Research on the Process-level Production Scheduling Optimization Based on the Manufacturing Process Simplifies

Transcription

1 Internatonal Journal of Smart Home Vol.8, No. (04), pp Research on the Process-level Producton Schedulng Optmzaton Based on the Manufacturng Process Smplfes Y. P. Wang,*, L. Gao, Z. M. Sh, K. X. B and J. H. Ge Harbn Unversty of Scence and Technology, Harbn, Chna, Qqhar Heavy CNC Equpment Corp. LTD., Qqhar Helongang Provnce, Chna,6005 wypbl@6.com Abstract Enterprses to acheve global optmzaton and better dversfcaton and personalzed product producton schedulng, from the unty of the tasks and resources manufacturng, based on the perspectve of global optmzaton n producton schedulng method. Wth global producton schedulng of matchng and process model resources as the foundaton for the realzaton of enterprse process producton schedulng optmzaton and level contnue workng process decouplng pont, ths paper manufacturng task and manufacturng process s dvded nto generalzaton and dfferental theory, applcatons from manufacturng process optmzaton results obtaned path, based on the manufacturng process optmzaton mathematcal model to study the producton schedulng, and by usng the mproved genetc algorthm, and fnally for example. Above schedulng method smplfed the schedulng process, reduce the schedulng of dffculty, shorten the tme schedulng, and fnally acheve enterprse resource effcent use. Keywords: Global optmzaton; Manufacturng process optmzaton; Process decouplng pont; Producton schedulng model. Introducton Wth the development of socety and the development of technology of, manufacturng producton way also has the evoluton and change constantly. In order to pursung hgh effcency and low cost. Henry Ford put forward "a sngle product prncple", namely mass producton mode. However, ths need for a sngle product market factors support the producton of a stable market demand and long product lfe cycle s no longer applcable n today's market demand for ncreasngly dverse []. As products become ncreasngly promnent personalty needs, and the gradual shortenng of product lfe cycle, the mass customzaton mode appears. Its products have become dversfed and customzed through flexble and rapd response [-]. However completely custom servce cost the demand s hgh, therefore, mass producton and mass customzaton of combnng the new method appeared.for manufacturng the mass customzaton shorter lead tme and reduce the cost of the two aspects of the research, above ths new method was put forward based on the predcton of the mass producton stage and order based on the batch manufacturng stage and fully customzed producton stage producton of the ndvdualzed product of the three phase models, t to the model as appled to varous envronmental factors were analyzed to obtan under mass customzaton born of the optmzaton of the scheme []. Wth customers on the enterprse producton demand ncrease, the correspondng rapd ISSN: IJSH Copyrght c 04 SERSC

