基于互联网的连锁企业的物流管理系统本科学位论文.docx

基于互联网的连锁企业的物流管理系统本科学位论文.docx

《基于互联网的连锁企业的物流管理系统本科学位论文.docx》由会员分享，可在线阅读，更多相关《基于互联网的连锁企业的物流管理系统本科学位论文.docx（19页珍藏版）》请在冰豆网上搜索。

基于互联网的连锁企业的物流管理系统本科学位论文

Aninternet-basedlogisticsmanagementsystemforenterprisechains

N.Prindezis,C.T.Kiranoudis

SchoolofChemicalEngineering,NationalTechnicalUniversity,15780Athens,Greece

Received13September2003;receivedinrevisedform20December2003;accepted27January2004

Availableonline10December2004

Abstract

ThispaperpresentsanInternet-BasedLogisticsManagementSystemtocoordinateanddisseminatetasksandrelatedinformationforsolvingtheheterogeneousvehicleroutingproblemusingappropriatemetaheuristictechniques,foruseinenterprisechainnetworks.ItsarchitectureinvolvesaJAVAWebappletequippedwithinteractivecommunicationcapabilitiesbetweenperipheralsoftwaretools.ThesystemwasdevelopedindistributedsoftwarefashiontechnologyforallcomputerplatformsutilizingaWebbrowser,focusingonthedetailedroadnetworkofAthensandtheneedsoftheAthensCentralFoodMarketenterprises.2004ElsevierLtd.Allrightsreserved.

Keywords:

Decisionsupportsystem;e-Logistics;Transportation;Vehicleroutingproblem

1.Introduction

Enterprisechainsarethebusinessmodelofthepresentandfutureregardingmarketsthatinvolvesmallandmediumcompanysizes.Clearly,groupingactivitiestowardsafocusedtargetfacilitatesanunderstandablyimprovedmarketpenetrationguaranteedbyasuccessfultrademarkofaleadingcompanyinthefield.Severalcollaborationmodelsthatbasicallyincludefranchisingareintroducedasapartofthisintegratedprocess.Whensuchanetworkisintroducedinordertoexploitacommercialideaorbusinessinitiativeandsubsequentlyexpandedasmarketpenetrationgrows,severalmanagementissuesariseregardingtheoperationsoftheentirenetwork.SuchanetworkistheidealplacefororganizingandevaluatinginamorecentralizedwayseveralordinaryoperationsregardingsupplychainandlogisticsInfact,toolsdevelopedfororganizingmanagementprocessesandoperationalneedsofeachindividualcompany,canbedevelopedinamorecentralizedfashionandtheservicesprovidedbythetoolcanbeofferedtoeachnetworkmembertofacilitatetransactionsandtackleoperationssimilarly.Web-basedapplicationsareanidealstartingplacefordevelopingsuchapplications.Typicallysuchsystemsserveasacentraldepotfordistributingcommonservicesinthefieldoflogistics.Thecommercialapplicationisstoredinacentralserverandservicesareprovidedforeachmemberofthegroup.Aprototypeofsuchaserverisdescribedinapreviouswork（Prindezis,Kiranoudis,&Marinos-Kouris,2003）.ThispaperpresentsthecompletedinternetsystemthatisinstalledinthecentralwebserveroftheAthensCentralFoodMarketthatdealswiththeintegratedproblemofdistributionfor690companiesthatcompriseauniquelogisticsandretailchainofenterprises.Theneedsofeachcompanyareunderlinedandthealgorithmsdevelopedaredescribedwithintheunifiedinternetenvironment.Theproblemsolvedandservicesprovidedforeachcompanyistheoneinvolvingdistributionofgoodsthroughaheterogeneousfleetoftrucks.Newinsightsofthemetaheuristicsemployedareprovided.Acharacteristiccasestudyispresentedtoillustratetheeffectivenessoftheproposedapproachforareal-worldproblemofdistributionthroughthedetailedroadnetworkofAthens.

2.Distributionthroughheterogeneousvehiclefleets

Thefleetmanagementproblempresentedinthispaperrequirestheuseofaheterogeneousfleetofvehiclesthatdistributegoodsthroughanetworkofclients

（Tarantilis,Kiranoudis,&Vassiliadis,2003,2004）.Therefore,thesystemwasdesignedinordertoautomaticallygeneratevehicleroutes（whichvehiclesshouldde-

livertowhichcustomersandinwhichorder）,usingrational,quantitative,spatialandnon-spatialinformationandminimizingsimultaneouslythevehiclecostandthetotaldistancetravelledbythevehicles,subjecttothefollowingconstraints:

●eachvehiclehasapredeterminedloadcapacity,typicallydifferentfromallothervehiclescomprisingthefleet（heterogeneousnature）,

●thecapacityofavehiclecannotbeexceeded,

●asinglevehiclesupplieseachcustomersdemand,

●thenumberofvehiclesusedispredetermined.

TheproblemhasanobviouscommercialvalueandhasdrawntheattentionofORcommunity.Itsgreatsuccesscanbeattributedtothefactthatitisaveryinterestingproblembothfromthepracticalandtheoreticalpointsofview.Regardingthepracticalpointofview,thedistributionprobleminvolveddefinitelyplaysacentralroleintheefficiencyoftheoperationalplanninglevelofdistributionmanagement,producingeconomicalroutesthatcontributetothereductionofdistributioncosts,offeringsimultaneouslysignificantsavingsinallrelatedexpenses（capital,fuelcosts,driversalaries）.ItsImportanceinthepracticallevel,motivatedintensetheoreticalworkandthedevelopmentofefficientalgorithms.

