《系统工程》课程教学资源（英文文献）A Research on Express Logistics System Simulation

A Research on Express Logistics System SimulationBased on Multi-Resolution ModelingWith conversion ofentities among differentresolutions,AbstractMostof simulationisbasedonideal situation,butthereal logistics system is complex and variable. Multi-resolutionaggregation and disaggregation method is to ensure themodeling can meet needs of different environments on theinteractionofdifferententities in the sameresolutionsimulation, and adapt to the diversity, from different angles andModels aggregation is converting a high-resolution modelaspects of simulation modeling. Based on the research of MRM,to a low-resolution model, or to convert a high-resolutionthe method of aggregation and disaggregation is analyzed, thesubmodel to a low-resolution submodel, or to change aparadigm of aggregation and disaggregation was structured, andcomplicatemodel to a simple model.12i Models disaggregationthe conversion problem between MRMS is explained. MRM wasapplied in the modeling of Express Logistics System with ais the direct opposite.Aggregation usually takes place duringtwo periods, which indicate serial period and concurrentnetwork covering the whole country.period.[3l Serial period aggregation refers that high-resolutionKeywords:Multi-Resolution Modeling;ExpressLogisticmodel transfer data to low-resolution model after severalSystem; Aggregation and Disaggregationturns. Concurrent period aggregation refers to models runningsimultaneously,which concludes unidirectional data transferIIntroductionaggregation and two-way data transfer aggregation Theinteractivity of two-way data transfer in concurrent periodWith the simulation system complexity continuing toaggregation is the strongest, and the serial period aggregationimprove,avariety of requirements isputforward toexpressis the weakest.resolution of the logistics system simulation model. If aseparate model is made for the same systemtomeet each ofII.Theaggregation anddisaggregationparadigmof multi-the different needs, it will cost a lot of resources and time.Therefore,the reusabilityand conversion between differentresolutionmodelingresolutionmodelsoftheexpresslogisticssystemmodelnowIn this paper, the aggregation and disaggregation is dividedbecome keys to logistics modeling and simulation.into four parts, which are the aggregation and disaggregationof model, entity,attributeand the conversion of connectorStarting with analysis of the overall carrying capacity ofAggregation and disaggregation of model means thethe logistics network, based on aggregation and disaggregationcentralization and decentralization of models or the abstractparadigm of multi-resolution modeling,this paper has thespecification of thewholemodelunder differentandobject of improving customs responsiveness of logisticsresolutions.Entity aggregation and disaggregation means thenetworktoachievemulti-resolutionmodelingoftheexpressenlargementand deflation of a single entity,or themerge andlogisticssystem.dissolve of entities under different resolutions. Connectorconversion means the join, transition, and integration ofII.aggregation and disaggregation method of multi-modelsunderdifferentresolutions.Thispaperprovidestheresolution modelingaggregation and disaggregation paradigm of muiti-resolutionmodeling,by meansof tree-likevariables network modelModels with different resolution can't interact in differentTherefore,models in different resolutions mustbestates.convertedtobeinthesameresolution.[]Figure2. Treelike variables network modelThis paper assumes thas the low-resolution model,Figurel,The interactions between modelswhich is the ending moderof disaggregation and the startingLow reselutionIatenmediate中978-1-4244-6581-1/11/S26.00C2011IEEE

A Research on Express Logistics System Simulation Based on Multi-Resolution Modeling Abstract—Most of simulation is based on ideal situation, but the real logistics system is complex and variable. Multi-resolution modeling can meet needs of different environments on the simulation, and adapt to the diversity, from different angles and aspects of simulation modeling. Based on the research of MRM, the method of aggregation and disaggregation is analyzed, the paradigm of aggregation and disaggregation was structured, and the conversion problem between MRMS is explained. MRM was applied in the modeling of Express Logistics System with a network covering the whole country. Keywords: Multi-Resolution Modeling; Express Logistic System; Aggregation and Disaggregation I. Introduction With the simulation system complexity continuing to improve, a variety of requirements is put forward to express resolution of the logistics system simulation model. If a separate model is made for the same system to meet each of the different needs, it will cost a lot of resources and time. Therefore, the reusability and conversion between different resolution models of the express logistics system model now become keys to logistics modeling and simulation. Starting with analysis of the overall carrying capacity of the logistics network, based on aggregation and disaggregation paradigm of multi-resolution modeling, this paper has the object of improving customs responsiveness of logistics network, to achieve multi-resolution modeling of the express logistics system. II. aggregation and disaggregation method of multi￾resolution modeling Models with different resolution can`t interact in different states. Therefore, models in different resolutions must be converted to be in the same resolution.[1] Figure1. The interactions between models With conversion of entities among different resolutions, aggregation and disaggregation method is to ensure the interaction of different entities in the same resolution. Models aggregation is converting a high-resolution model to a low-resolution model, or to convert a high-resolution submodel to a low-resolution submodel, or to change a complicate model to a simple model.[2] Models disaggregation is the direct opposite. Aggregation usually takes place during two periods, which indicate serial period and concurrent period.[3] Serial period aggregation refers that high-resolution model transfer data to low-resolution model after several turns. Concurrent period aggregation refers to models running simultaneously, which concludes unidirectional data transfer aggregation and two-way data transfer aggregation. The interactivity of two-way data transfer in concurrent period aggregation is the strongest, and the serial period aggregation is the weakest. III. The aggregation and disaggregation paradigm of multi￾resolution modeling In this paper, the aggregation and disaggregation is divided into four parts, which are the aggregation and disaggregation of model, entity, attribute and the conversion of connector. Aggregation and disaggregation of model means the centralization and decentralization of models or the abstract and specification of the whole model under different resolutions. Entity aggregation and disaggregation means the enlargement and deflation of a single entity, or the merge and dissolve of entities under different resolutions. Connector conversion means the join, transition, and integration of models under different resolutions. This paper provides the aggregation and disaggregation paradigm of multi-resolution modeling, by means of tree-like variables network model.[4] Figure2. Treelike variables network model This paper assumes that Z is the low-resolution model, which is the ending model of disaggregation and the starting 978-1-4244-6581-1/11/$26.00 ©2011 IEEE

model of aggregation; X is the high-resolution model, which isoperating efficiencyoftheworkers,unreasonabledistributionthe ending model of aggregation and the starting model ofroute,andlowefficiencyoflogisticsbases,andthesefactordisaggregation; Y is between X and Z.X describes the mostmust be considered synthetically, which do not exist inisolation. Express logistics system analysis model is proposed,details, and Z describes the least. F, G, and G, stand for thebasedonthe actual situationof Company A,models ofaggregation function. Aggregation relation can be expressed,different resolutions are build,and express logistics systemtaking advantage of the following equation.simulation is achieved by aggregation and disaggregationZ= F(Y, Y2, Y)(1)Analsis efthe actY,= G(XI, X12)(2) ae(3)Y,= G(X31, X32, X3)Determinethemnmberofmulri-resolution levelMx,Ex,Ax,Cxstand for model,entity,attribute,andDefinesimulation objectives ofcopnector separately. X can be expressed in matrix [M, Ex, Axmmlti-resolution levelsCxEstablish iligh (low)-Xae(4)-[ Mx, Ex, Ax, Cxn]Simubation resultynchronizationAggregation(5)X12-[Mx12, Ex12, Ax12, Cx12]DtherCdisagercgation）or专informationmutti resotutioa models(6)X3/=[Mx31, Ex31, Ax31, Cx]consistencyEstablish low (hieglo)(7)X32=[Mx32, Ex32, Ax32, Cx32]eslobal(8)X=[Mx33, Ex3, Ax3, Cx3]optimirationAnalogously,YandZcanbeexpressed in matrixFigure3.Analysis model ofexpress logistics system(9)Y,-[Myi, Eyi, Ayi, C]ADetermine the numberof resolution levelsY2-[ My2, E y2, A y2, C2](10)Set up models of Company A in three different resolutions:operation-level model, logistics base-level model, network-(11)Y,=[ My3, E v3, A y3, C y3]levelmodel.Operating-level modelreflectsthesituationofZ=[ Mz, Ez, Az, C](12)each individual operational links, such as operating sequence,process time; logistics base level model reflects theF can also be expressed in matrix [Mr, Er, Ar, Cr], inoperational status of the logistics base, such as throughput,whichMr,Er,Ar,Crstandfortheaggregationfunctionoflayout reasonability,network-level model reflects themodel, entity,attribute, and connector during aggregating Yefficiencyofdeliverylogisticsnetwork,suchastherationalityintg eodiognbeed inmatrix[Mof transport routes,theproducts availability of terminal salesCGOperation-levelmodelrequireshighsimulationaccuracy,andF=[Mneeds more detailed description, which is the high-resolutionF, Er, Ar, C](13)model. While network-level mode requires low simulation(14)G,-[MG1, EG1, AG1, CG]]accuracy,and needs less detailed description, which is thelow-resolution model.(15)G,=[MG3, EG3, AG3, Cc3]Aggregation paradigm from high-resolution model to low-0000resolution model is asfollowed.(16)Z=[Mr, Er, Ar, C]'[Yi, Y2, Y](17)Y,-[MGl, EGl, AG, CG]'[X1, Xi2]Y3=[MG3, EG3, AG3, CG3]'[ X3, X32, X33](18)6IV.Express logistics system simulation based on multi-resolution modelingFigure4. Multi-resolution model levelsCompanyAisalargedomesticexpress transportationservicescompanywithanationwidedeliverylogisticssystem.B.Define simulationobjectives ofmulti-resolution levelsHeadquartered inChengduthe country has6 first-classModels of Three resolution levels have differentlogistics bases, 11 second-class logistics bases and 23 deliverycharacteristics, and so are simulation goals. Network-levelcenters.model needs to reduce overall inventory,by demand analysisIncreasingportfolio leadsto deliverynetworkproblems,and path optimization.Logistics base-level model needs toand customers often complain about the low efficiency ofoptimizethelayout,reducetheexpressinventoryoflogisticsdelivery.Through analysis, express delivery and lowbase,andincreaseturnoverratio.Operation-levelmodelneedsefficiency may be caused by many factors,such as low

model of aggregation; X is the high-resolution model, which is the ending model of aggregation and the starting model of disaggregation; Y is between X and Z. X describes the most details， and Z describes the least. F, G1 and G3 stand for the aggregation function. Aggregation relation can be expressed, taking advantage of the following equation. Z= F( Y1 , Y2 , Y3) (1) Y1= G1(X11, X12) (2) Y3= G3(X31, X32, X33) (3) Mx , Ex , Ax , Cx stand for model, entity, attribute, and connector separately. X can be expressed in matrix [Mx , Ex , Ax , Cx ]. X 11=[ MX11, EX11, AX11, CX11] (4) X12=[ MX12, EX12, AX12, CX12] (5) X31=[ MX31, EX31, AX31, CX31] (6) X32=[ MX32, EX32, AX32, CX32] (7) X33=[ MX33, EX33, AX33, CX33] (8) Analogously, Y and Z can be expressed in matrix. Y1=[ MY1, E Y1, AY1, C Y1] (9) Y2=[ MY2, E Y2, AY2, C Y2] (10) Y3=[ MY3, E Y3, AY3, C Y3] (11) Z=[ MZ, EZ, AZ, CZ] (12) F can also be expressed in matrix [MF, EF, AF, CF], in which MF, EF, AF, CF stand for the aggregation function of model, entity, attribute, and connector during aggregating Y into Z. Accordingly, G can be expressed in matrix [MG，EG，AG， CG ]. So, the following equations are got: F=[M F, EF, AF, CF] (13) G1=[MG1, EG1, AG1, CG1] (14) G3=[MG3, EG3, AG3, CG3] (15) Aggregation paradigm from high-resolution model to low￾resolution model is as followed. Z=[MF, EF, AF, CF] T [ Y1 , Y2 , Y3] (16) Y1=[MG1, EG1, AG1, CG1] T [ X11, X12] (17) Y3=[MG3, EG3, AG3, CG3] T [ X31, X32, X33] (18) IV. Express logistics system simulation based on multi￾resolution modeling Company A is a large domestic express transportation services company with a nationwide delivery logistics system. Headquartered in Chengdu, the country has 6 first-class logistics bases, 11 second-class logistics bases and 23 delivery centers. Increasing portfolio leads to delivery network problems, and customers often complain about the low efficiency of delivery. Through analysis, express delivery and low efficiency may be caused by many factors, such as low operating efficiency