Transport Logistics Optimization Model Of Passenger Car Based On Heuristic Algorithm

Transcription

1 MATEC Web of Conferences 44, ( 2016) DOI: / matecconf/ C Owned by the authors, publshed by EDP Scences, 2016 Transport Logstcs Optmzaton Model Of Passenger Car Based On Heurstc Algorthm Juan Zhou 1,2, Mao Sheng Zhong 3 Dao Le Cheng 1 and Xe Xn 3 1.School of Software 2.Intellgent Optmzaton & Informaton Processng Lab 3.School of Informaton Engneerng,East Chna Jaotong Unversty, Nanchang Abstract Passenger car logstcs transportaton problem s at the prmary stage of development n Chna, great economc loss and waste of resources caused n the process of logstcs transportaton,because most of enterprses n Chna rely on artfcal experence for gudance at present.and use of a partcular genetc algorthm to solve ths problem by prevous also exst many defects,such as programmng complexty, nstablty and low effcency etc.in vew of ths,the heurstc algorthm desgnng based on the greedy thought and the 0-1 knapsack thought was proposd, accordng buld a unversal model of the logstcs transportaton to the passenger car.expermental results show, ths model not only recordng the concrete scheme of loadng,also can make the optmzaton by changng the loadng sequence.etc.ths model s applcable to solve smlar problems of transport logstcs,and contanng great practcal applcaton value and economc value. Keywords Passenger car logstcs transportaton problem The greedy thought The 0-1 knapsack thought INSTRUCTIONS Passenger transport logstcs problem refers to the producton of passenger cars off the assembly lne after fnshedby means of transport, and dstrbuton to all dstrbuton warehouse, and fnally to a seres of storage and transportaton process n the hands of consumers.how to use the lmted means of transport, and to mnmze the cost of [1] to be complete ths process s a thorough study of the problems.at present, most of Chna's passenger car stowage schedulng logstcs companes manly rely on manual operaton, dependng on experence dspatchers, the prevalence of the followng questons:1) lack of scentfc methods to gude; 2) work-ntensve, low effcency; waste 3) car carrers capacty, hgh cost.for car carrers stowage plan and the lack of n-depth study of the lack of strong support software tools lead to the root cause of ths problem. Passenger transport logstcs problems can be dvded nto loadng problem and vehcle routng problem two sub-problems to solve.loadng Problem belong combnaton [2-3] Mathematcs category, the paper loadng problem based on rules that CLP (Contaner Loadng Problems) ssues.to solve ths problem, heurstc algorthm foregn scholars have George & Robnson (referred to as the G & R) algorthm [4], Ngo algorthm [5] and Loh & Nee algorthm [6].Domestc, He Dayong [7] Based on Space Decomposton layout heurstc genetc algorthm processng contaner loadng problem;wanch [8] proposed a hybrd genetc algorthm based on stereotypes and adaptve heurstc technques to solve the actual operaton of the most common weak heterogeneous contaner loadng problem;guo Hongwe [9] proposed the concept of "spatal layout pont", and are usng heurstc algorthms, genetc algorthms and smulated annealng algorthm.they common drawback of programmng complexty, nstablty, and the soluton process s slow.the dffculty of the problem s how to resolve the ssue as one-dmensonal mult-dmensonal problem and how to solve the ultmate problem of sub-problems, whch requres the use of dynamc programmng. Ths artcle based on deology and greed 0-1 knapsack [10-12] dynamc programmng deas to construct heurstc algorthm to construct logstcs model, and fnd the correspondng reprnt and dstrbuton programs to passenger transport and logstcs ssuesit s converted to P class of problems, and accurately solved n polynomal tme.the paper s arranged as follows: Secton I ntroduces the background passenger transport and logstcs problems, research status and the purpose and sgnfcance of the study; Secton ntroducng practcal problems to set forth the proposed Ths s an Open Access artcle dstrbuted under the terms of the Creatve Commons Attrbuton Lcense 4.0, whch permts dstrbuton, and reproducton n any medum, provded the orgnal work s properly cted. Artcle avalable at or

