picksolved
solved 时间:2021-01-17 阅读:()
JournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org50A0-1MODELFORFIREANDEMERGENCYSERVICEFACILITYLOCATIONSELECTION:ACASESTUDYINNIGERIAAROGUNDADEO.
T.
,AKINWALEA.
T.
;ADEKOYAA.
F.
ANDAWEOLUDAREG.
DepartmentOfComputerScience,UniversityOfAgriculture,P.
M.
B2240,Abeokuta,OgunState,Nigeria.
E-mail:arogundade@acm.
org,atakinwale@yahoo.
com,lanlenge@yahoo.
comandaweolu@yahoo.
comABSTRACT:Facilitylocationselectionproblemisavariantofsetcoveringproblem.
Setcoveringproblemisaclassicalproblemincomputerscienceandcomplexitytheory.
Inthispapertwodifferenttechniquesareappliedtofacilitylocationproblems.
First,amathematicalmodeloffacilitylocationisintroducedandsolvedbyusingoptimizationsolver,TORA.
Secondly,thebalasadditivealgorithmofbranchandboundtechniquesisusedtosolvethefacilitylocationproblem.
TestsweremadeusingreallifedatafromacityinNigeria.
Wethenobservedthatbothalgorithmsindicatethesamenumberoffirestationsindifferentlocations.
Alsotheresultsobtainedbyapplyingandimplementingbalasadditiveweremoreexplanatorybyspecifyingthenamesofthelocationswherethefacilitiesaretobelocatedandthenamesofthelocationstobeservedbyeachofthefacilities.
Keywords:Setcoveringproblem,firestation,emergencyservice,branchandbound,integerlinearprogramming.
1INTRODUCTIONSetcoveringproblemisaclassicalproblemincomputerscienceandcomplexitytheory,andisoneofthemostimportantdiscreteoptimizationproblembecauseitservesasamodelforrealworldproblems.
Realworldproblemsthatcanbemodeledassetcoveringproblemincludeairlinecrewscheduling,nurseschedulingproblems,resourceallocation,assemblylinebalancing,vehiclerouting,facilitylocationproblemwhichisthemainfocusofthiswork.
Etc.
Setcoveringproblemisaproblemofcoveringtherowofanm-row/n-columnzero-onematrixwithasubsetofcolumnsatminimalcost[1].
ThesetcoverproblemisaclassicNP-hardproblemstudiedextensivelyinliterature,andthebestapproximationfactorachievableforitinpolynomialtimeis(logn)[2,3,4].
Arichliteraturehasbeendevelopedandseveralmodelshavebeenformulatedandappliedtothefacilitylocationproblemsoverthelastfewyears.
Thecomplexityoftheseproblemsisduetothemultitudesofquantitativeandqualitativefactorsinfluencinglocationchoices.
However,investigatorshavefocusedonbothalgorithmsandformulationindiversesettingintheprivatesector(e.
g.
industrialplants,retailfacilities,telecommunicationmastetc)andthepublicsectors(e.
g.
schools,healthcenters,ambulances,clinicsetc).
Inthiswork,ourinterestisononeofthepublicsectorfacilitylocationproblem,thefireandemergencyservicelocationproblem.
Infact,fireandemergencyserviceiscrucialinsavinglivesandvaluablepropertiesandthereforemustprovidehighlevelofqualityservicestoensurepublicsafety.
Butprovidingthesefacilitieseffectivelyisacomplexissuethatespeciallydependsonsomefactorsandmostespeciallyonthebestgeographicallocationofthefirefightingandemergenciesservicefacilities.
TheaimofthispaperthereforeistouseaSetCoveringmodeltoselecttheminimumfirestationsthatcouldserveallareasinabigcityinsuchawaythateachwardwillhaveequalbenefitsintermsofservicesfromthefirestationsandalsothefacilitywillbestrategicallyplaced.
TheprocessinvolvesgatheringdataaboutallthewardsinthecityusingtheGPS(GlobalPointSystem)soastogettheirdistancesfromeachotherusingGISsoftware(GeographicalInformationService).
WethendevelopedadecisionsupportsytemthatdeterminetheminimumnumberoffirestationsneededtoserveallthewardssuchthattheJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org51distancebetweeneachwardandatleastonestationislessorequal10kilometersbysolvingthemathematicalmodelofthesetcoveringproblemusingtheBalasAdditivealgorithmaspecialcaseofbranchandboundthathandlesbinarylinearprogrammingproblem.
TheresultobtainedwascomparedtotheresultobtainedfromTORAsolver.
2LITERATUREREVIEWTheClassicalLocationSetCoveringProbleminvolvesfindingthesmallestnumberoffacilitiesandtheirlocationsothatdemandiscoveredbyatleastonefacility.
Itwasfirstintroducedby[12].
Theproblemrepresentseveraldifferentapplicationsettingincludingthelocationofemergencyserviceandtheapplicationsettingincludingthelocationofemergencyservicesandtheselectingofconservativesites.
Theproblemiscalledcoveringprobleminthatitrequiresthateachdemandbeservedor"covered"withinsomemaximumtimeanddistancestandards.
Ademandisdefinedascoveredifoneormorefacilitiesarelocatedwithinthemaximumdistanceortimestandardsofthatdemand.
ThesecondtypeofcoveringproblemiscalledtheMaximalCoveringLocationProblem[13].
Sincethedevelopmentofthesetwojuxtaposedproblemswereformed,therehavebeennumerousapplicationsandextensions.
SetCoveringProblemisoneofthemostprominentNP-completeproblem.
(Anexhaustivealgorithmmustsearchthroughall2msubsetsofStofindthosewhicharecoveringsubsetsandthenpicktheminimalfromamongthese[4]andcanformallybedefinedasfollow:Uistheuniversalset,SisacollectionofsubsetsofU,andc:S->Nisacostfunction.
ThegoalistofindacollectionS1,S2.
.
.
,SKofelementsofSsuchthatS1US2U.
.
.
USk=Uwithminimaltotalcost.
[16].
Significantresearchhasbeendirectedtowardstheproblemoflocatingandcoveringproblemsandseveralmethodshavebeenmadetoprovidesolutionsspecificallytothefacilitieslocationproblemandthesemethodsgenerallyinvolvetheuseofqueuingmodels[5],simulationandmathematicalprogramming,alsoacombinationofsimulationmodelandheuristicsearchroutines[6].
