物流配送車輛路徑優(yōu)化研究

物流配送車輛路徑優(yōu)化研究

1sostributepolusssiphing在中國(guó)，50%的樣本從懸崖和維修費(fèi)用中投入，占45%。LogisticsisattheinitiatestagecurrentlyinChina.Ittakesup40%ofthetotalcommoditycost.Directlossisover50billionRMBperyearandlossduetothelackoftransportationcapacityis10billionRMBperyear.InAmerica,theoccupancyratewas80%decadeagoandnowreachingupto90%.Theproblemisobviouslyserious.Tosolvetheproblem,GPSwillplayasignificantrole.VariousapplicationsofGPSfunctionshavebeenprovedtobeeffective,especiallylogisticsdistribution.HerearethemainfunctionsofGPSinfreightdistributionasfollows:1)Navigation.2)Trackingvehicle.3)Goodsdistributeroutineplanning:(1)Automotiveroutineplanning;(2)Artificialroutineplanning;(3)Others.Informationrequirement,urgentaid.2“三自”—AnalysisofcurrentresearchonvehicleroutineproblemVehicleRoutingProblem(VRP)isaproblemthatdesignsaleastcostroutineofvehicleteamservingaknowngroupofcustomers,whichbeginandendatthesamepointandmustmakesureeachcustomerhasbeenservedoncewithinthevehicle′scapacity.2.1ExistedresearchEversince1958DantziandRamserfirstlypromotedVRP,numerousresearchershasbeenattracttothisfield.Anumberofheuristicsandsubheuristicsalgorithmshaveexisted,suchasGi11ettandMi11er′ssweepingalgorithm,ClarkandWright′ssavingalgorithm,Lin′s22optand32optexchangealgorithm,BrameltandSmchilevi′sbasedonlocationchoosingproblemLBHalgorithm,andrecentyearexistedgeneticalgorithm,neuralnetwork,simulatedannealing,Abusearchandsoon.RelatedliteraturesincludeMalandraki(1992,1996),RussJ(1995),Joojung(2000),YangByung(2000),Sorrelm(2003),andAmbardella(2003)etal.AnumberofexpertsresearchonVRPinChinaatpresent,forinstance,“StudyontheLogisticsDistributionVRPBasedonImmuneAlgorithm”(2003),“StudyontheOptimizationofLogisticsDistributionVRPBasedonImmuneCloneAlgorithm”(2004),“AGA-basedAlgorithmClusteringAnalysisinDeliveryVehicleScheduling”(2004),“ResearchofVehicleRoutingProblemUnderConditionofFuzzyDemand”(2004)andsoon.2.2dynamicrnp,atis,sicam.udeExistedstudiesmostlyfocusonanalysisoflongtermrecordstofindstatisticalrulesofthestatisticVRP.AfewstudieshavebeencarriedoutinTimeDependentVRP(TDVRP)modelandalgorithm,anditisstillattheinitialstage.Shortcomingsofcurrentresearchinclude:1)Existalgorithmsarenotapplicablefornewproblem.2)ResearchresultofdynamicVRPshouldbeevaluatedbydataunderrealtransportationenvironment.3)Noenhancementhasbeenproposedyet.4)ResearchofdynamicrandomVRP,isfarfromenough.Inthispaper,wepromotedanalgorithmbasedonfuzzypossibilityandstudiedontheimpactofGPSinformationonfinaldecisionobjectives.3whichdematchLogisticsdistributionVPRcanbedescribedasfollows:severalcapacitydefinitevehiclestakegoodsfromlogisticscenterbywayofreasonablearrangedroutinewhichmakethetotalhauldistanceshortest,toseveraldemandpoints,positionanddemandofwhicharecertain,andmeetingthefollowingconditions:1)Totaldemandofallpointsoneachpathnomorethanvehiclecapacity.2)Distributiondistanceofeachpathshouldwithinthedistributionrangeofvehicles.3)Eachdemandpointservedbyonevehicle.3.1et顯著etw的鑒定SetdynamicVRPmodelhypothesisasfollows:(1)Inadistributionnetwork,eachoftheNcustomer′slocationanddemandisknown;(2)Vehiclecapacityisdefinite,butquantityisuncertain;(3)VRPmodelistomeetallthecustomer′sdemandandminimizecostofwholetrip.3.2通過(guò)轉(zhuǎn)色機(jī)制的抗混合物本設(shè)計(jì)型心理表1.Factorsaffecttheproblemincluding:warehouse,vehicle,time-window,customer,roadinformation,goodsinformation,transportregulationandsoon.Theobjectivesaremaximizingloadrateandminimizetotalcost.Set:(1)Cistotalcost;(2)Dkisdrivingrangeofthevehiclek;(3)Lisnumberofdemandpoints;(4)dijisdistancebetweeni,j(i,j=1,2,…,L);(5)nkisdemandpointofktoserve(nk=0Meansitisnotused);(6)Rkisroutinepathcollectionofvehiclek.ThenwegettheoptimizationmodelasfollowsminZ=∑k=1N[∑i=1nkdrk(i?1)rki+drknkrkosign(nk)](1)minΖ=∑k=1Ν[∑i=1nkdrk(i-1)rki+drknkrkosign(nk)](1)Thatisthebasicmodeltooptimizeshortestpath;anywayitdoesnotmeancostminimization.Herebywecombinemaximizedloadrateandminimizedistancetorealizelowestcost.Setαasloadrate,meeting0≤α≤1,(1)Whenα=0isfullload;(2)Whenα=1isempty;(3)Whenαisbetween0and1.SetCEpermileisthelostofnotfullload,αCEisthelostofunload,andCFisthefullloadcost.Themodifymodelisasfollows:minC=αCEDE+CFDF(2)minC=αCEDE+CFDF(2)Where:DEisnotfullloaddrivingdistance;DFisfullloaddrivingdistance;Zistotaldistance,Z=DE+DF.minC=CFZ+(αCE?CF)DE(3)minC=CFΖ+(αCE-CF)DE(3)(1)(αCE-CF)≥0,pathshortest,costleast;(2)(αCE-CF)<0,otherminimumvalue.Obviously,thedatafromGPSdevicesgreatlyaffectthemodel,withtherefreshmentofwhichdynamicpathoptimizationwillberealized.DistributionVPRiscomplexNPproblem,hereinweintroduceanewrestrictintothefunction.4通過(guò)網(wǎng)絡(luò)效應(yīng)檢測(cè)出面向營(yíng)造良好的風(fēng)險(xiǎn)意義,并基于sqp.rox.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.4.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3日起算價(jià),性別較高3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.3.Supposethereare5demandpoints,3vehicleswith5tonscarryingcapacity.0standsforthecenterand1-5arethedestinationpoints.DemandandDistancesareshowninthefollowingTable1andTable2.Onlypoint5couldnotdirectlyconnectwiththecenter.seta,b,cstandforthetheretrucksseparatelyLT=La+Lb+Lc=15(4)whereLa≤5,Lb≤5,Lc≤5.minC=CFZ+(αCE?CF)DE(5)minC=CFΖ+(αCE-CF)DE(5)SupposingGPSinformationcanaccuratelyreflecttherealstateoftruckload,andthenwecangetrealtimevalueofα.Table3showstheprogramsthetruckscanchoosefrombeforedeparture.Rijstandsforthejpathofiprogram.(Ignorethecaseofanytrucknotwork):Thenwecalculatedthecostofeachprogram,takingR12inprogram1forinstance.Fig.1showstheroutineofR12.a,b,cstandsforeachtruck:purplerepresenttrucka,blackb,orangec.Arrowsaretowardsdrivingdirection.Consideredoilconsuming,wearingandlabor,supposeitcosts1￥perkmwhenfullloadand0.9whenempty.Fig.2showsthatvacancyrunningcosttakesmorethanhalfoftotalvalue,fromwhichmostincreasesofcostcome.Eveniffullloadcostdecreasedthetotalcostincreaseduetoemptyrunning,suchasR22.Thatmeansreducetheemptyrunningcostcouldbringgreatprofitandmanymeasurementscoulddotorealizeit.Atthesametime,wecanalsoseeformthesketch,R12costleast884.54￥withtotalroutinelength9.41km,whileR24′slengthis10.82kmwhichcostsmost1250.9￥.Thatmeanstheroutineoptimizationcouldreducethecostalot.