2 Internatonal Journal of Smart Home Vol.8, No. (04) response to the customer order needs of agle manufacturng producton mode [4-5]and effectvely reduce the producton cost [6-7] and make full use of network resources sharng between enterprses manufacturng producton mode and ts correspondng schedulng method arse and gradually development [8]. However, there have been many schedulng not ust need to a sngle producton envronment and the mode of dong research, and s more need to consder n two or more envronment and mode of the actual producton schedulng problem, and producton plannng and schedulng are enterprse actual producton of mportant research obect. Throughout the two perspectves, both macro and mcro level of nformaton and ssues to meet the global producton schedulng and parallel to the upper producton plannng, you can acheve global optmzaton of the entre producton process. Manufacturng process and producton schedulng process s closely lnked. There are a lot of manufacturng process modelng, optmzaton, but not to make ts and producton schedulng problem closely and full consderaton of the manufacturng process and the relatonshp between the producton schedulng process. For smplfcaton of the manufacturng process can acheve the optmzaton of producton schedulng to smplfy and thus facltate the schedulng. Through the above analyss, consdered from the perspectve of corporate global optmzaton, mult-enterprse resource optmzaton confguraton, wth each low-level producton schedulng, and through the manufacturng process s smplfed, so as to acheve the complexty of products n dfferent producton mode parallel producton schedulng multple obectves, and the smplfed producton schedulng optmzaton process has yet to be carred out comprehensve and n-depth study. Ths thess s n the enterprse global optmzaton to focus research on the bass of ts producton schedulng, shop-level schedulng mathematcal model and ts soluton to the completon of the process-level producton schedulng optmzaton.. Based on the Optmzaton of the Manufacturng Process Producton Schedulng Model s Establshed.. The Idea of Model s Establshed Exstng producton schedulng method n dfferent classfcaton accordng to order the schedulng method. However enterprse processng producton s the smallest unt process. Producton schedulng research n ths artcle focused on the process perspectve, the process accordng to the characterstcs of the process as a unt producton schedulng problems and producton schedulng optmzaton. Decouplng pont of the ntroducton of processes n ts upper part, the paper manufacturng process s dvded nto generc and dfferentaton, and drawn on the bass of the ndvdual parts processng operatons to choose from a number of optonal path, accordng to the process decouplng pont n the manufacturng process the poston of each of the tme and cost to establsh a unversal and dfferentated two-stage producton schedulng mult-obectve mathematcal model s solved to arrve at all the product components, producton schedulng optmal path and the path length and path processng costs. The schedulng method has the advantage as follows: () Complexty s sgnfcantly reduced. Ths s because the process decouplng pont postonng all the processes are dvded nto generc and dfferentated; shop schedulng and process the optmal soluton of the decouplng pont get a complex problem nto a smple 8 Copyrght c 04 SERSC

3 Internatonal Journal of Smart Home Vol.8, No. (04) queston and complex ssues. Combnaton, whch makes producton schedulng problem solvng greatly smplfy. () Reduce the dynamc randomness of manufacturng systems. When the arrval tme or processng tme of a ob change, the tradtonal producton schedulng process requres a lot of changes, process decouplng pont locaton-based producton schedulng process s composed of two phases, correspondng by changes caused by such problems wll be much smaller... The Obectve Functon Selecton and Model Parameter Settngs. The obectve functon Manufacturng producton schedulng s necessary to meet the delvery perod, but also ensure that the processng costs of the mnmum characterstcs to meet customer demand for personalzed on the bass of takng nto account the schedulng between multple process equpment occupy the tme and herarchy between the processng operatons. Therefore, n ths paper the schedulng obectve and constrants obectve are mnmze processng costs and mnmze processng tme.. Model parameters () Producton plant processng equpment, collecton of resource nodes are R N ' R N, R N,..., R N,..., R N, where m m the total number of processng nodes s. N l s the number of the actvtes of the workshop machnng tasks MT. k s the argument n the processng tasks cable, k,,..., N l.() C s the total cost of the processng tasks MT.(),, C k s the cost of processng task n MT, k actvtes usng the resource.(4) t, k, s Start tme of the frst k actvtes n the processng task n MT.(5) t, k, e s the completon tme of the frst k actvtes n the processng task n MT.(6) t s the executon tme of the workshop machnng tasks MT.(7) T Completon date for the end of the workshop machnng tasks MT.(8) DT s the delvery of the processng tasks MT.() Just lke the Formula -, k s the ndex of system., P r o c e s s n g a c tv te V? n M T c o m p le te d n e q u p m e n t r e s o u r c e s n o d e R N 0, o th e r w s e k k (-) (0)If there s to e E, P R E S E T ( T ) T, followng the actvtes set, S U C S E T ( T ) s T successor collecton actvtes. And T s the follow-up actvtes of T, T s T former after actvtes.processng actvtes n the prorty tasks defned as Formula -. p r 0, If P R E S E T (T ), T m a x P r ( T ), o th e r w s e T p reset (T ) (-) ()If T s T former followng, Must be after the completon of T for T,s pr ( T ) pr ( T ).Does not exst a connectng edge between the two tasks, these two actvtes can be performed n any order. Just lke the Formula -. Copyrght c 04 SERSC