of the workers, unreasonable distribution route, and low efficiency of logistics bases, and these factor must be considered synthetically, which do not exist in isolation. Express logistics system analysis model is proposed, based on the actual situation of Company A, models of different resolutions are build, and express logistics system simulation is achieved by aggregation and disaggregation. Figure3. Analysis model of express logistics system A. Determine the number of resolution levels Set up models of Company A in three different resolutions: operation-level model, logistics base-level model, network￾level model. Operating-level model reflects the situation of each individual operational links, such as operating sequence, process time; logistics base level model reflects the operational status of the logistics base, such as throughput, layout reasonability; network-level model reflects the efficiency of delivery logistics network, such as the rationality of transport routes, the products availability of terminal sales. Operation-level model requires high simulation accuracy, and needs more detailed description, which is the high-resolution model. While network-level mode requires low simulation accuracy, and needs less detailed description, which is the low-resolution model. Figure4. Multi-resolution model levels B. Define simulation objectives of multi-resolution levels Models of Three resolution levels have different characteristics, and so are simulation goals. Network-level model needs to reduce overall inventory, by demand analysis and path optimization. Logistics base-level model needs to optimize the layout, reduce the express inventory of logistics base, and increase turnover ratio. Operation-level model needs

tooptimizetheoperation process,reduce wasted time,andC(Connector parameter)-[operatingparameters of theimproveprocessingpowerpreceding step, the number ofexpress in process, machineutilization,productivity...JC.Establishhigh-resolutionmodelofoperation levelLow-resolution model aggregated from high-resolutionEstablish high-resolution simulation model after definingD.modelmodeling goals. Operational-level model requires a higherdegree of accuracy, and needs to reflect the production linksI) Establishlogisticsbase-levelmodelfor more details.Enterprise Dynamic is suitable for discreteLogistics base-level model with low resolution isevent simulation, in particular for the simulation of theaggregated from several operation-level models with lowerlogisticsproductionsystem,sooperationalaspectsoftimeresolution. Take an example of Wuhan logistics base whichwasting,inefficient, unreasonable action and other issues arehas 19 operation station, combined with 19 operation-levelsolved by use of the software. FigureI shows a high-model. Because location is an essential way to improveresolution simulation model of theoperational link.logistics circumstance,save time,reduce cost, and betterlogistics, so the logistics base stocks are reduced by thefacility layout. ExtendSim simulation software has strongscalability and compatibility to establish the logistics baselevel model.M(Model parameter)-[logistics basearea, regionalclassification ofthe logistics base,run time...JE(Entity parameter)-[physical area, position relationshipFigure5.High-resolution model of operation leveltheexpress amountof input and output...Parameters of operation-level model can be expressed inA(Attribute parameter)-[equivalent amount of logistics,vectorformvacancy rate of the operation, utilization, headcount, staffM(Model parameter)-[run time, operation sequence oftime...Jstuff, operation area..JC(Connector parameter)-[The amount of output and inputE(Entitylocation,parameter)-[entityentityfromthe logistics base,Thetotal amountof logisticsbase...Jrelationship...JWith operation-level model aggregating,data transfer toA(Attribute parameter)-[operating parameters of machine,the logistics base-level model.Aggregation relationshipidentificationinformation ofprocessability.express,betweenthetwomodelsisinterpretedinthetablesbelow.headcount,staff time...JTABLEIAGGREGATIONRELATIONSHIPBETWEENOPERATION-LEVELMODELANDLOGISTICSBASE-LEVELMODELOperation-levelLogistics base-levelOperation-levelLogistics base-levelModel running timeEntity running timeOperation areaEntity physical areaTotalnumberofemployeesatAmountofemployeesVacancyrateof staff ofVacancyrateof staff ofoperationrequired for entityoperationentityAmount of