简介:Weconsidertheclassicalpolicyiterationmethodofdynamicprogramming(DP),whereapproximationsandsimulationareusedtodealwiththecurseofdimensionality.Wesurveyanumberofissues:convergenceandrateofconvergenceofapproximatepolicyevaluationmethods,singularityandsusceptibilitytosimulationnoiseofpolicyevaluation,explorationissues,constrainedandenhancedpolicyiteration,policyoscillationandchattering,andoptimisticanddistributedpolicyiteration.Ourdiscussionofpolicyeva...
简介:NewalgorithmsbasedonartificialneuralnetworkmodelsarepresentedforcubicNURBScurveandsurfaceinterpolation.Whenallthknotspansareidentical,theNURBScurveinterpolationproceduredegeneratesintothatofuniformrationalB-splinecurves.Ifalltheweightsofdatapointsareidentical,thentheNURBScurveinterpolationproceduredegeneratesintotheintegralB-splinecurveinterpolation.
简介:Thispapershowsthemethodofestimatingspatiotemporaldistributionofpedestriansbyusingwatchcameras.Weestimatethedistributionwithouttrackingtechnology,withpedestrian’sprivacyprotectedandinUmedaundergroundmall.Latelyspatiotemporaldistributionofpedestrianshasbeingincreasinglyimportantinthefieldofurbanplanning,disasterpreventionplanning,marketingandsoon.Althoughmanyresearchershavetriedtocapturetheinformationoflocationasdealingwithsomesensors,someproblemsstillremain,suchastheinvestmentofsensors,therestrictionofthenumberofpeoplewhohasthedevicetheyareabletocapture.Fromsuchbackground,wedevelopanoriginallabellingalgorithmandestimatethespatiotemporaldistributionofpedestriansandtheinformationofthepassingtimeandthedirectionofpedestriansfromsequentialimagesofawatchcamera.
简介:ThispaperproposesandevaluatestwoimprovedPetrinet(PN)-basedhybridsearchstrategiesandtheirapplicationstoflexiblemanufacturingsystem(FMS)scheduling.Thealgorithmsproposedinsomepreviouspapers,whichcombinePNsimulationcapabilitieswithA*heuristicsearchwithinthePNreachabilitygraph,maynotfindanoptimumsolutionevenwithanadmissibleheuristicfunction.Toremedythedefectsanimprovedheuristicsearchstrategyisproposed,whichadoptsadifferentmethodforselectingthepromisingmarkingsandreservestheadmissibilityofthealgorithm.Tospeedupthesearchprocess,anotheralgorithmisalsoproposedwhichinvokesfasterterminationconditionsandstillguaranteesthatthesolutionfoundisoptimum.TheschedulingresultsarecomparedthroughasimpleFMSbetweenouralgorithmsandthepreviousmethods.Theyarealsoappliedandevaluatedinasetofrandomly-generatedFMSswithsuchcharacteristicsasmultipleresourcesandalternativeroutes.
简介:Thispaperdealswiththevectorcontrol,includingboththedirectvectorcontrol(DVC)andtheindirectvectorcontrol(IdVC),ofinductionmotors.ItiswellknownthattheestimationofrotorfluxplaysafundamentalroleintheDVCandtheestimationofrotorresistanceisvitalintheslipcompensationoftheIdVC.Intheseestimations,theprecisionissignificantlyaffectedbythemotorresistances.Therefore,onlineestimationofmotorresistancesisindispensableinpractice.Forafastestimationofmotorresistances,itisnecessarytoslowdowntheconvergencerateofthecurrentestimate.Ontheotherhand,forafastestimationoftherotorflux,itisnecessarytospeedupitsconvergencerate.Itisverydifficulttorealizesuchatrade-offinconvergenceratesinafullorderobserver.Inthispaper,weproposetodecouplethecurrentobserverfromthefluxobserversoastorealizeindependentconvergencerates.Then,theresistanceestimationalgorithmisappliedtobothDVCandIdVC.Inparticular,intheapplicationtoIdVCthefluxobserverneedsnotbeused,whichleadstoasimplerstructure.Meanwhile,independentconvergenceratesofcurrentobserverandfluxobserveryieldanimprovedperformance.Asuperiorperformanceinthetorqueandfluxresponsesinbothcasesisverifiedbynumeroussimulations.
简介:AMonteCarloAnalysisofnodesdeploymentforlarge-scaleandnon-homogeneouswirelesssensornetworks,hasbeendone.Throughsimulationsofrandomdeploymentsofnodesoverasquareareausingdifferentdensities,assumingthatournetworkiscomposedbyAnchornodes(specialsensorswithknownposition)andsimpleSensornodes,thelatteraresupposedtoestimatetheirownpositionafterbeingplacedwithinthecoverageareawiththeminimumAnchornodesneededto'feed'themwiththenecessaryinformation.Thegoalisthentoassistdecision-makersinselectingamongdifferentalternativestodeploythenetworks,accordingtoresourcesfeaturesandavailability,hencethismethodprovidesanestimatevalueofhowmanyAnchornodesshouldbedeployedinagivenareatotriggerthelocationalgorithminthegreatestpossiblenumberofSensornodesinthenetwork.