基于多特征和层级网络的LPI雷达信号识别

打开文本图片集

中图分类号：TN974-34 文献标识码：A 文章编号：1004-373X（2025）17-0055-06

Abstract：AnLPIradar signal modulation patern recognition method basedonmulti-featureand hierarchical networksis proposedtometthechallengesbroughtbyLPIradarsignalstosignaldetectionandrecognition.Thecharacteristicsofradar signalswithdiffrentmodulationpaternsaretakenintoaccountcomprehensively.Firstlythesignalambiguityfunctinwith strongrepresentationabilityisusedasthemain feature todistinguish modulationpaterns.OnthebasisofGausianfiltering noisereduction，theatentionmechanismisconstructedtoidentifytheclasificationofnetworksignalsinadvance.Secondly sincethesignalambiguityfunctionbasedonautocorelationfunctionisdificulttodistinguishsomesignalmodulationpaterns efectively，thshort-timeFouriertransforimages（Ss）areselectedastheriginalfeaturesofthisindofsigals，anda secondary network including stackeddepth encoders isconstructedfor noisereductionreconstruction，featureenhancemntand clasificationrecogitionandtheodulationaterecogitionofdarsigalsisompleted.Finalltheproposedthodwas verifiedby simulation experiments.Thesimulationresults show that when the SNR （signal-to-noise ratio）is above -7 dB，the average recognition rate of the proposed methodis higher than 93.25% for8 typical LPI signals including polyphase coding and polytime coding.The validity of the proposed method has been verified.

KeyWords：LPIadarsignal;modulationpaternrecognition;ambiguityfunction;STF;mult-feature;hierarchicaletwork

0 引言

LPI雷达信号因具备低截获和低识别概率的优势，成为了当前雷达信号的主要发展方向，也给作为雷达对抗中重要环节的雷达信号识别带来了严峻挑战。（剩余6870字）