processing expressAmount of processing expressOverallmachineUtilization of entityin the modelin the entityutilization of operationExpress completionrateofExpressarrivalrateoExpress completion rateExpress arrival rate of entityoperationofoperationentityExpress type processing oProcessing cost ofProcessing cost of entityoperationentityoperationMulti-resolution aggregation is achieved by use of multi-=G[X11, X12,.., X19]resolution aggregation paradigm This paper assumes that=[MG1, El, AG1, CG]' [Xi, Xi12,., X19]Wuhan logistics is represented byY,and model parameter.entity parameter,attributeparameter,and connector parameterMol'EC oare expressed asMyi,EyisSYl,Cyi.Operation model is..+Me s)12AX12X12ACFIM'TT.EMXI2Arepresented as Xi1, Xi2,,Xi19. Aggregation functionx119x119x119G11X119between them is Gr,G,-[MGi,EGi,SGl,CGi], in which Mgt,Plus sign does not mean addition, but aggregation. It isEGl, SGi, and CGr stands for aggregation format of model,plusand minus,multiply,ortakethemaximumnumberofentity,attribute,and connector.So, these equations arevariables,specific meaning ofwhich is dependent on theavailable.Y,=[Myi, Ey, Ayi, Ci]

to optimize the operation process, reduce wasted time, and improve processing power. C. Establish high-resolution model of operation level Establish high-resolution simulation model after defining modeling goals. Operational-level model requires a higher degree of accuracy, and needs to reflect the production links for more details. Enterprise Dynamic is suitable for discrete event simulation, in particular for the simulation of the logistics production system，so operational aspects of time wasting, inefficient, unreasonable action and other issues are solved by use of the software. Figure 1 shows a high￾resolution simulation model of the operational link. Figure5. High-resolution model of operation level Parameters of operation-level model can be expressed in vector form. M(Model parameter)=[run time, operation sequence of stuff, operation area .] T E(Entity parameter)=[entity location, entity relationship.] T A(Attribute parameter)=[operating parameters of machine, process ability, identification information of express, headcount, staff time.] T C(Connector parameter)=[ operating parameters of the preceding step, the number of express in process, machine utilization, productivity.] T D. Low-resolution model aggregated from high-resolution model 1) Establish logistics base-level model Logistics base-level model with low resolution is aggregated from several operation-level models with lower resolution. Take an example of Wuhan logistics base which has 19 operation station, combined with 19 operation-level model. Because location is an essential way to improve logistics circumstance, save time, reduce cost, and better logistics, so the logistics base stocks are reduced by the facility layout. ExtendSim simulation software has strong scalability and compatibility to establish the logistics base level model. M(Model parameter)=[logistics base area, regional classification of the logistics base, run time .] T E(Entity parameter)=[physical area, position relationship, the express amount of input and output.] T A(Attribute parameter)=[equivalent amount of logistics, vacancy rate of the operation, utilization, headcount, staff time.] T C(Connector parameter)=[The amount of output and input from the logistics base, The total amount of logistics base.] T With operation-level model aggregating, data transfer to the logistics base-level model. Aggregation relationship between the two models is interpreted in the tables below. TABLE I. AGGREGATION RELATIONSHIP BETWEEN OPERATION-LEVEL MODEL AND LOGISTICS BASE-LEVEL MODEL Operation-level Logistics base-level Operation-level Logistics base-level Model running time Total number of employees at operation Amount of processing express in the model Express arrival rate of operation Express type processing of operation Entity running time Amount of employees required for entity Amount of processing express in the entity Express arrival rate of entity Express type processing of entity Operation area Vacancy rate of staff of operation Overall machine utilization of operation Express completion rate of operation Processing cost of operation Entity physical area Vacancy rate of staff of entity Utilization of entity Express completion rate of entity Processing cost of entity Multi-resolution aggregation is achieved by use of multi￾resolution