简介:BPandRBFneuralnetworktopredictforeststockvolumewerestudied,butthestudyinevaluatingbothnetworks’applicationeffectswasnotconducted.Inordertofindahigherforecastprecision,morestrongapplicativemethod,thecomprehensiveanalysisandevaluationonthetwomethodswerecarriedoutinthepracticalapplication.Bythecorrelationanalysis,crowndensity,shady-slopeandsunny-slope,TM1,TM2,TM3,TM5,TM7,NDVI,TM,(4-3),TM4/3wereselectedasinputvariables,andtheforestvolumeofMiyunCountyasoutputvariables,RBFandBPneuralnetworkmodelsforforecastingtheforestvolumewereestablished.Andtheneuralnetworktrainingsteplength,trainingtime,predictionaccuracyandtheapplicabilitymodelofthetwomethodswerecomprehensivelyanalyzed.TheresultsshowthattheRBFneuralnetworkmodelissuperiortotheBPneuralnetworkmodel.
简介:
简介:Afuzzyneuralnetworkcontrollerforunderwatervehicleshasmanyparametersdifficulttotunemanually.Toreducethenumerousworkandsubjectiveuncertaintiesinmanualadjustments,ahybridparticleswarmoptimization(HPSO)algorithmbasedonimmunetheoryandnonlineardecreasinginertiaweight(NDIW)strategyisproposed.OwingtotherestraintfactorandNDIWstrategy,anHPSOalgorithmcaneffectivelypreventprematureconvergenceandkeepbalancebetweenglobalandlocalsearchingabilities.Meanwhile,thealgorithmmaintainstheabilityofhandlingmultimodalandmultidimensionalproblems.TheHPSOalgorithmhasthefastestconvergencevelocityandfindsthebestsolutionscomparedtoGA,IGA,andbasicPSOalgorithminsimulationexperiments.ExperimentalresultsontheAUVsimulationplatformshowthatHPSO-basedcontrollersperformwellandhavestrongabilitiesagainstcurrentdisturbance.ItcanthusbeconcludedthattheproposedalgorithmisfeasibleforapplicationtoAUVs.
简介:Amethodologyisproposedtohandleproblemthatunderequiprobleaddressofpackettrafficattheinputport,GeneralizedShuffle-ExchangeNetwork(GSEN)routestrafficunevenlybecauseoftheunbalancedroutingtags.Theideaistouseroutingtagaccordingtoprobability,whichcanbeevaluatedbyusingMoore-Penroseinverseinmatrixanalysis.Aninstanceisusedtoillustratetheidea,andthesimulationisdonetoshowtheimprovementinperformanceissues.
简介:ThecharacteristicofEDFAinWDMsystemisanalyzed.ThesimulationsystembasedonthearchitectureoftheChinaInformationNetwork(CAINONET)isintroduced,andtheinfluenceofnoisefigure,gainunflatnessandoutputpowerofEDFAontheBERperformanceandscaleofnetworkisinvestigated.Moreover,theinfluenceofnoisefigureisstudiedwithdifferentcrosstalk.Meanwhile,itisindicatedthatitisimportanttoapplythe"nodeisolationprinciple"intheopticalnetwork.Theresearchresultscanprovidethevaluablereferenceforthepracticalconstructionofallopticalnetwork.
简介:FORUMON“WOMENANDSOCIALFORESTRY”HELDBYTHEFORESTRYANDSOCIETYNETWORK¥ByLiWeichangAforumon"WomenandSocialForestry"washeldattheIns...
简介:Anopticalsignal-to-noiseratio(OSNR)awarelightpathprovisioningmechanism(OSNR-LPM)isproposedfordistributedopticalnetworks.ThisOSNR-LPMtakestheOSNRvalueofthelightwavealongthelightpathintoconsiderationwhenestablishingthelightpathfortheconnectionrequestusingresourcereservationprotocol-trafficengineering(RSVP-TE).Moreover,theOSNR-LPMmakesfulladvantagesoftheOSNRmonitoringfunctionineachnodeandassignsthelightwavebyjudgingtheOSNRvaluecarriedbythesignalingmessageinordertoguaranteethereliableestablishmentofthelightpathinthedataplane.ThesimulationresultsshowthattheOSNR-LPMoutperformsotherlightpathassignmentmechanismsintermsofrealblockingprobabilityinthedataplane.
简介:
简介:InthispapertheΔ-Ytransformationofalinearelectriccircuitisintroducedtosimplifythecomputationofnetradiantheatexchangeonasurfaceofanenclosurewiththreesurfaces.Theillustrationofthetransformationisgivenbysolvingsixproblemstakenfromthetextbooks.Somediscussionconcerningthepossibilityofextendingthetransformationmethodtoenclosureswithmorethanthreesurfaces,andtheconditionforapplyingthenetworkmethodarealsopresented.
简介:Thispaperdescribesamodifiedspeed-sensorlesscontrolforinductionmotor(IM)basedonspacevectorpulsewidthmodulationandneuralnetwork.AnElmanANNmethodtoidentifytheIMspeedisproposed,withIMparametersemployedasassociatedelements.TheBPalgorithmisusedtoprovideanadaptiveestimationofthemotorspeed.Theeffectivenessoftheproposedmethodisverifiedbysimulationresults.TheimplementationonTMS320F240fixedDSPisprovided.
简介:Anovelcomplex,[Cu2(phen)(sal)(Hsal)2]n(1),wassynthesizedandstructurallycharacterized.Thebasicdimericunitsareholdbysalligandsandextendedinto1-Dnetwork.Thecarboxylategrougsofsalicylatescoordinatetothecentralioninthreedifferentcoordinationmodes:chelating,bridgingandbridging-chelating.Inthecaseofbridging-chelatingofthecarboxylategroupofthesalicylate,allthreeoxygenatomsofsalicylatearebidentatelycoordinatedtocopperion,namely,μ4-η^3bindingmode.
简介:InthispaperweconsideraMarkovchainmodelinanATMnetwork,whichhasbeenstudiedbyDagandStavrakakis.OnthebasisoftheiterativeformulasobtainedbyDagandStavrakakis,weobtaintheexplicitanalyticalexpressionofthetransitionprobabilitymatrix.ItisverysimpletocalculatethetransitionprobabilitiesoftheMarkovchainbytheseexpressions.Inaddition,weobtainsomeresultsaboutthestructureofthetransitionprobabilitymatrix,whicharehelpfulinnumericalcalculationandtheoreticalanalysis.
简介:Anewsystemisdevelopedtorecognizepromotersequencesfromnon-promotersequencesbasedonpositionweightmatrixandbackpropagationneuralnetworkinthispaper.Thesystemperformssignificantlybetteronthetrainingsetandthetestset,themeanrecognitionrateisashighas99%onthetrainingsetand97%onthetestingset.Experimentalresultsdemonstratetheeffectivenessofthesystemtorecognizethepromotersequencesthathavebeentrainedandthepromotersequencesthathavenotbeenseenpreviously.
简介:OnJanuary10,ademonstrationprojectofSGCCinZhejiangProvinceforelectricvehicles-asmartcharging&exchangingservicenetworksuccessfullypassedthescientific&technologicalachievementinspectionorganizedbyChinaElectricityCouncilinHangzhou,whichturnsouttobethefirsttrans-cityinterconnectedsmartcharging&exchangingserviceoftheworldandsignifiesitsleadinglevelinternationally.