4 Internatonal Journal of Smart Home Vol.8, No. (04) p r 0, If P R E S E T (T ), T m a x P r ( T ), o th e r w s e T p reset (T ) (-).. Producton Schedulng Model. Schedulng obectves- processng tasks cost-optmal m N M n C s.t. C C, k, ) k k m N n C (-4) k ( C ( t (, k, ) t (, k, )) k, N, m (-5) For the purpose of the smplfed model, assumng the same equpment processng task executon unt cost coeffcent C are settng value, the use of the equpment and the mplementaton of the processng tme s proportonal to the cost.. Constran obectves-under the premse to meet delvery the task average completon tme optmal. Frst make the followng assumptons on the producton schedulng problem n the process:. Each work pece processng route and procedure of processng tme s known, the tme of transfer of the work pece or raw materals or ready s neglgble.. Each machne n a certan moment can only process a work pece n whch a process and the operaton are not free to termnate.. One of the work pece on procedure before unfnshed, next procedure can't start. v. Each work pece only processng on the same machne.v.wth a work pece on a process s not completed before the next process can not be started. ()The producton schedulng model target to tme before the process decouplng pont e s M n t n t (-6) s.t. m N l k e s (-7) k t ( t (, k, ) t (, k, )) T t N T (,, ) e l D T (-8) t (, m ) M ( ) t (, n ) (-) s m n e t (, k ) M ( ) t (, k ) (-0) s k e pr T ) pr ( T ) k ( N 0, k, k, N,, m l k 0 (-) ()The producton schedulng model target to tme after the process decouplng pont M n t n t (-) s.t. m N l k el sl (-) k t ( t (, k, ) t (, k, )) T t N (,, ) T e l D T (-4) 0 Copyrght c 04 SERSC

5 Internatonal Journal of Smart Home Vol.8, No. (04) t (, m ) M ( ) t (, n ) (-5) s m n e t (, k ) M ( ) t (, k ) (-6) s k e p r ( T ) p r ( T ) (-7) k N l k 0 k 0,, k N, N,, m (-8) Therefore, the mathematcal model of the tme constrants of the producton schedulng problem s as follows, m n t m n t m n t (-). Based on Hybrd Genetc Algorthm for Shop Schedulng Soluton of the Model In ths paper, an mproved genetc algorthm to solve the structure shown n Fgure, the genetc algorthm desgn of the schedulng model of producton schedulng s as follows: ready Parent groupsp(t). Cross. Varaton Genetc manpulaton Progeny groupsp(t+) N Ftness evaluaton The obectve functon end Y Stoppng crtera? Matng pool Select Gamblng theory () Chromosome descrpton Fgure. GA Solve Structure Based on the actvtes of manufacturng tasks genetc, each genetc contans the seral number of the processng equpment used by the label of manufacturng tasks and perform the task label. Such a reasonable sort of understandng of all processng actvtes n space by a chromosomal. Is that an effcent soluton of the problem, Processng actvtes n the prorty level from left to rght n turn reduce. Ths type of genetc algorthm for the transformaton of the processng equpment n the same processng actvtes wth a manufacturng task relatvely easy to mplement. Operaton of the thrd codng genes can be modfed. Therefore, the dynamc advantages, the excepton occurs when the processng equpment, n order to ensure the normal executon of the processng actvtes. The only change to the resource nodes n the same schedulng scheme can acheve ts functon. () Intalze groups The classc genetc algorthm to optmze the number of ndvduals and groups at the same tme. the frst to complete the task s selected n the ntal soluton. Heurstc algorthm to Copyrght c 04 SERSC