2 modelng scheme; s three, elaborated on the establshment and soluton models, and detals the correspondng algorthm; the fourth quarter, ths model s summarzed. Through modelng and solvng a practcal problem as an example,to elaborate modelng of the model soluton process, the queston summarzed as follows: passenger car manufacturer accordng to the customer's purchase orders natonwde (Table 1): NOTE: A representaton A land demand Transportaton logstcs company ssued to passenger cars to the task throughout the country, the logstcs companes plan to develop transport and dstrbuton n accordance wth the mandate ssued by these passenger cars.to ths end, the logstcs company frst they were selected from a number of vehcle car carrers can call "car carrers" (Table 2 below): Table 2 has a lst of car carrers Whch gves each passenger a car transport car loadng programs and destnatons, n order to ensure the completon of the transport task, the paper nvolved n car carrers n three models: the next upper each load a passenger, remember 1-1 type; n the upper load 1,2 respectvely column, referred to as 1-2 type; n upper varous load 2, recorded as 2-2. Load requrements are as follows: Each car transport car, the lower the loadng area can be consdered equvalent to a rectangle, each longtudnal row of passenger cars were placed longtudnally and transversely between adjacent passenger cars are at least as safe dstance between vehcles 0.1 m, lower strve to fll the upper two symmetrcal strve to ensure a smooth rde car carrers, the lne shown n Fgure 1. MATEC Web of Conferences C 76 E D 160 O Prevous secton detals the problems to be solved, ths secton wll establsh the correspondng mathematcal model of the problem.frst, the modelng process need to make the followng assumptons: (1) car carrers on standby after arrval, wthout emptyng return; (2) the cost per Unloadng almost neglgble; and (3) that the cost per unt dstance smplfy and model-ndependent; (4) does not consder the specal crcumstances that occurred durng the actual transport; (5) car carrers can not turn back. Ths problem has several dfferent types of car carrers and a number of dfferent models of cars, and there s a trangular network path shown n Fgure 1 [13]. Because of the lmted number of car carrers, and therefore should be the frst consderaton s to be able to hold all the goods vehcles, followed by the total dstance before consderng the ssue, the node D s a key node problem.in order to smplfy the problem, tryng to FIG dvded nto two lnes, but because of too many models of goods vehcles and not be able to enumerate how to assgn to two branch road, so t can be dvded nto two sub-path path drectly to the pont and the pont together transport, after transport drecton before gong to demand, as shown n Fgure E A C B 76 D 280 O Then you can start wth the frst one frst car carrers, accordng to the order for loadng, and the frst shpment of drecton, untl demand for B A p.2

3 ICEICE 2016 passenger cars and a varety of pont all 0, and then further drecton for shpment.durng shpment, due to the sze of the car models are dfferent, so the vehcle when nstalled, should determne the sze meets the requrements, and then decde to nstall or not nstall.after all the goods vehcles Bahran, accordng to backpack load process the recorded program, the calculaton of specfc per car carrers actual dstrbuton lnes.frst prncple and pont to stop at car carrers allocated, allocaton: go after the order enumerate drecton of the car, accordng to varous dstrbuton shpment demand for passenger cars, all