多传感器数据融合中多目标跟踪关键技术.研究

《多传感器数据融合中多目标跟踪关键技术.研究》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、多传感器数据融合中多目标跟踪关键技术研究arebothappliedtorebuildingnodecompatibilityfunctionsandedgecompatibilityfunctions．Theproposedalgorithmimprovesdataassociationperformanceinthesensornetworks；thet11irdmethod：themulti—characterinformationandthekinematicinformationarcfuseddirectly,bothusedto

2、FuzzyC-meansclusterbasedOUdataassociationalgorithm．Indensemeasurements，whenthekinematicinformationisonlyusedtocluster,theassociationresultisalwaysnotcorrect．Sothatintheproposedalgorithm，thecalculationofkeyparameter-distanceandsubjectionfunctionsisbasedonbothkenematicinforma

3、tionandmultiplefeaturesofthetarget．Furthermore．theeffectoffeaturescanbeadjusted,thusdataassociationisbeRerrealizedbasedonthemultiplefeatures．necomprehensiveinformationgreatlyraisestheperfornmceofdataassociation．Finally,outlierandout—of-sequencemeasurement，twopossiblekindsof

4、uncertainsituationindatafusionsystem,areresearched．Foroutlier,whenitarrivesinfusioncenterdirectly,itwouldinducelargedeviationtodatafusion．Soamethodofreal-timeeliminatingoutliermulti—sensordatafusionisproposed．Inthismethod,firstlythetrackofmulti—sensorisclusteredbyFCM，andthe

5、ntheoutliersaredetectedandeliminated，accordingtothecompactdegreeandthemembership．Finallyforecastedestimationtakestheplaceofoutliers，andissenttofusion．Thismethodsolvesoutlierproblemrapidlyandeffectively,aswellasdataassociationproblem；Forout-of-sequencemeasurement(OOSM)，tradi

6、tionalfilterscanhardlydealwithit．AccordingtoextendedKalmanfilter(EKF)based00SMfilterhaslowprecisionforthenonlinearsystem．anoptimalandasub—optimalUnscentedKalmanfilterforout-of-sequencemeasurementsarcproposed．11bprocedureofout-of-sequencemeasurementupdating,thatconvertingmul

7、tiplestepslagproblemtoonesteplagproblem,isdeducedagainbasedonI，KF．Andthemethodneedslittlememoryburden．Simulationresultsshowthatthepresentedalgorithmsaremoreeffectiveinutilizingout—of-sequencemeasurementsandtargettrackingperformancethanthoseinEKFfilterfornonlinearsystem．Atla

8、st，thetreatmentofmultiple00SMs，whichhavethesamelags，isresearched．AFCMbaseddataasso