6 Internatonal Journal of Smart Home Vol.8, No. (04) operate or randomly generated ntal soluton. Constrants to test randomly generated ntal soluton to determne whether t s feasble. f selected, otherwse remove. Ths ongong process n accordance wth the pre-set teraton, untl the group number of the number of predetermned soluton space. At ths stage nclude the followng four steps:. Topologcal sort converson genome sequencng.. Feasble solutons from these sorts of problems.. Calculate the target value of each of ths schedulng program. v. Target value nto ftness value. () The ntersecton of the genetc algorthm The crossover operator of ths artcle s based on processng the sequence of tasks, the order crossover nclude the followng steps: Step : randomly select a substrng; Step : copy the above substrng nto the correspondng poston, and produce an offsprng. Step : Remove the exstng symbols n the substrng, the remanng sequence contans symbols of the prototype offsprng; Step 4: In accordance wth left to rght the symbols nto the rest of the remanng vacances, producng an offsprng. In ths paper, n accordance wth the process sequence of the cross n Fgure. The proposed approach to meet n front of restrctve condtons and easy to mplement. Shown n Fgure based on the allocaton of crossprocessng equpment, a sngle-pont crossover n genetc cross, one of the ntersecton of the two parent bodes have been dentfed, randomly determned, and transform the dstrbuton by the processng actvtes at the pont n front of parent body processng equpment. P P O O Parent ndvdual Cross Offsprng Fgure. Based on the Sequence of Actvtes of the Cross Parent ndvdual Actvty P Resources P Actvty Resources Cross Offsprng Actvty O Resources O Actvty Resources Fgure. Based on the Intersecton of the Allocaton of Resources (4) Varaton of the genetc algorthm desgn Neghborhood search-based mutaton Step : Start, 0.Step : p o p _ sze * p random selecton of a varaton of m chromosomes to pck out the gene to construct the neghborhood, choose a good neghborhood and the neghborhood schedulng assessment chromosomes as the offsprng,step ; otherwse, perform step 4.Step : The Executve, proceed to Step. Step 4: end. Copyrght c 04 SERSC

7 Internatonal Journal of Smart Home Vol.8, No. (04) (5) Select the control parameters Ths paper uses the control parameters nclude populaton sze, crossover probablty and mutaton probablty. Larger the group, the greater sample capacty, easy to mprove the qualty of the genetc algorthm search, but ncreased the computaton of the ndvdual adaptve assessment, reducng the convergence speed, the scale of the general populaton to take The hgher the crossover probablty, the faster the formaton of a new structure n groups, whch s fne gene structure, s lost, the more. Crossover probablty s too small, wll lead the search block, general crossover probablty of 0.6 to.0.mutaton probablty s too large, and the genetc search wll evolve nto a random search. Mutaton probablty s too small, the earler genetc nformaton wll not be restored, the general varaton of probablty of Instances of the Model Valdaton In order to verfy the effectveness of the producton schedulng process decouplng pont postonng method. Use ths method to test for a workshop producton data. The 0 dfferent products n the workshop producton of the same product famly. The 0 processng procedures on 0 dfferent machnes, processng tme matrx T and processng sequence matrx O as follows: T [0 0 ] O [0 0 ] The matrx lne represents number from left to rght order of product, product... product 0.column represents from top to bottom, to machne a processng machne, machne... the machne 0.The devce processng tme fee(refers to, ncludng he sum of equpment deprecaton costs, devce processng costs,equpment mantenance costs, the cost of the operator's workng hours and busness management costs)informaton such as shown n Table, by calculatng the dozens of products processng costs s 77.0 Yuan. Control parameter values shown n Table. Table. The Devce Processng Tme Fee Machne M M M M4 M5 M6 M7 M8 M M0 cost Table. Control Parameter Values Termnate the Parameter Populaton sze Crossover probablty Mutaton probablty evoluton algebra Value Ths paper presents the mproved genetc algorthm to carry out the test on the computer, parameter selecton as follows: the populaton sze s 00, teratve algebra s 00, crossover probablty s 0., mutaton probablty s 0..Processng tme matrx and processng order matrx can make process decouplng pont located n the ffth process. Therefore, the fve Copyrght c 04 SERSC