Alsoanextensivenumberofpapershavebeendedicatedtothesetcoveringproblem(SCP)andmanyexactalgorithms[7,10]whichcansolveinstanceswithuptofewhundredrowsandcolumns.
Acomparisonofsomeexactalgorithmscanbefoundin[9].
ApproximationalgorithmsplaysanimportantroleinsolvingSCP,giventhelimitationofexactmethodsandthelargelistofapplicationsusinglargesizeSCP[12].
Virtuallyeveryheuristicapproachforsolvinggeneralintegerproblemhasbeenappliedtosetcoveringproblems.
Thesetcoveringformulationnaturallylendsthemselvestogreedystart(i.
e.
anapproachthatateveryiterationmyopicallychoosesthenextbestsolutionwithoutregardsforitsimplicationonfuturemoves).
Interchangeapproacheshavealsobeenapplied;hereaswapofoneormorecolumnistakenwheneversuchaswapimprovestheobjectivefunctionvalue.
Newerheuristicapproachessuchasgeneticalgorithm,probabilisticsearch[8],simulatedannealing[11]andneuralnetworkhavealsobeentried.
Unfortunately,therehasnotbeenacomparativetestingacrosssuchmethodstodetermineunderwhatcircumstancesaspecificmethodmightperformbest.
Inaddition,onecanembedheuristicwithinanexactalgorithmsothatonecaniterativelytightentheupperboundandatthesametimeoneisattemptingtogetatightapproximationtothelowerboundforthisproblem.
Problemsarisinginpracticedonothoweverhaveperfectoridealmatrices.
Nevertheless,ithasbeenobservedincomputationalpracticethataslongastheproblemtobesolvedarerelativelyofmediumsize,linearprogrammingwithbranchandboundwillprovideintegersolutionquicklyandoptimally.
Howeverasthesubprogramsizeincreases,thenonintegralityofthelinearprogrammingsolutionincreasesdramaticallyanddoesthelengthandsizebranchingtree.
Itisforthislargeinstanceofproblemthatapproximationtechniques,reformulationandexactprocedureshavebeendevelopedthatexploittheunderlyingstructureoftheproblem.
IntegerLinearProgramming(ILPs)arelinearprogramsinwhichsomeorallofthevariablesarerestrictedtointeger(ordiscrete)values.
ILPhasimportantpracticalapplication.
Unfortunately,despitedecadesofextensiveresearch,computationalexperienceswithILPJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org52havebeenlessthansatisfactory.
TodatetheredoesnotexistanILPcomputercodethatcansolveintegerlinearproblemsconsistently[15].
2.
1ProblemStatementConsiderafirestationlocationandallocationproblemhavingthefollowingfeatures:Afirestationlocatedinawardhastoserveasetofwards.
Eachwardtobeservedmustbelocatedatfixeddistancetothelocationofthefirestation.
Theminimumnumberoffirestationsthatcanserveallthewardsmustbedetermined.
Themathematicalmodelofthisproblemisformulatedasfollow.
MinZ=∑Cjxjj={1,2,…n}Subjectedto:∑aijxj≥1i={1,2,…m}xj={0,1}whereCjisthecostofinstallation,xirepresentsacoveringi.
xjwhichcantakethevalue0or1dependingonifwardiisincoveringxj.
3MethodologyThispaperaimstoobtainanoptimalsolutiontofireandemergencyfacilitieslocationproblem.
WeusetheGPS(GlobalPointSystem)equipmenttogetthecoordinatesofallthewardsinthecityunderconsideration.
Fromthescreenoftheequipment,wegottheNorth-axisandtheEast-axisofeveryparticularplacewevisited(37wards).
Afterthecollectionofthecoordinates,weinstalledtheGIS(GeographicalInformationSystem)softwareforanalysis.
WethensupplythecoordinatesofeachwardintotheGISwhichthenlocatethepositionofthewardsonthemapofOgunstate(seefigure1)andthereafterobtainedthedistancereadingsforeachwardtotheother.
Figure1.
WardslocationonthemapJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org53Theresultobtainedfromthedistancereadingisa37by37matrixwhichwethentransformedintocoveringsaccordingtoaspecifieddistance(precisely10kmfromeachwards).
Forexample,thefirstcoverwhichis{1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}indicatethosewardthatcanbecoveredwithintherangeof10kmfromward1.
Thepartoftheresultofthisprocessisshowninfigure2.
Accordingtoourfirstdefinitionofsetcoveringproblem,theuniversalsetUis{1,2…37}andF={C1,C2,…………….
C37},nowouraimistofindtheminimumSasubsetofFsuchthatitsunionwillgiveusU,andatthisstagethewardsareallcoveredwithequaldistancesandtheCipickedarethewardswherethefirestationshouldbelocated.
Thesedatawerethenslottedintothebalasadditiveandtorasolvertosolvethefacilitylocationproblem.
Theresultsfromthetwoalgorithmswerethencomparedtodeterminetheoptimalcase.
Figure2.
Wardscovering3.
1ModelsUsedToSolveFireAndEmergencyFacilityLocationProblem3.
1.
1.
BalasAdditiveAlgorithmTheadditivealgorithmwasoneoftheapproachesknownasbranchandboundandisusedtosolvelinearprogramsinn0-1variablesbysystematicallyenumeratingasubsetof2npossiblebinarynvectors,whileusingthelogicalimplicationofthe0-1propertytoensurethatthewholesetisimplicitlyexamined.
Thetechniqueemployedinthisalgorithmisbasedonsystematicallyassigningthevalue0and1tocertainsubsetofvariablesandexploringtheimplicationsoftheseassignmentsbyasequenceoflogicaltests.
Thesimplicityoftheprocedureanditseffectivenesswhendataarenottoolargemakesitabetterchoiceforthisresearchwork.
BalasAdditivealgorithmrequiredthattheproblembeputinstandardform:1.
{1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}fromObantoko.
2.
{1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}fromIkija3.
{1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}fromAgoOko16{16,24}fromAlagbagba18{18,21,23}fromOsiele37{1,2,3,4,5,6,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}fromPansekeJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org54TheobjectivefunctionisaformofminimizationThemconstraintsareallinequalitiesoftheform(≤)AllthevariablesxjarebinaryvariablesAllobjectivefunctioncoefficientsarenonnegativeAlgorithm:BalasAdditiveAlgorithm1Standardizetheproblemtotheform:MinZ=∑j∈NCjxjs.
t∑j∈Naijxj≤biforalli∈M.
whereM={1,2,…m}andN={1,2,…n}xj={0,1},forallj∈N2SetaninitialupperboundtoZ=+∞,seti=0,andJ={}.