FortheproblembyacademicresearchersandprofessionalsocietiesinOR/MS,resultinginanumberofpapersconcerningthedevelopmentofanumberofVehicleRoutingInformationSystems（VRIS）forsolvingtheproblem.TheproblemdiscussedisanNP-hardoptimizationproblem,thatistosaytheglobaloptimumoftheproblemcanonlyberevealedthroughanalgorithmofexponentialtimeorspacecomplexitywithrespecttoproblemsize.Problemsofthistypearedealtwithheuristicormetaheuristictechniques.Researchonthedevelopmentofheuristicalgorithms（Tarantilis&Kiranoudis,2001,2002a,2002b）forthefleetmanagementproblemhasmadeconsiderableprogresssincethefirstalgorithmsthatwereproposedintheearly60s.Amongthem,tabusearchisthechampion（Laporte,Gendreau,Potvin,&Semet,2000）.Themostpowerfultabusearchalgorithmsarenowcapableofsolvingmediumsizeandevenlargesizeinstanceswithinextremelysmallcomputationalenvironmentsregardingloadandtime.Onthealgorithmicside,timehasprobablycometoconcentrateonthedevelopmentoffaster,simpler（withfewparameters）andmorerobustalgorithms,evenifthiscausesasmalllossinqualitysolution.Theseattributesareessentialifanalgorithmistobeimplementedinacommercialpackage.

Thealgorithmbeyondthesystemdevelopedisoftabusearchnature.Asmentionedbefore,sincethealgorithmscannotrevealtheguaranteedglobaloptimum,thetimethatanalgorithmislefttoproposeasolutiontotheproblemisofutmostimportancetotheproblem.Certainly,thereisatrade-offbetweentimeexpectedfortheinductionofthesolutionanditsquality.Thispartwasimplementedinastraightforwardway.Ifthesystemisaskedbytheusertoproduceasolutionofveryhighqualityinstantly,thenanaggressivestrategyistobeimplemented.Iftheuserrelaxesthetimeofsolutiontobeobtained,thatistosayifthealgorithmislefttosearchthesolutionspacemoreeffciently,thenthereisroomformoreelaboratealgorithms.

Thealgorithmemployedhastwodistinctparts.Thefirstoneisageneralizedrouteconstructionalgorithmthatcreatesroutesofverygoodqualitytobeimprovedbythesubsequenttabuphase.Theconstructionalgorithmtakesintoaccountthepeculiaritiesoftheheterogeneousnatureoffleetandthedesireoftheusertousevehiclesofhisowndesire,ownedorhired,accordingtohisdailyneeds.

TheGeneralizedRouteConstructionAlgorithmemployed,isatwo-phasealgorithmwhereunroutedcustomersareinsertedintoalreadyconstructedpartialsolutions.Thesetofpartialsolutionsisinitiallyempty,andinthiscaseaseedrouteisinsertedthatcontainsonlythedepot.Rivalnodestobeinsertedarethenexamined.

Allroutesemployedinvolvesingleunroutedcustomers.Theinsertionprocedureutilizestwocriteriac1（i,u,j）andc2（i,u,j）toinsertanewcustomerubetweentwoadjacentcustomersiandjofacurrentpartialroute.Thefirstcriterionfindsthebestfeasibleinsertionpoint（i*,j*）thatminimizestheClarkandWrightsavingcalculationforinsertinganodewithinthisspecificinsertionpoint,

C1（i,u,j）=d（I,u）+d（u,j）-d（I,j）

（1）

Inthisformula,theexpressiond（k,l）standsfortheactualcostinvolvedincoveringthedistancebetweennodeskandl.TheClarkandWrightsavingcalculationintroducedinthisphaseservesasanappropriatestrongintensificationtechniqueforproducinginitialconstructionsofextremelygoodquality,acomponentofutmost

necessityintabuimprovementprocedure.

Thesecondphaseinvolvestheidentificationoftheactualbestnodetobeinsertedbetweentheadjacentnodepair（i*,j*）foundinthefirstphase（Solomon,1987）.Fromallrivalnodes,theoneselectedistheonethatmaximizestheexpression

C2（i*,u,j*）=[d（0,u）+d（u,0）]-C1（i*,u,j*）

（2）

where0denotesthedepotnode.Theexpressionselectedisthetravellingdistancedirectlyfrom/tothedepotto/fromthecustomerandtheadditionaldistanceexpressed

bythefirstcriterion.Inall,thefirstphaseoftheconstructionalgorithmseeksforthebestinsertionpointinallpossiblerouteseedsandwhenthisisdetected,theappropriatenodeisinserted.Ifnofeasiblenodeisfound,anewseedroute,containingasingledepot,isinserted.

Thealgorithmiteratesuntiltherearenounroutednodes.Itmustbestretchedthatthewayroutesarefilledupwithcustomersisguidedbythedesireoftheuserregarding

theutilizationofhisfleetvehicles.Thatistosay,vehiclesaresortedaccordingtothedistributionandutilizationneedsofthedispatcher.Vehiclestobeusedfirst（regardingtousercostaspectsandvehicleavailability）willbeloadedbeforeothersthatareoflowerimportancetotheuser.Typically,allusersinterviewedexpressedthedesirefortheutilizationofgr