aggregation paradigm. This paper assumes that Wuhan logistics is represented by Y1 , and model parameter, entity parameter, attribute parameter, and connector parameter are expressed as MY1, E Y1, S Y1, C Y1.Operation model is represented as X11, X12, ., X119. Aggregation function between them is G1 , G1=[MG1, EG1, SG1, CG1], in which MG1, EG1, SG1, and CG1 stands for aggregation format of model, entity, attribute, and connector. So, these equations are available. Y1=[ MY1, E Y1, AY1, C Y1] = G1[X11, X12,., X119] =[MG1, EG1, AG1, CG1] T [X11, X12,., X119] = [MG1, EG1, AG1, CG1] T [ MX11, EX11, AX11, CX11]+ [MG1, EG1, AG1 , C G1 ] T [ M X12 , E X12 , A X12 , C X12 ]+ . +[MG1 , E G1 , A G1 , C G1 ] T [ M X119 , E X119 , A X119 , C X119 ] (19) Plus sign does not mean addition, but aggregation. It is plus and minus, multiply, or take the maximum number of variables, specific meaning of which is dependent on the

function.Wuhan logistics base-level model is established by-aggregation andExtendSim simulation software.国司OAccording to process routes, the logistics is analyzed.CAD drawings are used to get the distance between theDNoperating,and simulation model are used to get logisticscapacity.-Figure6.Facility layout of Wuhan logistics base:TABLE I.LOGISTICS CAPACITY-DISTANCEbetweenlogisticsbetweenlogisticsdistanceproductdistanceproductpointsquantitypotms200925909.1418-64917366429.343124.1151974.69-4-5411532431560117426452913.5262134.4497725.21237624.711058120334546.8320078.723-7237077156885.64134862.878101358120366838.6238846.816960688104540.8727675.813-448322897110.038-898849446926.293122587.6121177.141255248997413.2453819.7491824237017545.2872390.4851934417098588800.02242370102388.9248161215181615851933761442.2516621113683961.9317727.391183095-85-16258676087743.49226971.4621113627205.055743.965161219874644665.0298207.034852440050362.826410.2535880021340773206168420.6477041.5627642.5112178According to the logistics and distance between points, theNetwork-level model has the lowest resolution, and it does notmap logistics capacity-distance is draw..Thehorizontal axishavespecific description of the operating status of eachdenotes the amount of logistics,axisindicateslogistics base, but coordinates the express deployment of thethe verticaldistance.whole network, optimizes the logistics network,reducesoverall inventory levels, increases turnover.momeThe parameters of the network-level model are given,-sooosimilar to the conversion from operation-level1modeltologistics base-level modelBOXEEEAx000M(Model parameter)-[running time, delivery area.. ]"o100000.E(Entity.parameter)-[entityphysicalarea,entitysol15e000relationship, entity connection style..JFigure7. Logistics capacity-distant mapA(Attribute parameter)-[input and output of entity,From above it is concluded that, the overall layout of theinventory of entity, unit transport costs, utilization, maximumlogistics base in Wuhan is reasonable, but the local operatorcapacity...Jspaces need to be further improved.C(Connectorparameter)=[production,,transportcost2)Establish network-level modelcustomdemands...JNetwork-level model of expresslogistics system isestablished by use of logistics base-level model.Each logisticsTheaggregation relationship from logistics base-levelbase is a network node.Network-level model consists of 17model to network-level model has been showed as followinglogistics base-level model and23 delivery centermodelsTABLE III.ANALYSIS MODEL OF EXPRESS LOGISTICS SYSTEMSLogistics base-levelNetwork-levelLogisties base-levelNetwork-level Model running timeEntity running timeLogistics base areaEntity physical area

function. Wuhan logistics base-level model is established by aggregation and ExtendSim simulation software. According to process routes, the logistics is analyzed. CAD drawings are used to get the distance between the operating, and simulation model are used to get logistics capacity. Figure6. Facility layout of Wuhan logistics base between logistics points quantity TABLE II. distance product LOGISTICS CAPACITY-DISTANCE between logistics distance product points quantity 18—4 649173 66429.3 43124.11 2—5 2006037 25909.14 51974.69 4—11 1174264 52913.52 62134.44 5—3 2431560 97725.21 237624.7 11—10 581203 34546.83 20078.72 3—7 2370771 56885.64 134862.8 10—13 581203 66838.62 38846.81 7—8 6960688 104540.8 727675.8 13—4 483228 97110.03 46926.29 3—8 988494 122587.6 121177.1 4—12 552489 97413.24 53819.74 9—18 242370 72390.48 