3SelectthenextpartialsolutionJ,solvetheLPiofJandattempttofathomusingoneofthethreeconditionslistedbelow.
a.
Allcompletionviolatesoneormoreconstraints.
i.
ecomputei.
A={j:j∈N-J,aij≥0foralli∈MsuchthatSi≤0}ii.
N1=N–J–AIfNI={}thenfathomthepartialsolutionJb.
Allcompletionareinferiortotheincumbentz'i.
ecomputei.
B={j:j∈N1,Z+Cj≥Z'}ii.
N2=N1–BIfN2={}thenfathompartialsolutionJc.
IfconstraintiisviolatedbythezerocompletionofthepartialsolutionsothatSi{}thenfathomthepartialsolutionJIfallthefathomtestfail,Gotostep64.
Ifbettersolutionisfound,thenupdateZ5IfallelementsofJisfathomedi.
eunderlined,thenZisoptimalGotostep7ElsesetJJ,{-j}andrepeatfromstep36Performbranchingby:i)Selectfreevariableforforwardstepii)SetJJ,{+j}Seti=i+1andrepeatstep37Terminate3.
1.
2.
TORAOneofthepowerfulfeaturesofTORAisitsgraphicaluserinterface(GUI)whichenablesuserstoexpresstheirproblemsinanaturalwaythatisverysimilartostandardmathematicalnotation.
ThisfeatureofGUIallowsuserstochoosethenextactionbeingmenudriven.
Thisoffersflexibilitytouserstoincreaseordecreasethedatasizeortoremoveaparticularvariablecompletely.
TORAoptimizationsolverhasthefollowingattributes:a.
Sets,whichcompriseofobjectsinprogrammingmodelb.
Objectivefunctionoftheproblemc.
ConstraintsofProblemd.
inputdataJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org554IMPLEMENTATIONANDRESULT4.
1FormatofinputdataInthispaper,theinputis38x380-1matrixwherecolumn2-38representseachcoveringandrow2-38representseachward.
Therefore,foreachcolumnandrow,theelementis1ifthewardiscoveredand0ifnotcovered.
E.
gthenameofthematrixisa,ifa[2][3]=1,itimpliesthatward3iscoveredbycovering2,otherwiseitisnotcoveredandthevaluewillbe0.
Thewholeinputfileformatforthisworkisshowninfigure4.
Theformatoftheoutputisinformofasolutionvectorcontainingonlyzerosandonesi.
e.
1ifacoveringisselectedand0ifnotselected.
Eachcoveringhasspecificnameofwardscoveringotherwardsthatarewithinthespecifieddistance.
(Thenameofeachwardandthenumberattachedtothemisshowninfigure3.
Figure3NamesofwardsandtheirnumberofidentificationTheinputmatrixshowninfigure4aandfigure4bwassavedastextfileandthebalasadditivealgorithmwasimplementedusingJavaprogramminglanguage.
Figure4aInputFileFormat1Obantoko,2Ikija,3Agooko,4ElegaHousing,5Iberekodo,6Agoika,7Ayetoro,8Okeago,9Totoro,10Itaosin,11Olorunda,12ImalaOrile,13IbaraOrile,14Ilewo/isaga,15Itaota,16Alagbagba,17Alabata,18Osiele,19Olodo,20Ilugun,21Agoodo,22Opeji,23Odeda,24Itesi,25Lafenwa,26Saje,27Itoko,28Ake,29Lantoro,30Ijemo,31Iporosodeke,32Irunbe,33Ijaye,34Okeitoku,35IjehunTitun,36Sabo,37Panseke.
JournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org56Figure4bInputFileFormatOncetheinputfilehasbeenselected,andthentheprogramcanberuntogeneratetheoutputrequired.
Theresultaftertheclickofthe"run"buttonisshowninfigure5below.
Figure5SetCoveringoutput.
JournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org57Thebalasadditiveresultaboveshowedthatcoveringisfoundandalsodisplayedthesolutionvector.
Itindicatedthatsixfirestationsareneededtoserveallthewardsandthelocationsofthosestationsareclearlystated.
ThesameresultwasobtainedfromTORAsoftwareintermsofoptimality,butthelocationsaredifferentandnotclearlystatedthoughitcanbetracedout.
Figure6ResultfromTORAsolverFigure6showstheresultoftheTORAsolver,theresultsissuchthatsixfirestationsarealsoneededtoserveallthewardseffectivelybutthelocationsofthefireservicestationarequitedifferentfromthatofBALASalgorithmthatwasimplemented.
Thelocationsindicatedbythesolverarelocations11,14,17,19,21,24whichcorrespondstothenamesofthefollowingwards(Olorunda,IlewoIsaga,alabata,Olodo,Ago-OdoandItesi)asshowninfigure3.
Torasolverdoesnotlistthenamesornumbersofthevillagesinitscovering.
5DISCUSSIONThenecessityofthedevelopmentoffacilitieslocationsoftwareforenhancingthedecisionmakingprocessandeventuallyproductivitycannotbeover-emphasized.
Theresultsobtainedinthisworkshowedthatsixfirestationsareneededtoserveeverywardsuchthatthemaximumdistancethatafirestationservicecangois10kilometers.
ItalsoshowedthatthelocationofthestationshouldbeObantoko,Olorunda,Ibaraorile,Olodo,OpejiandOdeda.
ThefirestationsatObantokowillrenderservicestotwentyfivewardswhichare:Obantoko,Ikija,Ago-oko,Elega,iberekodo,Ago-Ika,Ayetoro,Oke-ago,totoro,Ita-osin,Ita-ota,Ago-odo,lafenwa,Saje,Itoko,Ake,Lantoro,Ijemo,Iporo-sodeke,irunbe,Ijaye,Oke-itoku,Ijeun-titun,SaboandPanseke.
Olorundaservicestationwillservetwowardswhichare:OlorundaandImala-Orile.
Ibara-OrileservicestationwillserveIbara-orileandIlewo-Isagarespectively.
