基于机器学习与空间映射的滤波器优化方法  

Filter Optimization Method based on Machine Learning and Spatial Mapping

作　　者：胡聪 赵建平[1] 张仕顺 杨君[1] 徐娟[1] HU Cong;ZHAO Jian-ping;ZHANG Shi-shun;YANG Jun;XU Juan(College of Physics and Engineering,Qufu Normal University,Qufu Shandong 273165,China)

机构地区：[1]曲阜师范大学,山东曲阜273165

出　　处：《通信技术》2019年第3期745-749,共5页Communications Technology

基　　金：国家青年科学基金项目(No.61701278);山东省博士基金项目(No.ZR2017BF017);毫米波国家重点实验室开放课题(No.K201929)~~

摘　　要：滤波器是电子设备中的常见模块。随着通信技术的快速发展,系统对滤波器性能的要求越来越高,滤波器的高效优化成为重点。使用传统的空间映射方法对滤波器进行优化时,优化过程比较复杂,耗时较长,因此寻找一种简单快速的优化算法至关重要。通过使用机器学习中的支持向量机(SVM)算法寻找粗模型和细模型之间的对应关系,可以减少传统优化方法的复杂性,进一步节省计算时间。最后,通过微带带通滤波器优化设计的实例,证明了该方法的可行性。Filters are common modules in electronic devices.With the rapid development of communication technology,the system has higher and higher requirements for filter performance,and efficient optimization of the filtersbecomes the focus.When the traditional spatial mapping method is used to optimize the filter,the optimization process isfairly complicated and time-consuming.Therefore,it is very important to find a simple and fast optimization algorithm.Via using SVM(support vector machine)algorithm in machine learning and searching for the corresponding relation between rough model andfine model,the complexity of traditional optimization method may be reduced,and the calculation time be further saved.Finally,the feasibility of the proposed method is verified by an example of microstrip bandpass filter optimization design.

关 键 词：优化方法 空间映射 机器学习 支持向量机 

分 类 号：TN92[电子电信—通信与信息系统]