to meet the farthest pont, the remanng amount of car all wll be docked at the D pont.but there are some car though docked n pont, but there s remove the part of the goods at the pont, ths may count for each tem separately vehcles remanng amount, whch n turn fnd the pont after the prevous, wth the remanng amount of demand for ths pont subtracton, f they oversupply, wll contnue to fnd ts former relay node.snce the backpack obtaned durng shpment program must be just to meet shpment demand, so eventually wll fnd ts most start unloadng pont. Obtanng the specfc dstrbuton plan, we got the number n each node docked car carrers, then you can calculate the total dstance of the program. 3.2 Model establsh Smplfed dagram of the transport and two ndependent drectons OC OA, the defnton y represents OC the knd of demand for passenger cars, empathy x s defned OA demand for passenger cars, sad frst knd, and car ( n) s defned as a passenger n pont of the frst speces on the passenger the demand for the knd of car, thereby obtanng the followng equaton: y car(c) car(d) x car (A) car (B) car (E) OC hypothess needs drecton k cars car carrers need to accomplsh ths k cars car carrers may nclude three models, In order to meet the needs of ts car carrer below flled wth as much as possble, let's start from the lower load, n order to meet the requrements for the bottom of the two columns to symmetrc, we try to fll up the columns, maxmze space utlzaton. Therefore, the ntroducton of the 0-1 knapsack algorthm to solve ths problem, and wth the dea of dynamc programmng to solve ths problem. Defnton methods: soluton(, s T,, l w, h ), That the use of a gven car carrers loaded the remanng -type passenger car program. Where s s currently used model car carrers, T represents the number of the remanng need to be loaded -type rde transport vehcles, lwhcar carrers currently usng the length and breadth. But ths s only the loadng program car carrers n a column poston. So for the 1-1 type car carrers, adjustment shall be co-loaded sequence twce from bottom to top: 2* soluton( s, T, l, w, h) 1-2 bass havng car from bottom to top, from left to rght and then co-loaded three tmes: 3* soluton( s, T, l, w, h) 2-2 type car as from bottom to top, from left to rght n all loadng four tmes: 4* soluton( s, T, l, w, h) G were used car transport car, you can get the followng formula: k R soluton(, s x,, l w, h )* t g 1 soluton( s, y, l, w, h ) * t, t {2, 3, 4} 3 jk T from the value of the s. As soluton(, s T,, l w, h) s a sngle sngle loadng problem, so the car transport the poston the lower sngle aspect hgh value can be determned accordng to lw h. We wll be the car car carrer sngle row as a contaner for l long, wth the gven passenger car flled accordng to the specfc space requrement.let dp( v) 1 to v represents the sze of the space has been flled, to obtan the followng formula: dp( v) dp( v - v ) 0?1: 0 If the sze s expressed v- v space has been flled, then the passenger load length of v, v space sze wll also be flled Note that ntally dp(0) 1. In the calculaton of the loadng process, the need to record the specfc crcumstances of each of the next load.after that, accordng to the route problem solvng concrete delvery vehcle specfc programs, ncludng the ntal program loadng, and unloadng volume ntermedate stops, as well as the fnal stops.let number( n ) represents the fnal stop n n number of car carrers, t can be calculated, the total number of mles traveled by car carrers to obtan the followng formula: D S ds( n, O) * number( n ) 5 A Where S represents the total number of mles traveled, ds( n, O ) represents the shortest p.3

