22,IJ2(1.!KLM"> 22,IJ2(1.!KLM" />
欢迎来到天天文库
浏览记录
ID:53736278
大小:742.17 KB
页数:7页
时间:2020-04-21
《状态时间序列预测的贝叶斯最小二乘支持向量机方法.pdf》由会员上传分享,免费在线阅读,更多相关内容在应用文档-天天文库。
1、!44"!10#!"#$%&&'Vol.44№102010$10%JOURNALOFXI′ANJIAOTONGUNIVERSITYOct.2010!"#$%&'()*+,-./01234567-.1,3451,->22,IJ2(1.!KLMNOFPCQRSNO7,710025,TU;2.!KLMNOFPVWS;XYZ[],710025,TU)89:()*+,-./0123456789:,;<=>?@ABCDEFG6HIJKLMNOPQR9:ST.USTVWXYZH[01E]^_H`01E6Sab[HIJKLM
2、NOP9:cd,efghijklmnopq[`01E#rYstuv69:cdwxyz,${ABCDEFG)*9:cd
3、}6QR~1p.f@P,60123459:),UST69:n8cdSTg9.52%73.26%,79:、9:n86,f@,-./QR0123459:.:;<:HIJKLMNOP;ABCDEFG;,-./;P;0123459:=>?@A:TP277BCDEF:ABGHA:0253?987X(2010)10?0042?05犃犔犲犪狊狋犛狇狌犪狉犲狊犛狌狆狆狅狉狋犞犲犮狋狅狉犕犪犮犺犻狀犲犳狅狉犆狅狀
4、犱犻狋犻狅狀犜犻犿犲犛犲狉犻犲狊犘狉犲犱犻犮狋犻狅狀犅犪狊犲犱狅狀犅犪狔犲狊犻犪狀犈狏犻犱犲狀犮犲犉狉犪犿犲狑狅狉犽1,WANGHongli1,ZHANGJinsheng2,SUNYuan2ZHANGXian(1.DepartmentofAutomaticControlEngineering,TheSecondArtilleryEngineeringCollege,Xi′an710025,China;2.LaboratoryofAccurateGuidanceandSimulation,TheSecondArtil
5、leryEngineeringCollage,Xi′an710025,China)犃犫狊狋狉犪犮狋:AmethodusingBayesianevidenceframework(BEF)andleastsquaressupportvectormachine(LSSVM)isproposedtopredictelectronicsystemconditiontimeseriesaccurately.ALSSVMmodelforpredictionistrainedwithallthecurrentconditionti
6、meseriesdata.Then,theLSSVMmodelisiterativelyupdatedbyadoptingthelatestdataandpruningtheoldestdata.MatrixtransformisappliedtoreducethecomputationalcostofretrainingtheLSSVMmodel.Finally,theupdatedLSSVMmodelisdynamicallyoptimizedbyBEF.Numericalexperimentsonrada
7、rtransmitterconditiontimeseriespredictionarecarriedouttotesttheeffectivenessoftheproposedmethod.Theexperimentalresultsandcomparisonswiththeconventionaladaptivegreymodelshowthattheproposedmethodhasbetterperformanceinpredictionaccuracy,predictionstabilityandcomp
8、utationalefficiency,andthatthepredictionaccuracyandthecomputationalefficiencyforelectronicsystemconditiontimeseriespredictionareraisedby9.52%and73.26%respectively.犓犲狔狑狅狉犱狊:leastsquaressupportvectormachine;Bayesianevidenceframework;electronicsystem;radartransm
9、itter;conditiontimeseriesprediction^_`abcdefXghijk7lm,n~.`qr^jk7lqrmbopqrss^tuvOmwxyz{
10、ph}n,opqr,abip,&'(#:2010?03?23.)*+,:-.(1982-),/,012;345(678),/,9:,012;<.=>?@:ABCDEF=>GH?@(609
此文档下载收益归作者所有