17545.28 12—15 519344 170985 88800.02 18—16 242370 102388.9 24816 15—8 519337 118309 61442.25 16—6 211136 83961.93 17727.39 5—16 2586760 87743.49 226971.4 6—8 211136 27205.05 5743.965 16—1 2198746 44665.02 98207.03 4—8 524400 50362.8 26410.25 1—2 2134077 32061 68420.64 17—8 358800 77041.56 27642.51 According to the logistics and distance between points, the map logistics capacity-distance is draw. The horizontal axis denotes the amount of logistics, the vertical axis indicates distance. Figure7. Logistics capacity-distant map From above it is concluded that, the overall layout of the logistics base in Wuhan is reasonable, but the local operator spaces need to be further improved. 2) Establish network-level model Network-level model of express logistics system is established by use of logistics base-level model. Each logistics base is a network node. Network-level model consists of 17 logistics base-level model and 23 delivery center models. Network-level model has the lowest resolution, and it does not have specific description of the operating status of each logistics base, but coordinates the express deployment of the whole network, optimizes the logistics network, reduces overall inventory levels, increases turnover. The parameters of the network-level model are given, similar to the conversion from operation-level model to logistics base-level model. M(Model parameter)=[running time, delivery area.] T E(Entity parameter)=[entity physical area, entity relationship, entity connection style.] T A(Attribute parameter)=[input and output of entity, inventory of entity, unit transport costs, utilization, maximum capacity.] T C(Connector parameter)=[production, transport cost, custom demands.] T The aggregation relationship from logistics base-level model to network-level model has been showed as following. TABLE III. ANALYSIS MODEL OF EXPRESS LOGISTICS SYSTEMS Logistics base-level Network-level Logistics base-level Network-level Model running time Entity running time Logistics base area Entity physical area

Express arrival rate of logisticsThe rate of processing expressIssue rate of the expressExpress arrival rate of entitybaseExpress capacity of logisticsStaff amountStaff amount of entitybaseMaximum capacity of entityExpress amount inExpress amount in processing ofCapacity oflogistics baseUtilization of entityentityVariety of express in entityLocation of logistics baseOperating costs of logistics baseOperating costs of entityThe network-level model is established in application ofFTheoveralloptimizationofmodelsaggregation paradigm.Assumed that, network-level model isExpress logistics systemmodel is established,modelingrepresented as Z, and model parameter, entity parameter,time is reduced, cost is cut down, and good results are got,attribute parameter,connector parameter are represented astaking advantage of multi-resolution simulation technologyMz, Ez, Sz, Cz.Y,Y2, .,Yi7 stand for logistics base-leveland aggregation paradigm. With multi-resolution modeling.models, containing 6 first-level logistics bases and 11 second-the bottleneck operator spaces are eliminated, and the stafflevel lpsistisskaseseAggregationfuntionseate-MetiSnsSorWithrelocatingWuhanoperation actionsareimproved.model, entities, Attributes, interfaces from logistics base-levelBeijing logistics base,logistics is smootherand efficiencyismodel to a network-level model of.Thefollowing equations canimproved.By changing delivery rout of logistics network, thebe gottaking advantage of theparadigm,networkisoptimizedandcustomssatisfactionisimprovedAfter initial improvement,the average processing capacityZ=F[YI, Y.,Yi]increased by 15%, cost logistics of the base reduced by 23%,while the express logistics network capacityincreased by24% [M, E, Ar, C][Y, Y2..,Y]overallandcustomersatisfactionincreasedby26%= [M, Er, Ar, C]' [ My, Eyl, A y, C]+V.Conclusionn ErAr,' [Mn, Ev,A ye, .+[M, r2?(20)In the establishment of multi-resolution simulationmodelCETLY17AY17CY17Y17conversion of different resolution models is the core.By useIn theseequations,plus sign means aggregation.Further more,of aggregation paradigm,EnterpriseDynamics andExtendSimnetwork-level model also includes the delivery center models. By7.0 simulation software,the express logistics systemacrossuseofnetwork-levelmodelaggregationandExtendSimthe country is simulated to improve processes, optimizesimulation software,models are builtlogistics base layout, and enhance the network's overalloperational capability, based on the existing capacity of the高山network.