基于KNN和ANN算法的微带天线尺寸优化方法  

Microstrip antenna size optimization method based on KNN and ANN algorithms

作　　者：窦江玲 李聃 宋健 王青旺 沈韬 DOU Jiangling;LI Dan;SONG Jian;WANG Qingwang;SHEN Tao(Yunnan Key Laboratory of Computer Technologies Application,Kunming Yunan 650500,China;School of Information Engineering and Automation,Kunming Yunan 650500,China;Graduate School,Kunming University of Science and Technology,Kunming Yunan 650500,China)

机构地区：[1]昆明理工大学云南省计算机技术应用重点实验室,云南昆明650500 [2]昆明理工大学信息工程与自动化学院,云南昆明650500 [3]昆明理工大学研究生院,云南昆明650500

出　　处：《太赫兹科学与电子信息学报》2025年第1期61-65,共5页Journal of Terahertz Science and Electronic Information Technology

基　　金：国家自然科学基金资助项目(61971208);云南省万人计划青年拔尖人才资助项目(201873);云南省基础研究计划资助项目(202401AT070351,202301AV070003);云南省计算机技术应用重点实验室开放基金资助项目(2022202)。

摘　　要：为解决传统天线优化复杂度较高的问题,提出一种基于K-最近邻(KNN)算法和人工神经网络(ANN)算法的微带天线尺寸优化方法。该方法通过分析天线表面电流分布,将高敏感度参数设置为变量,低敏感度参数设置为常量,利用KNN和ANN对天线的尺寸参数进行优化,最终实现宽带性能的提升。为验证优化算法的有效性,加工了两款天线进行测试。结果表明,与传统天线设计方法相比,KNN和ANN算法使天线的阻抗带宽分别提高了20.8%和18.4%。其中,ANN算法在训练阶段耗时较长,但其阻抗匹配特性在多个频段上表现出显著改进。A microstrip antenna size optimization method based on K-Nearest Neighbors(KNN)and Artificial Neural Network(ANN)algorithms is proposed to solve the problem of high optimization complexity of traditional antennas.By analyzing the surface current distribution of the antenna,high-sensitivity parameters are set as variables,while low-sensitivity parameters are set as constants.The KNN algorithm and ANN algorithm are then utilized to optimize the size parameters of the antenna,ultimately enhancing broadband performance.To validate the effectiveness of the optimization algorithms,two antennas were fabricated and tested.The results indicate that compared to traditional antenna design methods,the KNN and ANN algorithms increase the impedance bandwidth by 20.8%and 18.4%,respectively.Although the ANN algorithm requires longer training time,it demonstrates significant improvements in impedance matching across multiple frequency bands.

关 键 词：K-最近邻(KNN) 人工神经网络(ANN) 机器学习 尺寸优化 微带天线 

分 类 号：TN82[电子电信—信息与通信工程]