范围为0.1 ~ 13.5 GHz，介电常数实部为1 ~ 100，虚部为0 ~ 0.2。与现有的逆方法相比，这种建模方法降低了不确定性。这种方法证明了一个通用的框架，可以用于转换实验或计算导出的散射参数到复杂的介电常数。
Abstract This study investigates the application of artificial neural networks to determine the complex dielectric material properties derived from experimental VNA scattering parameter measurements. The study utilizes a finite element approach to synthetically generate data to train the neural network. The neural network was trained using a supervised learning approach and validated using experimental measurement data. The frequency range of interest was between 0.1 and 13.5 GHz with the real part of the dielectric constants ranging from 1 − 100 and the imaginary part ranging from 0 − 0.2. This modelling approach decreases the uncertainty when compared to existing inverse approaches. This approach demonstrates a general framework that can be used for converting experimental or computational derived scattering parameters to complex permittivities.