Boundary restricted genetic algorithm for multi-layer equivalent dielectric parameters retrieval of artificial gradient structure

作     者：Guo, Shaojun Shen, Tongsheng Zou, Chunrong 

作者机构：Acad Mil Sci PLA Natl Inst Def Technol Innovat Beijing Peoples R China 169 Yongding Rd Beijing Peoples R China 

出 版 物：《INTERNATIONAL JOURNAL OF MODERN PHYSICS C》 (国际现代物理学杂志，C: 计算物理学和物理计算)

年 卷 期：2021年第32卷第1期

页      面：2150004-2150004页

核心收录：

学科分类：07[理学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 0702[理学-物理学] 

基　　金：National Nature Science Foundation of China [51802348  41906169] 

主　　题：Equivalent dielectric parameter retrieval S-curve inversion boundary restricted genetic algorithm optimization 

摘      要：With the equivalent dielectric parameters of an artificial gradient structure consisting of a kind of dielectric material as inputs of FDTD multi-layer equivalent simulation, there are big nonuniform differences between the S-curve of retrieval methods and the corresponding full structure. In order to decease these differences, here, a boundary restricted genetic algorithm is proposed. In our method, Smith S method is employed to find the rough values of dielectric parameters, and at the same time, the up and low limit cases are introduced to calculate the boundary parameter values for each layer of the artificial structure and form the searching areas for genetic algorithm to get high-precision inversion of S-curve. The FDTD S-curve of the retrieval parameters and full structure of cone gradient and moth eye were performed experimentally, the maximum deviation of inversion S21 curves corresponding to the cone and moth eye with full structure is limited within 0.0028 and 0.0024 in the X-band (8-12 GHz) range, respectively, which shows us the promising application of our method in dielectric parameter retrieval and may be helpful for electromagnetic field analysis.