Olodoservicestationwillservethreewards.
Theyare:alagbagba,OlodoandIlugun.
Opejiwillservefourwardswhichare:alabata,OsieleandOpeji.
FinallyodedastationwillserveOdedaandItesi.
ThoughtheresultfromTORAsolveralsoindicatedthatsixfirestationsareneeded,itdidnotspecifytheactuallocationswherethestationsshouldbelocated.
ThisshortcominginTORAmakesouroutputandimplementationabetterone.
Theseresultsarepresentedinthetablebelowformoreclarity.
Itshowsthelocationswherethefacilitiesaretobeinstalledandalsothevillagestobecoveredbyeachofthefacility(coverings)onlyforbalasadditivealgorithm.
Theoutputfrombalasadditivealgorithmdoesnotshowfairdistribution.
InthecaseofthefacilityinlocationObantokowhichistoserve25locationswhileothersserveminimumoftwolocationsandmaximumofthree.
ThefacilityinObantokowillbeoverused.
JournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org58Table1:OUTPUTOFBALASADDITIVEALGORITHMLocationidentificationnumberLocationnameCoverings1Obantoko1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,3711Olorunda11,1213Ibaraorile13,1419Olodo16,19,2022Opeji17,18,2223Odeda23,24TABLE2:OUTPUTOFTORASOLVERLocationIdentificationNumberLocationNameCoverings11OlorundaNotindicated14Ilewo-IsagaNotindicated17AlabataNotindicated19OlodoNotindicated21Ago-odoNotindicated24ItesiNotindicated6CONCLUSIONSANDRECOMMENDATIONSThisresultthereforeshouldraiseawarenessandcontributetotheaimofourgovernmenttoadoptthistoolwhichwilldefinitelyimprovethefunctionalityoffirestationsinNigeriabysavingalotofcitizen'slivesandproperties.
Itshouldalsobenotedthattheuseofthissystemisnotlimitedonlytofirestationsallocationalone,butalsotootherpublicfacilitieslikeschools,policestationsoastoincreaseresponsetimeandthereforereducecrime.
Itcanalsobeusedbyprivateestablishments.
REFERENCES[1]J.
EBeasleyandP.
C.
Chu,"Ageneticalgorithmforthesetcoveringproblem",EuropeanJournalofOperationalResearch,vol.
94,1996,pp.
392-404.
[2]V.
Chvatal.
Agreedyheuristicforthesetcoveringproblem.
MathematicsofOperationResearch,4(3):233-235,1979.
[3]U.
Fiege.
Athresholdforlnnforapproximatingsetcover.
JournaloftheACM,45(4):634-652,July1998.
[4]D.
S.
Johnson.
Approximationalgorithmsforcombinatorialproblems.
J.
Compute.
SystemScience9:256-278,1974.
[5]LarsonRichardC.
,"AHypercubeQueuingModelforfacilitylocationandredistrictinginurbanemergencyservices",ComputersandOperationsResearch1(1974)67-95.
[6]SavasES,"Simulationandcost-effectivenessanalysisofNewYork'semergencyambulanceservice",ManagementScience14(1969)608-627.
[7]M.
L.
FisherandP.
Kedia.
OptimalSolutionofsetcoveringproblemsusingdualheuristics.
ManagementScience,36:674-688,1990.
[8]FeoA.
andG.
C.
MauricioandA.
Resende,(2002)"AProbabilisticHeuristicforaComputationallyDifficultSetCoveringProblem",OperationsResearchLetters,8,67-71.
[9]A.
Caparara,M.
Fischetti,andP.
Toth.
AlgorithmsforsetcoverinproblemstechnicalreportOR-98-3,DEIS,UniversityofBologna,Italy1998.
[10]N.
ChristofidesandJ.
PPaixao.
Algorithmsforlargescalesetcoveringproblems.
AnnalsofOperationResearch,43:261-277,1993.
[11]M.
JBrusco,L.
W.
Jacobs,andG.
MThompson.
Amorphingproceduretosupplementasimulatedannealingheuristicsforcostandcoverage-correlatedsetcoveringproblems.
AnnalsofOperationResearch,86:611-627,1999.
[12]ToregasC.
,SwainR.
,RevelleC.
,andBergmanL.
,(1988)"ThelocationofJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org59emergencyservicefacilities",Operationresearch19,pg1363-1373.
[13]ChurchRichardL.
,GerardRossa.
,(2003)"Themultilevellocationsetcoveringmodel"GeographicalAnalysisPublication,pg76-79.
[14]GareyM.
R.
andJohnsonD.
S.
,(2005)"Computerandinteractability:aguidetothetheoryofNP-completeness",pg34-38.
[15]Handya.
Taha.
,(2005)'OperationResearch:Anintroduction",Pearsoneducation(Singapore)Pte.
Ltd.
,pg391-397.
[16]Danielgulotta,(2006)"ApplicationoflinearProgrammingtosetcoveringandrelatedproblems"OnlinejournalofOperationResearch,pg1-2.
Allrightsreserved.
www.
jatit.
org50A0-1MODELFORFIREANDEMERGENCYSERVICEFACILITYLOCATIONSELECTION:ACASESTUDYINNIGERIAAROGUNDADEO.
T.
,AKINWALEA.
T.
;ADEKOYAA.
F.
ANDAWEOLUDAREG.
DepartmentOfComputerScience,UniversityOfAgriculture,P.
M.
B2240,Abeokuta,OgunState,Nigeria.
E-mail:arogundade@acm.
org,atakinwale@yahoo.
com,lanlenge@yahoo.
comandaweolu@yahoo.
comABSTRACT:Facilitylocationselectionproblemisavariantofsetcoveringproblem.
Setcoveringproblemisaclassicalproblemincomputerscienceandcomplexitytheory.
Inthispapertwodifferenttechniquesareappliedtofacilitylocationproblems.
First,amathematicalmodeloffacilitylocationisintroducedandsolvedbyusingoptimizationsolver,TORA.
Secondly,thebalasadditivealgorithmofbranchandboundtechniquesisusedtosolvethefacilitylocationproblem.
TestsweremadeusingreallifedatafromacityinNigeria.
Wethenobservedthatbothalgorithmsindicatethesamenumberoffirestationsindifferentlocations.
