简介:Thispaperpresentsanoveltimedelayestimation(TDE)methodusingtheconceptofentropy.Therelativedelayisestimatedbyminimizingtheestimatedjointentropyofmultiplesensoroutputsignals.Whenestimatingtheentropy,theinformationaboutthepriordistributionofthesourcesignalisnotrequired.Instead,theParzenwindowestimatorisemployedtoestimatethedensityfunctionofthesourcesignalfrommultiplesensoroutputsignals.Meanwhile,basedontheParzenwindowestimator,theRenyi’squadraticentropy(RQE)isincorporatedtoeffectivelyandefficientlyestimatethehigh-dimensionaljointentropyofthemultichanneloutputs.Furthermore,amodifiedformofthejointentropyforembeddinginformationaboutreverberation(multipathreflections)forspeechsignalsisintroducedtoenhancetheestimator’srobustnessagainstreverberation.
简介:Byutilizingtotalmagneticfluxφoftheprimaryandsecondarywindingsoftheflybacktransformerasastatevariable,thediscrete-timemodelofcurrent-modecontrolledflybackconverterisestablished,uponwhichthebifurcationbehaviorsoftheconverterareanalyzedandtwoboundaryclassificationequationsoftheorbitstateshiftingareobtained.Theoperation-stateregionsofthecurrent-modecontrolledflybackconverterarewellclassifiedbytwoboundaryclassificationequations.Thetheoreticalanalysisresultsareverifiedbypowerelectronicssimulator(PSIM).Theestimationofoperation-stateregionsfortheflybackconverterisusefulforthedesignofcircuitparameters,stabilitycontrolofchaos,andchaos-basedapplications.
简介:Thispaperdiscussestheproblemofdirectionofdeparture(DOD)anddirectionofarrival(DOA)estimationforabistaticmultipleinputmultipleoutput(MIMO)radar,andproposesanimprovedreduced-dimensionCaponalgorithmtherein.Comparedwiththereduced-dimensionCaponalgorithmwhichrequirespairmatchingbetweenthetwo-dimensionalangleestimation,theproposedalgorithmcanobtainautomaticallypairedDODandDOAestimationwithoutdebasingtheperformanceofangleestimationinbistaticMIMOradar.Furthermore,theproposedalgorithmhasalowercomplexitythanthereduced-dimensionCaponalgorithm,anditissuitablefornon-uniformlineararrays.ThecomplexityoftheproposedalgorithmisanalyzedandtheCramer-Raobound(CRB)isalsoderived.Simulationresultsverifytheusefulnessoftheproposedalgorithm.
简介:Inthispaper,weproposeanewmethodthatcombinescollageerrorinfractaldomainandHumomentinvariantsforimageretrievalwithastatisticalmethod-variablebandwidthKernelDensityEstimation(KDE).TheproposedmethodiscalledCHK(KDEofCollageerrorandHumoment)anditistestedontheVistextexturedatabasewith640naturalimages.ExperimentalresultsshowthattheAverageRetrievalRate(ARR)canreachinto78.18%,whichdemonstratesthattheproposedmethodperformsbetterthantheonewithparametersrespectivelyaswellasthecommonlyusedhistogrammethodbothonretrievalrateandretrievaltime.