4 MATEC Web of Conferences dstance O of n.requrements ds( n, O ), accordng to the followng formula (Floyd algorthm): ds(, j) Mn( ds(, k) ds( k, j), ds(, j)) 6 Loadng process of the algorthm s descrbed as follows: 1The cumulatve total for each model of each destnaton passenger demand OC drecton; 2Enumeraton of the -th car carrers j-cars, ts loadng vehcles. Accordng to the -th model car carrers, the electon of the correspondng load means (such as descrbed n 3.2); 3Enumerate each task s not yet complete loadng of the passenger, assumng k numbers vk passenger cars left, try to mount the frst knd of car carrers j -cars. And record the loadng programs; 4Analyzng the -th car carrers j-vehcle whether there s space, f there s, the mplementaton of k k 1, skp to Procedure 3), and f not, skp to process 5); 5Determne the drecton of the loadng task s completed, f not yet complete, execute j j 1, f j exceeds a gven range, executon jumps to 1, process 2) If you have completed skp to 6); 6To determne whether there s a path dd not complete loadng, f there s, skp to procedure 7), otherwse skp to process 8); 7Accumulated OA drecton of each destnaton the total demand for passenger cars for each model, the Executve j j 1, f j exceeds a gven range, executon 1, skp to Procedure 2); 8The loadng program records, calculates per car carrers transport path and calculate the fnal number of car carrers docked at each destnaton, the statstcal total dstance. Accordng stowage plans, calculate specfc delvery routes algorthm s descrbed as follows: 1) per enumeraton packed car carrers, startng from the current path farthest destnaton stowage, provded the current destnaton node ; 2) enumerate car carrers loaded passenger j, set up a total of v j vehcles; 3) Analyzng the current v j passenger cars exceeds n -pont demand for j passenger cars, f not exceed, home need[ n]. car[ j] 0, execute j j 1, skp to Procedure 2); otherwse, the process jumps to 4); 4) n calculated pont before a node pre[ n ], update v v need[ n]. car[ j] j j, whch need[ n]. car[ j ] represents node of n demand for passenger car j,put need[ n]. car[ j] 0 Prnt out the dstrbuton record n nodes, and set n pre[ n], such as n pre[ n] does not exst, skp to procedure 7); v j Analyzng s 0, and f yes, then execute j j 1, the process jumps to 10); f not, skp to Procedure 4); Determne whether there are car carrers useless statstcs, f there s, do 1, skp process 2), otherwse the algorthm ends. Accordng to the above models and algorthms to obtan results as shown n Table 3: Table 3 car carrers specfc dstrbuton pla Loadng mode (a, b) Unlouadng capacty n mddle pont Vehcle order Car name models Car layer A passenger car (X, Y, Z) whch unloudng Z termn number b vehccles of Y types n place X Upper layer (3,2)(5,3) C (3,2)(5,3) Upper layer (3,2)(5,3) (D,3,3),(D,5,5) C (3,2)(5,3) Upper layer (3,2)(5,3) (D,3,4), (D,5,6) D (3,2)(5,3) Upper layer (4,4) (D,3,2), (D,5,3) C (3,2)(5,3) p.4

5 ICEICE Upper layer (4,4) D (4,4) Upper layer (1,1)(2,2)(3,1) D (1,2)(2,1)(4,1) Upper layer (5,3) D (2,1)(5,4) Upper layer (3,2)(5,3) E (3,2)(5,3) Upper layer (3,2)(5,3) (A,3,1),(A,5,6) E (3,2)(5,3) Upper layer (3,2)(5,3) (A,3,4),(A,5,6) A (3,2)(5,3) Upper layer (3,2)(5,3) (A,3,4),(B,5,6) A (3,2)(5,3) Upper layer (4,4) (A,4,2) E (4,4) Upper layer (4,4) (A,4,8) A (4,4) Upper layer (4,4) (A,4,5),(B,4,3) A (4,4) Upper layer (1,2)(2,1)(4,1) (A,1,1)(A,4,5) E (4,4) Upper layer (1,1)(2,2)(3,1) (A,1,2)(A,2,2)(A,3,2) (B,2,1), E (1,1)(2,2)(3,1) Upper layer (1,2) (3,2) (B,1,2) (B,2,2)(B,3,2) A (1,1)(2,2)(3,1) Upper layer (5,2) B (3,4) Total dstance 5520 By adjustng the car carrers loadng order, you can get a dfferent load transport programs, the followng comparson between dfferent optons of ts exemplar 4 s as follows: Program Sequence Number of car carrers Total dstance Through Table 4 we found that the program loadng sequence 5 and 6 can be used to make car carrers number and total dstance are relatvely small. Thus show that the model can adjust the load order to acheve the effect of optmzaton models. The tme complexty of the model analyss as follows: Suppose a total of M cars need to use car carrers and dstrbuton N cars, car carrers whch length s L, then every - tme durng loadng tme complexty s ONL ( ), Assumng a postve nteger 2 k 4, the complexty of the whole loadng process there s M O( knl) 1. A number n for the mddle of the unloadng pont, and then calculate ts car carrer consumed by a specfc dstrbuton path tme complexty s OnM ( ). Let the path a total of Z stes, f Z 100,Algorthm Floyd [14] to solve the shortest path between ponts, tme p.5