Alsotheresultsobtainedbyapplyingandimplementingbalasadditiveweremoreexplanatorybyspecifyingthenamesofthelocationswherethefacilitiesaretobelocatedandthenamesofthelocationstobeservedbyeachofthefacilities.
Keywords:Setcoveringproblem,firestation,emergencyservice,branchandbound,integerlinearprogramming.
1INTRODUCTIONSetcoveringproblemisaclassicalproblemincomputerscienceandcomplexitytheory,andisoneofthemostimportantdiscreteoptimizationproblembecauseitservesasamodelforrealworldproblems.
Realworldproblemsthatcanbemodeledassetcoveringproblemincludeairlinecrewscheduling,nurseschedulingproblems,resourceallocation,assemblylinebalancing,vehiclerouting,facilitylocationproblemwhichisthemainfocusofthiswork.
Etc.
Setcoveringproblemisaproblemofcoveringtherowofanm-row/n-columnzero-onematrixwithasubsetofcolumnsatminimalcost[1].
ThesetcoverproblemisaclassicNP-hardproblemstudiedextensivelyinliterature,andthebestapproximationfactorachievableforitinpolynomialtimeis(logn)[2,3,4].
Arichliteraturehasbeendevelopedandseveralmodelshavebeenformulatedandappliedtothefacilitylocationproblemsoverthelastfewyears.
Thecomplexityoftheseproblemsisduetothemultitudesofquantitativeandqualitativefactorsinfluencinglocationchoices.
However,investigatorshavefocusedonbothalgorithmsandformulationindiversesettingintheprivatesector(e.
g.
industrialplants,retailfacilities,telecommunicationmastetc)andthepublicsectors(e.
g.
schools,healthcenters,ambulances,clinicsetc).
Inthiswork,ourinterestisononeofthepublicsectorfacilitylocationproblem,thefireandemergencyservicelocationproblem.
Infact,fireandemergencyserviceiscrucialinsavinglivesandvaluablepropertiesandthereforemustprovidehighlevelofqualityservicestoensurepublicsafety.
Butprovidingthesefacilitieseffectivelyisacomplexissuethatespeciallydependsonsomefactorsandmostespeciallyonthebestgeographicallocationofthefirefightingandemergenciesservicefacilities.
TheaimofthispaperthereforeistouseaSetCoveringmodeltoselecttheminimumfirestationsthatcouldserveallareasinabigcityinsuchawaythateachwardwillhaveequalbenefitsintermsofservicesfromthefirestationsandalsothefacilitywillbestrategicallyplaced.
TheprocessinvolvesgatheringdataaboutallthewardsinthecityusingtheGPS(GlobalPointSystem)soastogettheirdistancesfromeachotherusingGISsoftware(GeographicalInformationService).
WethendevelopedadecisionsupportsytemthatdeterminetheminimumnumberoffirestationsneededtoserveallthewardssuchthattheJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org51distancebetweeneachwardandatleastonestationislessorequal10kilometersbysolvingthemathematicalmodelofthesetcoveringproblemusingtheBalasAdditivealgorithmaspecialcaseofbranchandboundthathandlesbinarylinearprogrammingproblem.
TheresultobtainedwascomparedtotheresultobtainedfromTORAsolver.
2LITERATUREREVIEWTheClassicalLocationSetCoveringProbleminvolvesfindingthesmallestnumberoffacilitiesandtheirlocationsothatdemandiscoveredbyatleastonefacility.
Itwasfirstintroducedby[12].
Theproblemrepresentseveraldifferentapplicationsettingincludingthelocationofemergencyserviceandtheapplicationsettingincludingthelocationofemergencyservicesandtheselectingofconservativesites.
Theproblemiscalledcoveringprobleminthatitrequiresthateachdemandbeservedor"covered"withinsomemaximumtimeanddistancestandards.
Ademandisdefinedascoveredifoneormorefacilitiesarelocatedwithinthemaximumdistanceortimestandardsofthatdemand.
ThesecondtypeofcoveringproblemiscalledtheMaximalCoveringLocationProblem[13].
Sincethedevelopmentofthesetwojuxtaposedproblemswereformed,therehavebeennumerousapplicationsandextensions.
SetCoveringProblemisoneofthemostprominentNP-completeproblem.
(Anexhaustivealgorithmmustsearchthroughall2msubsetsofStofindthosewhicharecoveringsubsetsandthenpicktheminimalfromamongthese[4]andcanformallybedefinedasfollow:Uistheuniversalset,SisacollectionofsubsetsofU,andc:S->Nisacostfunction.
ThegoalistofindacollectionS1,S2.
.
.
,SKofelementsofSsuchthatS1US2U.
.
.
USk=Uwithminimaltotalcost.
[16].
Significantresearchhasbeendirectedtowardstheproblemoflocatingandcoveringproblemsandseveralmethodshavebeenmadetoprovidesolutionsspecificallytothefacilitieslocationproblemandthesemethodsgenerallyinvolvetheuseofqueuingmodels[5],simulationandmathematicalprogramming,alsoacombinationofsimulationmodelandheuristicsearchroutines[6].
Alsoanextensivenumberofpapershavebeendedicatedtothesetcoveringproblem(SCP)andmanyexactalgorithms[7,10]whichcansolveinstanceswithuptofewhundredrowsandcolumns.
Acomparisonofsomeexactalgorithmscanbefoundin[9].
ApproximationalgorithmsplaysanimportantroleinsolvingSCP,giventhelimitationofexactmethodsandthelargelistofapplicationsusinglargesizeSCP[12].
Virtuallyeveryheuristicapproachforsolvinggeneralintegerproblemhasbeenappliedtosetcoveringproblems.
Thesetcoveringformulationnaturallylendsthemselvestogreedystart(i.
e.
anapproachthatateveryiterationmyopicallychoosesthenextbestsolutionwithoutregardsforitsimplicationonfuturemoves).
Interchangeapproacheshavealsobeenapplied;hereaswapofoneormorecolumnistakenwheneversuchaswapimprovestheobjectivefunctionvalue.
Newerheuristicapproachessuchasgeneticalgorithm,probabilisticsearch[8],simulatedannealing[11]andneuralnetworkhavealsobeentried.
Unfortunately,therehasnotbeenacomparativetestingacrosssuchmethodstodetermineunderwhatcircumstancesaspecificmethodmightperformbest.