8 Internatonal Journal of Smart Home Vol.8, No. (04) processes can be mass-produced pre-producton form of sem-fnshed products are stored n the warehouse. So that the schedulng problem transformed nto sort of 0 knds of products n fve dfferent machnes. Usng MATLAB programmng for smulaton, Make span of the problem of the mnmum s 8.Gantt Chart of the optmzaton results as Fgure 4. If do not ntroduce the process decouplng pont, the schedulng problem s sort of the 0 knds of products n 0 dfferent machnes, through the above procedure to make the approprate changes, the mnmum make span s 65,Gantt Chart of the optmzaton results as Fgure 5. M M7 M8 M M Fgure 4. Schedulng Optmzaton Results after Introducng the Process Decouplng Pont M M M M4 M5 M6 M7 M8 M M Fgure 5. Schedulng Optmzaton Results Wthout Introducng the Process Decouplng Pont By comparng the above two test results, through the ntroducton of the process decouplng pont postonng can greatly smplfy the calculaton process of schedulng, to shorten the completon tme of the product, and mass producton way ahead of schedule producton processes n the process before the decouplng pont. Ths part of the processng tme can be greatly shortened based on the orgnal.in order to provde customers low-prce hgh-qualty products. 5. Conclusons Based on the global optmzaton pont of vew, on the bass of resources match and process model n the exstng overall producton schedulng, To acheve the producton schedulng optmzaton of the busness process level, use manufacturng process optmzaton to get the dfferentated parts several optonal path, establsh the mathematcal model of producton schedulng, ths model to the tme and cost as the obectve functon and use the mproved genetc algorthm to solve. The schedulng method smplfes the schedulng process, reduce schedulng dffculty, shortenng the schedulng tme, to acheve effcent use of resources, fnally applcaton examples demonstrate the feasblty of the method. Acknowledgements Ths paper s supported by Unversty Scence Park Entrepreneural Talent of Innovaton Talents of Scence and Technology of Harbn Applcaton Technology Research and Development Proect (0RFDXJ00), 0 Helongang provnce graduate nnovaton fund. 4 Copyrght c 04 SERSC

9 Internatonal Journal of Smart Home Vol.8, No. (04) References [] Y. J. Xe, X. J. Gu and G. N. Q, Mass producton and mass customzaton producton, Group Technology and Producton Modernzaton, vol. 4, (004). [] X. D. Zhou, G. S. Zou and F. J. Xe, Revew of Research on Mass customzaton, Computer Integrated Manufacturng Systems, vol., no., (00). [] W. Zhang and J. Ln, Mult-stage producton model based on mass customzaton, Computer Integrated Manufacturng Systems, vol., no., (007). [4] Y. L, J. Ln, Y. S. Ln and J. Q. Zheng, Mult-agent Based Hybrd Schedulng Strategy Agle Manufacturng System, Journal of System Smulaton, vol., no., (00). [5] G. L, Research on ntellgentzed agle manufacturng system n dstrbuted knowledge base archtecture, Proceedngs of 00 Thrd Internatonal Symposum on Knowledge Acquston and Modelng. 0, KAM00, (00). [6] A. Zhao and F. Qng, Optmzaton of Networked Manufacturng System Based on Bottleneck Resources, Internatonal Semnar on Busness and Informaton Management, vol., (008). [7] Z. Yong and F. Lu, Informaton ntegraton framework of workshop level DNC system for networked manufacturng, Journal of System Smulaton, vol. 4, no. 4, (008). [8] C. Lu, H. B. Sh and J. Yuan, Modelng and performance analyss of manufacturng processes wth stochastc machne falures, Computer Integrated Manufacturng Systems, vol., no. 4, (008). Copyrght c 04 SERSC 5

10 Internatonal Journal of Smart Home Vol.8, No. (04) 6 Copyrght c 04 SERSC