6 MATEC Web of Conferences 3 complexty OZ ( ) ; If Z, algorthm SPFA [15-16] Solvng the shortest tme complexty s O (ZE), where E s the fgure of the total number of edges, So the algorthm's tme complexty s M O( knl) O( nm) mn( O( Z ), O( ZE)) 1 3, the model s sutable for large-scale car car carrer load dstrbuton scheme. The proposed model can be solved to a certan extent n our asked the passenger vehcle logstcs transportaton problem today. In the future work, wll do the further research on the model, the mprovement and optmzaton, and the correspondng software system, and ts promoton to the commercal market. Acknowledgments Ths work was supported by the Natonal Natural Scence Fund of Chna (No , No , No ), and by scence& technology pllar program of Jangx provnce (20151BBE50055), and by Scence Fund of ECJT Unversty(No.13RJ04) References [1] L Da,Captal Unversty of Economcs and Busness,Bejng,2012. [2] Zhang Jangjn, Combnatoral Packng Problem: Models and Algorthms[D].Shangha Jao Tong Unversty,2012. [3] ZHANG Jangjng,CHEN Feng.One-dmensonal Packng Problem wth Loadng Combnaton Contrants.Industral Engneerng and Management.2012,17(3): [4] Eberhard E.Bschoff,Mchael D.Marrott.A comparatve evaluaton of heurstcs for contaner loadng[j].european Journal of Operatonal Research 1990,44(2): [5] B.K.A.Ngo, M.L.Tay, E.S.Chua.Applyng spatal representaton technques to the contaner packng problem[j].internatonal Journal of Producton Research,1994,32(1): [6] T. H. Loh, A. Y. C. Nee.A packng algorthm for hexahedral boxes[c].in:proceedngs of the Conference of Industral Automaton, Sngapore,February 1992: [7] HE Dayong,E Mngcheng,CHA Janzhong,WANG Chun X,JIANG Y Dong.A Heurstc Approach to Contaner Loadng Problem Based on Space Decomposton and A Rule for Usage Rate of Packng Space, Journal of Computer-Aded Desgn & Computer Graphcs. 2000,125: [8] Huang Chuan.The Research of the Optmzaton Algorthm of the Contaner Loadng Problem.Fujan Normal Unversty [9] Guo Hongwe, Development of arborne packng scheme and software development of the logstcs system.qngdao,qngdao Unversty2005. [10] Shan Xaojun,Wu Supng. Solvng 0-1 Knapsack Problems wth Genetc Algorthm Based on Greedy Strategy,Computer Applcatons and Software,2010,27(12): [11] SHI Lan,LV Jan-hu.New optmzaton algorthm of 0-1 knapsack problem.applcaton Research of Computers,2014,31(4): [12] YU Juan,HE Yuyao. Hybrd algorthm based on Genetc Algorthm and estmaton of dstrbuton algorthm for 0-1 knapsack problem. Computer Engneerng and Applcatons, 2014,509: [13] SONG Wenfeng.The Research of Fnshed Vehcle Logstcs Based on Trangle Dstrbuton Network[D].Hefe Unversty of Technology,2012. [14] WANG Hua,A mxed algorthm of shortest path n logstcs and dstrbuton. Scence of Surveyng and Mappng, [15] SHEN Ha-lan,WANG Yu-bn,CHEN Za-lang,CA,Z-wen Improved SPFA Algorthm Based on Layered Graph,Computer Engneerng, 2012,3813: [16] WU Xao-zhen,LI Bao-ku,DONG Z-yan,LI Bo-xng ZHAN Jun The Optmzaton and Solvng of Vehcle Logstcs Transport Routes and the Mode of Transportaton Based on SPFA. Logstcs Engneerng and Management, 2014,365: Zhou Juan s a full lecture of School of Software at East Chna Jaotong Unversty, female,born on September, 1977.She s the nstructor of the Mathematcal Contest n Modelng and the coach of ACM ICPC(Internatonal Collegate Programmng Contest).Her current research nterests nclude desgn and analyss of computer algorthms and Graph Theory p.6