Inaddition,onecanembedheuristicwithinanexactalgorithmsothatonecaniterativelytightentheupperboundandatthesametimeoneisattemptingtogetatightapproximationtothelowerboundforthisproblem.
Problemsarisinginpracticedonothoweverhaveperfectoridealmatrices.
Nevertheless,ithasbeenobservedincomputationalpracticethataslongastheproblemtobesolvedarerelativelyofmediumsize,linearprogrammingwithbranchandboundwillprovideintegersolutionquicklyandoptimally.
Howeverasthesubprogramsizeincreases,thenonintegralityofthelinearprogrammingsolutionincreasesdramaticallyanddoesthelengthandsizebranchingtree.
Itisforthislargeinstanceofproblemthatapproximationtechniques,reformulationandexactprocedureshavebeendevelopedthatexploittheunderlyingstructureoftheproblem.
IntegerLinearProgramming(ILPs)arelinearprogramsinwhichsomeorallofthevariablesarerestrictedtointeger(ordiscrete)values.
ILPhasimportantpracticalapplication.
Unfortunately,despitedecadesofextensiveresearch,computationalexperienceswithILPJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org52havebeenlessthansatisfactory.
TodatetheredoesnotexistanILPcomputercodethatcansolveintegerlinearproblemsconsistently[15].
2.
1ProblemStatementConsiderafirestationlocationandallocationproblemhavingthefollowingfeatures:Afirestationlocatedinawardhastoserveasetofwards.
Eachwardtobeservedmustbelocatedatfixeddistancetothelocationofthefirestation.
Theminimumnumberoffirestationsthatcanserveallthewardsmustbedetermined.
Themathematicalmodelofthisproblemisformulatedasfollow.
MinZ=∑Cjxjj={1,2,…n}Subjectedto:∑aijxj≥1i={1,2,…m}xj={0,1}whereCjisthecostofinstallation,xirepresentsacoveringi.
xjwhichcantakethevalue0or1dependingonifwardiisincoveringxj.
3MethodologyThispaperaimstoobtainanoptimalsolutiontofireandemergencyfacilitieslocationproblem.
WeusetheGPS(GlobalPointSystem)equipmenttogetthecoordinatesofallthewardsinthecityunderconsideration.
Fromthescreenoftheequipment,wegottheNorth-axisandtheEast-axisofeveryparticularplacewevisited(37wards).
Afterthecollectionofthecoordinates,weinstalledtheGIS(GeographicalInformationSystem)softwareforanalysis.
WethensupplythecoordinatesofeachwardintotheGISwhichthenlocatethepositionofthewardsonthemapofOgunstate(seefigure1)andthereafterobtainedthedistancereadingsforeachwardtotheother.
Figure1.
WardslocationonthemapJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org53Theresultobtainedfromthedistancereadingisa37by37matrixwhichwethentransformedintocoveringsaccordingtoaspecifieddistance(precisely10kmfromeachwards).
Forexample,thefirstcoverwhichis{1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}indicatethosewardthatcanbecoveredwithintherangeof10kmfromward1.
Thepartoftheresultofthisprocessisshowninfigure2.
Accordingtoourfirstdefinitionofsetcoveringproblem,theuniversalsetUis{1,2…37}andF={C1,C2,…………….
C37},nowouraimistofindtheminimumSasubsetofFsuchthatitsunionwillgiveusU,andatthisstagethewardsareallcoveredwithequaldistancesandtheCipickedarethewardswherethefirestationshouldbelocated.
Thesedatawerethenslottedintothebalasadditiveandtorasolvertosolvethefacilitylocationproblem.
Theresultsfromthetwoalgorithmswerethencomparedtodeterminetheoptimalcase.
Figure2.
Wardscovering3.
1ModelsUsedToSolveFireAndEmergencyFacilityLocationProblem3.
1.
1.
BalasAdditiveAlgorithmTheadditivealgorithmwasoneoftheapproachesknownasbranchandboundandisusedtosolvelinearprogramsinn0-1variablesbysystematicallyenumeratingasubsetof2npossiblebinarynvectors,whileusingthelogicalimplicationofthe0-1propertytoensurethatthewholesetisimplicitlyexamined.
Thetechniqueemployedinthisalgorithmisbasedonsystematicallyassigningthevalue0and1tocertainsubsetofvariablesandexploringtheimplicationsoftheseassignmentsbyasequenceoflogicaltests.
Thesimplicityoftheprocedureanditseffectivenesswhendataarenottoolargemakesitabetterchoiceforthisresearchwork.
BalasAdditivealgorithmrequiredthattheproblembeputinstandardform:1.
{1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}fromObantoko.
2.
{1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}fromIkija3.
{1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}fromAgoOko16{16,24}fromAlagbagba18{18,21,23}fromOsiele37{1,2,3,4,5,6,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,37}fromPansekeJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org54TheobjectivefunctionisaformofminimizationThemconstraintsareallinequalitiesoftheform(≤)AllthevariablesxjarebinaryvariablesAllobjectivefunctioncoefficientsarenonnegativeAlgorithm:BalasAdditiveAlgorithm1Standardizetheproblemtotheform:MinZ=∑j∈NCjxjs.
t∑j∈Naijxj≤biforalli∈M.
whereM={1,2,…m}andN={1,2,…n}xj={0,1},forallj∈N2SetaninitialupperboundtoZ=+∞,seti=0,andJ={}.
3SelectthenextpartialsolutionJ,solvetheLPiofJandattempttofathomusingoneofthethreeconditionslistedbelow.
a.
Allcompletionviolatesoneormoreconstraints.
i.
ecomputei.
A={j:j∈N-J,aij≥0foralli∈MsuchthatSi≤0}ii.
N1=N–J–AIfNI={}thenfathomthepartialsolutionJb.
Allcompletionareinferiortotheincumbentz'i.
ecomputei.
B={j:j∈N1,Z+Cj≥Z'}ii.
N2=N1–BIfN2={}thenfathompartialsolutionJc.
IfconstraintiisviolatedbythezerocompletionofthepartialsolutionsothatSi{}thenfathomthepartialsolutionJIfallthefathomtestfail,Gotostep64.
Ifbettersolutionisfound,thenupdateZ5IfallelementsofJisfathomedi.
eunderlined,thenZisoptimalGotostep7ElsesetJJ,{-j}andrepeatfromstep36Performbranchingby:i)Selectfreevariableforforwardstepii)SetJJ,{+j}Seti=i+1andrepeatstep37Terminate3.
