独立分量分析在雷达信号分选上的应用

:2006125海磊(解放军电子工程学院,合肥230037)传统的雷达信号分选主要是基于统计的思想,通过对脉冲序列的PRI分析,依据PRI的调制方式,将对应的雷达信号分选出来这种方法存在运算量大,速度慢,且准确率不高的缺点文章提出将含噪的快速独立分量分析算法应用于雷达信号的分选并分析了不同信噪比下分离的效果仿真试验证明了该算法可以取得很好的雷达信号分离效果盲源分离FastICA信噪比雷达信号分选TheApplicationofIndependentComponentAnalysisinRadarSignalClassificationHaiLei(ElectronicEngineeringInstituteofPLA,Hefei230037,China)Abstract:ThetraditionalmethodsofradarsignalclassificationrealizetheclassificationofradarsignalsbythemodulationmodeofPRI(PulseRepetitionInterval)basedontheprincipleofstatistics.However,theyhavesomedefectssuchasenormousdata,lowerrateandlowerprecision.Anewmethodofradarsignalclassificationbasedonfastindependentcomponentanalysisisdiscussedinthispaper.Thesimulationresultsshowthismethodcanobtainbettereffectofradarsignalclassification.Keywords:blindsourceseparation;FastICA;SNR;radarsignalclassification0,,,PRI,PRI,,,,,(independentcomponentanalysis,ICA),,200732007,No.3ELECTRONICWARFARE114SeriesNo.1141,,,:(1):;(2):(DOA)(AOA);(3):(TOA)(PW)(PRI);(4):,(),,,,,,PRI:(1),;(2),;(3),,2FastICAICA(IndependentComponentAnalysis)(blindsourceseparation,BSS),ICA[1][2]:mxi(t)(i=1,2,,m),nsj(t),:xi(t)=nj=1aijsj(t)+ni(t)(i=1,2,,m)(1),X(t)=AS(t)+N(t)(2),AN!M,mn,mXW,XY,Y(t)=WX(t)(3),Y(t),S(t)FastICAICA,,ICA,FastICA:,,3FastICA[3,4]mn,[5]ns1(t),s2(t),,sn(t)()m(mn)mx(t)=[x1(t)x2(t)xm(t)]T1420073()FastICA,31FastICAFastICA:X,,,;,(PCA)[6]X,X=BXE{XXT}=E{(BX)(BX)T}=I(4)B,I,BXRx=E{XXT},E,ERxET=∀(5),∀Rx,ERxBB=∀-12E(6),XXXX,W,W,(7),;W(n+1)=E{xG(WT(n)x)}-E{G#(WT(n)x)}W(n)(7)G#(∃)G(∃),G(∃)G(∃)G(u)=1alogcosh(au),a1!a!2,G#(u)=tanh(au),(8)W(n+1)=W(n+1)/%W(n+1)%(8),.,,.AWY=WXy1(t),y2(t),,yn(t)32FastICA,,::=AAT+(9)X,(9):R=AAT+(10)R,,R*=R-=AAT(11),,,A,:=diag(R-AAT)(12),A,X,X=BXE{XXT}=I+AAT(13)B(14)B=AT-1A-1AT-1(14)X,W,FastICA,W(),,,,,,114:15PRI,,,,:(),,,,,,,,,4,,(7),A=rand(3,3),123:;5dB,;-5dB,5,,,(a)(b)(c)1(a)1620073(b)(c)25dB(a)(b)(c)3-5dB1JuttenC,HeraultJ.Blindseparationofsources,PartI:Anadaptivealgorithmbasedonneuromimeticarchitecture[J].SignalProcessing,1991,24(1):1-102,..,2002,30(4):570-5763ParraL,SpenceC.ConvolutiveblindsourceseparationofnorstationarysourcesIEEETrans.onSpeechandAudioProcessing,2000,(5):320-3274HerbertBuchner,RobertAichner,WalterKellermannICA2003.Nara.Japan,20035..,2006,42(1):38-436,,..,2004,31(3):399-404(1979&),,,:2007∋(∋电子对抗(2007年全年定价60元,可通过编辑部订阅,也可以通过如下方式订阅:)邮局汇款天津市大寺泉集北里别墅17号联合征订服务部,邮编:300385电话:[phone]∗银行转帐户名:天津市河西区联合征订服务部开户银行:中国工商银行天津市尖山支行帐号:0302060509104619603114:17