基于多Agent的供应商管理库存中需求预测的研究

华东交通大学硕士学位论文基于多Agent的供应商管理库存中需求预测的研究姓名：魏伟申请学位级别：硕士专业：计算机应用技术指导教师：杜江洪20061026I21VMIAgentAgentAgentMASAgentVMIVMIVMISwarmjavaAgentSwarmabstractIIResearchondemandforecastofvendormanageinventorybasedonMulti-AgentABSTRACTCompetitioninthe21stCenturyisthecompetitionofthesupplychain.Inventorymanagementisaveryimportantaspectofsupplychainmanagement.Thispaperwillchooseanadvancedinventorymanagementmodelvendormanageinventory(VMI)todiscuss.Theenterpriseisinthefaceofthevariousproductsandthefactorofcustomers’demandchangemoreandmore.Inthiskindofsituation,theenterprisescant’tfullygrasptheaccurateinformation.Inthesametimeuncertaintyoftheproductshort-termdemandwillincreasethedifficultyofproductforcast.Soweestablishademandmodelwhichcanforecastaccuratelyandgetaccuratefilingstock.Andthisisacurrentlyresearchtopic.Andthesupplychainisacomplexnetworkandtypicallong-distanceheterogeneousdistributedsystemswhichismadeupofrawmaterialsupplier,theproducer,theretailmerchantandthewarehouse.TheAgenttechnologyisanimportantbreakthroughintheresearchofAI.UsingAgent’sinitiative,interaction,reactivity,independencytoconstructthedistributionalsystemstructuralmodelisaneffectivemethod.Atpresent,theMASmodelingmothodbasedonAgentistheoneoftheimportantresearchmethodswhichisusedmoreineconomic,themanagementscience,thebiologicalscienceandsoon.ThispaperdiscussesthemothodtomodelfromthebottomtotopandthemeritandtheadvantangeofVMI.Weusethemulti-spotofgeneticalgorithm,analysisthemanagementsystemstructureofVMIandthemothodofforecasttoestablishaVMImodel.Weaddtheforecastmethodwhichisconstructedbygeneticalgorithmtothemodel.Thencomputingtheforecastandthefillstocktofindthebetterforecastvalue.InthismodeleveryAgentintercoordinates,sharestheinformation,realizestheinformationsharingfunctioninthesupplychainbetweenthepoint.Finallywetaketheactualmarketconditionsasthebackgroundtomakeashort-timesilulationusejavatosimulateonSwarmplatform.Andtakingtheaverageastheappraisaltarget,keepingthechangesituationtoconfirmshowingagain.Andcarringonthecomparisonwiththetraditionalforecastmethod.Whenusingourmethod,thedemandforecastismoreaccuratethantheothers,itsstockchangesituationisstable,namelytheerroneousundulationissmall.Sowhensettingthesafestock,wecanestablishthelowsafestock,reducetheentiresupplychain’sstockcost.Therefore,thismodelmightalsodoforthefuturecorrelationresearchforthereference.KeyWordsMulti-Agent,vendormanageinventory,Swarm,geneticalgorithm,demandforecast____________________________________________________________2,(SupplyChainManagementSCM)()[1](VendorManagedInventoryVMI)[2]3EDIVMIVMIVMIAgentAgentAgentAgentAgentVMI[3]AgentVMIVMISwarm4java1.2.1MAS:MASAgentAgentMASMASInternetAgentMASMASInternetF.Perriolat1996ARCHONB.Burmeister1997R.Weihmayer1998PR(KAMRO)KAMARA[4]AgentMASMASSwarm[5]1998AgentAOSDEAgentSACL,AgentADLAgentAgentNDUIA(NanDaUserInterfaceAgent)Agent:1;21AgentAgentSwaminathan[6]Kalakota[7]AgentWhat-ifSwaminathanAgentAgent5[6]Agent[8]PetriAgentAgentAgentAgentSwarm[9]AgentSadehAgent[10][11]Agent2AgentAgentAgentAgentMahmoudTchikouEricGouardereAgentAgent[12][13]AgentAgentAgentAgentAgentAgentAgentAgentAgent1.2.221[14]AgentMASAgent[15]Agent[16]Agent6[17]AgentAgentAgent1.AgentAgentAgentSwarmAgent2.VMIAgent3.SwarmjavaAgentVMIAgentSwarmjavaAgentVMIVMIVMISwarmjava7AgentSwarmAgentVMISwarmAgentAgentAgentAgentAgentVMIVMIVMISwarmSwarmMASMASSwarmjavaAgentAgent82.1.1AgentAgentDAIDAIDistributedProblemSolvingAgentMulti-AgentSystemParallelArtificialIntelligent[20]AgentAI[20]Agent119771995AgentSmartAgent1977HewwitActor[21]ActorsActorsAgent1995WooldridgeAgentdeliberative-AgentAgentAgentAgentActorsDVM[22]MICE[23][25][26]21995AgentFrontOfficeSystemCallCenterAgent[27]StanfordCISAgent[28]-CMUVistorHostingAgentAgentAgentAgentAgentAgentAgent2.1.2AgentAgent9AgentAgentAgentAgentAgent[29]AgentAgentBratmanBDIBelief-Desire-IntentionTheoryKripkeMooreCohenLevesqueKonoligeRAO[30]MooreAgent[31]AgentAgentMooreAgent[30]BratmanBDIBelief-Desire-IntentionTheoryBratman[31][32]1BratmanBDI[32]AgentAgent1987BratmanDAIBratmanAgentAgentAgentAgent2RaoGeorgeffBDI[33]A.RaoM.GeorgeffBDIBDIRaoGeorgeffCTL3AgentAgent10S5KD45KDAgentAgentAgentShohamAgent(belief),(capability)(decision)(commitment)FreemanAgentmkailsrg1994MinskyCommunicationofACMAgentAgentAgentAgentWooldridgeJenningsAgentAgentMITPatieMaesAgent[35]2.1.3AgentAgentAgentAgentAgentAgentAgentAgentAgent3AgentAgent3Agent2-1Agent11AgentAgentFig2-1TheConstructionofAgentsAction2.1.4AgentAgentAgentAgentAgentAgentAgent[36]1.(DeliberativeArchitecture),AgentAgentAgentAgentAgentAgent12Fig2-2TheChartofPrudenceAgent2.ReactiveArchitectureBrooks[37]AgreAgent/AgentAgentFig2-3TheChartofReflectAgent3.HybridArchitectureAgent-Agent132.1.5AgentAgentAgentAgentAgentAgentAgentAgent:AgentAgentAgentAgentAgentAgentAgentAgentAgentAgentAgentObjectManagementGroupAgentAgentPlatformSpecialInterestGroupAgentTechnologyGreenPaperITAgentl:AgentlAgentlAgentlAgentlAgentAgentlAgentlAgentlAgentlAgentlAgentlAgentAgentlAgentAgentAgentlAgentAgentAgentAgent14AgentlAgentAgentAgentAgentAgentAgentAgentAgent3:AgentAgent