口口aFigure8. Network-level modelE.consistency and synchronization of the simulation resultBy use of aggregation paradigm, operation-level model,logistics base-level model and network-level model areconnected.Data is delivered from high-resolution simulationmodel to low-resolution simulation model, and simulationmodel of low-resolution data back to the high-resolutionmodel at the same time.Different resolution models areconverted to the same resolution using aggregation paradigm.Express network systemmodelachieves the mappingbyparadigmshift,result comparisonconsistency3andverification. In addition, time variable is set in the systemmodel, is called simulation time, to achieve the consistency ofsimulation clock under different resolutions by means oftransfer function

Issue rate of the express The rate of processing express Staff amount Staff amount of entity Express amount in Express amount in processing of processing entity Express arrival rate of logistics base Express arrival rate of entity Express capacity of logistics base Maximum capacity of entity Capacity of logistics base Utilization of entity Location of logistics base Variety of express in entity Operating costs of logistics base Operating costs of entity The network-level model is established in application of aggregation paradigm. Assumed that, network-level model is represented as Z1 , and model parameter, entity parameter, attribute parameter, connector parameter are represented as MZ, EZ, S Z, C Z. Y1 , Y2 , ., Y17 stand for logistics base-level models, containing 6 first-level logistics bases and 11 second￾level logistics bases. Aggregation function is F, F=[MF, EF, SF, CF ], in which M F , E F , S F , C F , it is the aggregate functions of model, entities, attributes, interfaces from logistics base-level model to a network-level model of. The following equations can be got taking advantage of the paradigm. Z=F[Y 1 , Y2 ,.,Y17] = [MF, EF, AF, CF] T [Y1 , Y2 ,.,Y17] = [MF, EF, AF, CF] T [ MY1, E Y1, AY1, C Y1]+ [MF, EF, AF, CF] T [ MY2, E Y2,A Y2, C Y2]+.+ [MF, EF, AF, CF ] T [ M Y17 , E Y17 , A Y17 , C Y17 ] (20) In these equations, plus sign means aggregation. Further more, network-level model also includes the delivery center models. By use of network-level model aggregation and ExtendSim simulation software, models are built. Figure8. Network-level model E. consistency and synchronization of the simulation result By use of aggregation paradigm, operation-level model, logistics base-level model and network-level model are connected. Data is delivered from high-resolution simulation model to low-resolution simulation model, and simulation model of low-resolution data back to the high-resolution model at the same time. Different resolution models are converted to the same resolution using aggregation paradigm. Express network system model achieves the mapping consistency by paradigm shift, result comparison and verification. In addition, time variable is set in the system model, is called simulation time, to achieve the consistency of simulation clock under different resolutions by means of transfer function. F. The overall optimization of models Express logistics system model is established, modeling time is reduced, cost is cut down, and good results are got, taking advantage of multi-resolution simulation technology and aggregation paradigm. With multi-resolution modeling, the bottleneck operator spaces are eliminated, and the staff operation actions are improved. With relocating Wuhan Beijing logistics base, logistics is smoother and efficiency is improved. By changing delivery rout of logistics network, the network is optimized and customs satisfaction is improved. After initial improvement, the average processing capacity increased by 15%, cost logistics of the base reduced by 23%, while the express logistics network capacity increased by 24% overall, and customer satisfaction increased by 26%. V. Conclusion In the establishment of multi-resolution simulation model, conversion of different resolution models is the core. By use of aggregation paradigm, Enterprise Dynamics and ExtendSim 7.0 simulation software, the express logistics system across the country is simulated to improve processes, optimize logistics base layout, and enhance the network's overall operational capability, based on the existing capacity of the network