1.
2.
TORAOneofthepowerfulfeaturesofTORAisitsgraphicaluserinterface(GUI)whichenablesuserstoexpresstheirproblemsinanaturalwaythatisverysimilartostandardmathematicalnotation.
ThisfeatureofGUIallowsuserstochoosethenextactionbeingmenudriven.
Thisoffersflexibilitytouserstoincreaseordecreasethedatasizeortoremoveaparticularvariablecompletely.
TORAoptimizationsolverhasthefollowingattributes:a.
Sets,whichcompriseofobjectsinprogrammingmodelb.
Objectivefunctionoftheproblemc.
ConstraintsofProblemd.
inputdataJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org554IMPLEMENTATIONANDRESULT4.
1FormatofinputdataInthispaper,theinputis38x380-1matrixwherecolumn2-38representseachcoveringandrow2-38representseachward.
Therefore,foreachcolumnandrow,theelementis1ifthewardiscoveredand0ifnotcovered.
E.
gthenameofthematrixisa,ifa[2][3]=1,itimpliesthatward3iscoveredbycovering2,otherwiseitisnotcoveredandthevaluewillbe0.
Thewholeinputfileformatforthisworkisshowninfigure4.
Theformatoftheoutputisinformofasolutionvectorcontainingonlyzerosandonesi.
e.
1ifacoveringisselectedand0ifnotselected.
Eachcoveringhasspecificnameofwardscoveringotherwardsthatarewithinthespecifieddistance.
(Thenameofeachwardandthenumberattachedtothemisshowninfigure3.
Figure3NamesofwardsandtheirnumberofidentificationTheinputmatrixshowninfigure4aandfigure4bwassavedastextfileandthebalasadditivealgorithmwasimplementedusingJavaprogramminglanguage.
Figure4aInputFileFormat1Obantoko,2Ikija,3Agooko,4ElegaHousing,5Iberekodo,6Agoika,7Ayetoro,8Okeago,9Totoro,10Itaosin,11Olorunda,12ImalaOrile,13IbaraOrile,14Ilewo/isaga,15Itaota,16Alagbagba,17Alabata,18Osiele,19Olodo,20Ilugun,21Agoodo,22Opeji,23Odeda,24Itesi,25Lafenwa,26Saje,27Itoko,28Ake,29Lantoro,30Ijemo,31Iporosodeke,32Irunbe,33Ijaye,34Okeitoku,35IjehunTitun,36Sabo,37Panseke.
JournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org56Figure4bInputFileFormatOncetheinputfilehasbeenselected,andthentheprogramcanberuntogeneratetheoutputrequired.
Theresultaftertheclickofthe"run"buttonisshowninfigure5below.
Figure5SetCoveringoutput.
JournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org57Thebalasadditiveresultaboveshowedthatcoveringisfoundandalsodisplayedthesolutionvector.
Itindicatedthatsixfirestationsareneededtoserveallthewardsandthelocationsofthosestationsareclearlystated.
ThesameresultwasobtainedfromTORAsoftwareintermsofoptimality,butthelocationsaredifferentandnotclearlystatedthoughitcanbetracedout.
Figure6ResultfromTORAsolverFigure6showstheresultoftheTORAsolver,theresultsissuchthatsixfirestationsarealsoneededtoserveallthewardseffectivelybutthelocationsofthefireservicestationarequitedifferentfromthatofBALASalgorithmthatwasimplemented.
Thelocationsindicatedbythesolverarelocations11,14,17,19,21,24whichcorrespondstothenamesofthefollowingwards(Olorunda,IlewoIsaga,alabata,Olodo,Ago-OdoandItesi)asshowninfigure3.
Torasolverdoesnotlistthenamesornumbersofthevillagesinitscovering.
5DISCUSSIONThenecessityofthedevelopmentoffacilitieslocationsoftwareforenhancingthedecisionmakingprocessandeventuallyproductivitycannotbeover-emphasized.
Theresultsobtainedinthisworkshowedthatsixfirestationsareneededtoserveeverywardsuchthatthemaximumdistancethatafirestationservicecangois10kilometers.
ItalsoshowedthatthelocationofthestationshouldbeObantoko,Olorunda,Ibaraorile,Olodo,OpejiandOdeda.
ThefirestationsatObantokowillrenderservicestotwentyfivewardswhichare:Obantoko,Ikija,Ago-oko,Elega,iberekodo,Ago-Ika,Ayetoro,Oke-ago,totoro,Ita-osin,Ita-ota,Ago-odo,lafenwa,Saje,Itoko,Ake,Lantoro,Ijemo,Iporo-sodeke,irunbe,Ijaye,Oke-itoku,Ijeun-titun,SaboandPanseke.
Olorundaservicestationwillservetwowardswhichare:OlorundaandImala-Orile.
Ibara-OrileservicestationwillserveIbara-orileandIlewo-Isagarespectively.
Olodoservicestationwillservethreewards.
Theyare:alagbagba,OlodoandIlugun.
Opejiwillservefourwardswhichare:alabata,OsieleandOpeji.
FinallyodedastationwillserveOdedaandItesi.
ThoughtheresultfromTORAsolveralsoindicatedthatsixfirestationsareneeded,itdidnotspecifytheactuallocationswherethestationsshouldbelocated.
ThisshortcominginTORAmakesouroutputandimplementationabetterone.
Theseresultsarepresentedinthetablebelowformoreclarity.
Itshowsthelocationswherethefacilitiesaretobeinstalledandalsothevillagestobecoveredbyeachofthefacility(coverings)onlyforbalasadditivealgorithm.
Theoutputfrombalasadditivealgorithmdoesnotshowfairdistribution.
InthecaseofthefacilityinlocationObantokowhichistoserve25locationswhileothersserveminimumoftwolocationsandmaximumofthree.
ThefacilityinObantokowillbeoverused.
JournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org58Table1:OUTPUTOFBALASADDITIVEALGORITHMLocationidentificationnumberLocationnameCoverings1Obantoko1,2,3,4,5,6,7,8,9,10,15,21,25,26,27,28,29,30,31,32,33,34,35,36,3711Olorunda11,1213Ibaraorile13,1419Olodo16,19,2022Opeji17,18,2223Odeda23,24TABLE2:OUTPUTOFTORASOLVERLocationIdentificationNumberLocationNameCoverings11OlorundaNotindicated14Ilewo-IsagaNotindicated17AlabataNotindicated19OlodoNotindicated21Ago-odoNotindicated24ItesiNotindicated6CONCLUSIONSANDRECOMMENDATIONSThisresultthereforeshouldraiseawarenessandcontributetotheaimofourgovernmenttoadoptthistoolwhichwilldefinitelyimprovethefunctionalityoffirestationsinNigeriabysavingalotofcitizen'slivesandproperties.
Itshouldalsobenotedthattheuseofthissystemisnotlimitedonlytofirestationsallocationalone,butalsotootherpublicfacilitieslikeschools,policestationsoastoincreaseresponsetimeandthereforereducecrime.
Itcanalsobeusedbyprivateestablishments.
REFERENCES[1]J.
EBeasleyandP.
C.
Chu,"Ageneticalgorithmforthesetcoveringproblem",EuropeanJournalofOperationalResearch,vol.
94,1996,pp.
392-404.
[2]V.
Chvatal.
Agreedyheuristicforthesetcoveringproblem.
MathematicsofOperationResearch,4(3):233-235,1979.
[3]U.
Fiege.
Athresholdforlnnforapproximatingsetcover.
JournaloftheACM,45(4):634-652,July1998.
[4]D.
S.
Johnson.
Approximationalgorithmsforcombinatorialproblems.
J.
Compute.
SystemScience9:256-278,1974.
[5]LarsonRichardC.
,"AHypercubeQueuingModelforfacilitylocationandredistrictinginurbanemergencyservices",ComputersandOperationsResearch1(1974)67-95.
[6]SavasES,"Simulationandcost-effectivenessanalysisofNewYork'semergencyambulanceservice",ManagementScience14(1969)608-627.
[7]M.
L.
FisherandP.
Kedia.
OptimalSolutionofsetcoveringproblemsusingdualheuristics.
ManagementScience,36:674-688,1990.
[8]FeoA.
andG.
C.
MauricioandA.
Resende,(2002)"AProbabilisticHeuristicforaComputationallyDifficultSetCoveringProblem",OperationsResearchLetters,8,67-71.
[9]A.
Caparara,M.
Fischetti,andP.
Toth.
AlgorithmsforsetcoverinproblemstechnicalreportOR-98-3,DEIS,UniversityofBologna,Italy1998.
[10]N.
ChristofidesandJ.
PPaixao.
Algorithmsforlargescalesetcoveringproblems.
AnnalsofOperationResearch,43:261-277,1993.
[11]M.
JBrusco,L.
W.
Jacobs,andG.
MThompson.
Amorphingproceduretosupplementasimulatedannealingheuristicsforcostandcoverage-correlatedsetcoveringproblems.
AnnalsofOperationResearch,86:611-627,1999.
[12]ToregasC.
,SwainR.
,RevelleC.
,andBergmanL.
,(1988)"ThelocationofJournalofTheoreticalandAppliedInformationTechnology2005-2009JATIT.
Allrightsreserved.
www.
jatit.
org59emergencyservicefacilities",Operationresearch19,pg1363-1373.
[13]ChurchRichardL.
,GerardRossa.
,(2003)"Themultilevellocationsetcoveringmodel"GeographicalAnalysisPublication,pg76-79.
[14]GareyM.
R.
andJohnsonD.
S.
,(2005)"Computerandinteractability:aguidetothetheoryofNP-completeness",pg34-38.
[15]Handya.
Taha.
,(2005)'OperationResearch:Anintroduction",Pearsoneducation(Singapore)Pte.
Ltd.
,pg391-397.
[16]Danielgulotta,(2006)"ApplicationoflinearProgrammingtosetcoveringandrelatedproblems"OnlinejournalofOperationResearch,pg1-2.
美国G口/香港CTG/美国T级超防云/物理机/CDN大促销 1核 1G 24元/月
[六一云迎国庆]转盘活动实物礼品美国G口/香港CTG/美国T级超防云/物理机/CDN大促销六一云 成立于2018年，归属于西安六一网络科技有限公司，是一家国内正规持有IDC ISP CDN IRCS电信经营许可证书的老牌商家。大陆持证公司受大陆各部门监管不好用支持退款退现，再也不怕被割韭菜了！主要业务有：国内高防云，美国高防云，美国cera大带宽，香港CTG,香港沙田CN2,海外站群服务，物理机，...
青云互联：洛杉矶CN2弹性云限时七折,Cera机房三网CN2gia回程,13.3元/月起
青云互联怎么样?青云互联是一家成立于2020年6月份的主机服务商，致力于为用户提供高性价比稳定快速的主机托管服务，目前提供有美国免费主机、香港主机、香港服务器、美国云服务器，让您的网站高速、稳定运行。目前，美国洛杉矶cn2弹性云限时七折，美国cera机房三网CN2gia回程 13.3元/月起，可选Windows/可自定义配置。点击进入：青云互联官网青云互联优惠码：七折优惠码：dVRKp2tP (续...
UCloud：美国云服务器,洛杉矶节点大促,低至7元起/1个月
ucloud美国云服务器怎么样?ucloud是国内知名云计算品牌服务商家，目前推出全球多地机房的海外云服务器。UCloud主打的优势是海外多机房，目前正在进行的2021全球大促活动参与促销的云服务器机房就多达18个。UCloud新一代旗舰产品快杰云服务器已上线洛杉矶节点，覆盖北美和亚太地区，火热促销中， 首月低至7元，轻松体验具备优秀性能与极高性价比的快杰云服务器。点击进入：ucloud美国洛杉矶...
solved为你推荐
-
cm域名注册什么是CM域名？.cm .cm域名asp主机asp虚拟主机租用哪里好？英文域名中文域名和英文域名有什么区别，越具体越好空间域名空间域名什么意思香港虚拟空间香港虚拟空间 好不、现在还有人买嘛免备案虚拟空间免备案的虚拟主机空间，买了以后会强制备案不？成都虚拟空间五星网络隶属于成都冠一科技有限公司，虚拟空间购买了不到一个月不能访问2次，质量真差啊！重庆虚拟空间重庆虚拟主机租用那家好?免费网站空间申请哪里有免费申请空间的（网页制作）虚拟主机控制面板虚拟主机管理面板与网站后台有什么区别?