基于神经网络的滤波天线单元优化技术研究

打开文本图片集

中图分类号：TN822+.4-34 文献标识码：A 文章编号：1004-373X（2025）15-0006-05

DOI：10.16652/j.issn.1004-373x.2025.15.002

Research on filtering antenna element optimization technology based on neural network

CHENJunda1，²，WUJie³，ZHAOJianing4，5，LUPei1，²，YANGRuoyang4，5，ZHANGHaichuan6

（1.CollegeofComputer Scienceand Engineering，Guilin Universityof Technology，Guilin 541006，China; 2.GuangxiKeyLaboratoryofmbededcholodIteligentystem，GuilinUnivesityofechologyGuilin4ina; 3.China Academy of Aerospace Electronics Technology，Beijing 1Ooo94，China; 4.YangtzeDeltaRegionInstitute（Huzhou），UniversityofElectronicScienceandTchnologyfChina，Huzhou330o9，Cina; 5.Huzhou KeyLaboratoryof Terahertz Integrated Circuitsand Systems，Huzhou 313oo9，China; 6.KeyLaboratoryofHighPowerMicrowaveTechnology，NorthwestInstituteofNulearTechnology，Xi’an7oO24，China）

Abstract：Thedesignoffilteringantennaelementsinvolvesmultidimensionalparameteroptimization，and traditional electromagneticsimulationmethodsarenotonlyslow，butalsostruggletoidentifytheglobaloptimuminacomplexparameter space，therebylimiting designaccuracyandperformanceenhancement.Therefore，amethodthatintegrates forward prediction with inverseoptimizationisproposed.Inthismethod，theneuralnetworksareemployedtopredictantennaelementperformance， andthe filteringantennaelementsareoptimizedgloballyinamulti-parameterspaceincombinationwiththegeneticalgorith. Simulation results indicate that the optimized S21 parameter exhibits an 82.65% increase in average value across the pass band from 11.5 GHz to 16.5 GHz， with auniform in-band response.In addition， S21 is improved by 87.5% after optimizing the average valueoftransmissonamplitudewiththecenterfrequencyof14GHz.Thetransmissonamplitudeisimprovedsignificantlyand thetransmisionphase shiftis more linear.Thesimulationresultsafteroptimizationshowthattheseimprovementsenhancethe overallperformanceof theantennasignificantlyespeciallyintransmissioneficiencyandfrequencyresponse.Tosumupthe proposed method provides valuable reference for the further development of the filtering antennas.

Keywords：neural network;genetic algorithm; forward prediction; inverse optimization; filtering antenna; S -parameter

0 引言

在现代通信系统中，天线作为信息传输的关键组件，其性能直接影响整个系统的效率和可靠性。（剩余6158字）