Evolutionary fragmentary algorithm for permutation flow shop problem

Рассматривается NP - трудная в сильном смысле задача Джонсона. Установлена фрагментарная структура задачи. Предложен эволюционно-фрагментарный подход для поиска оптимального решения. Ппроведено тестирование эволюционно-фрагментарного алгоритма на наборе тестовых задач из библиотеки ORLib. Ключевые с...

Full description

Saved in:
Bibliographic Details
Published in:Таврический вестник информатики и математики
Date:2009
Main Authors: Bondarenko, O.S., Kozin, I.V.
Format: Article
Language:English
Published: Кримський науковий центр НАН України і МОН України 2009
Online Access:https://nasplib.isofts.kiev.ua/handle/123456789/18229
Tags: Add Tag
No Tags, Be the first to tag this record!
Journal Title:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Cite this:Evolutionary fragmentary algorithm for permutation flow shop problem / O.S. Bondarenko, I.V. Kozin // Таврический вестник информатики и математики. — 2009. — № 2. — С. 47-51. — Бібліогр.: 18 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
_version_ 1859660254504026112
author Bondarenko, O.S.,
Kozin, I.V.
author_facet Bondarenko, O.S.,
Kozin, I.V.
citation_txt Evolutionary fragmentary algorithm for permutation flow shop problem / O.S. Bondarenko, I.V. Kozin // Таврический вестник информатики и математики. — 2009. — № 2. — С. 47-51. — Бібліогр.: 18 назв. — англ.
collection DSpace DC
container_title Таврический вестник информатики и математики
description Рассматривается NP - трудная в сильном смысле задача Джонсона. Установлена фрагментарная структура задачи. Предложен эволюционно-фрагментарный подход для поиска оптимального решения. Ппроведено тестирование эволюционно-фрагментарного алгоритма на наборе тестовых задач из библиотеки ORLib. Ключевые слова: задача Джонсона, фрагментарная структура, еволюционно-фрагментарный подход, NP - трудность. Розглядається NP - важка в сильному сенсі задача Джонсона. Встановлено фрагментарну структуру задачі. Запропоновано еволюційно-фрагментарний підхід для пошуку оптимального розв'язку. Проведено тестування еволюційно-фрагментарного алгоритму на наборі тестових задач з бібліотеки ORLib. Ключові слова: задача Джонсона, фрагментарна структура, еволюційно-фрагментарний підхід, NP - важкість. The article tackles the strongly NP - hard permutation flow shop problem. The flagmentary structure of the problem is pointed out. The evolutionary fragmentary approach for optimal solution is proposed. The testing of evolutionary fragmentary algorithm on instances' set from the ORLib library is conducted.
first_indexed 2025-11-30T09:41:29Z
format Article
fulltext UDC 519.8EVOLUTIONARY FRAGMENTARY ALGORITHM FORPERMUTATION FLOW SHOP PROBLEM Bondarenko O.S., Kozin I.V.Zaporizhzhya National UniversityThe Department of E onomi sThe Chair of E onomi Cyberneti sZhukovsky str., 66, Zaporizhzhya, 69600, Ukrainee-mail: buenasdiaz�gmail. omAbstra t. The arti le ta kles the strongly N P-hard permutation �ow shop problem. Thefragmentary stru ture of the problem is pointed out. The evolutionary fragmentary approa h foroptimal solution is proposed. The testing of evolutionary fragmentary algorithm on instan es' set fromthe ORLib [1℄ library is ondu ted. Introdu tionIn 1954 Selmer Martin Johnson proposed an algorithm for �nding optimal solution toa problem [2℄, whi h has been known sin e as �ow shop problem.Suppose that n jobs Ji(i = 1; : : : ; n) are to be exe uted on mma hines Mj(j = 1; : : : ; m). Ea h job onsists of m operations Oi1; : : : ; Oim. Ea hoperation Oij is asso iated with the pro essing time pij. Every job is pro essed on allma hines in the order of the ma hine indexing. It means that the exe ution of theoperation Oij+1 is not allowed until the exe ution of the operation Oij is �nished. By thes hedule we mean the mapping from the set of the job numbers (i; j) to the ompletiontimes set Cij. In the given problem, we are supposed to �nd su h a s hedule in whi h omletion time of the last job on the last ma hine is minimized. Preemption, pro essingmore than one job on one ma hine at a time, and exe ution one job on more than onema hine at a time are not allowed.A ording to the notation [3℄ the problem under onsideration is denoted F jjCmax.The problem has (n!)m feasible solutions, hen e, in the literature the easier ase is studied,whi h admits n! s hedules with the same order of pro essing on all ma hines (permutations hedules) and is denoted F jprmujCmax. The latter ase is onsidered in the following.The exa t algorithm proposed by Johnson solves the problem to optimality for m = 2 ase and for spe ial ase with m = 3. All the e�orts to �nd exa t polynomial algorithmform > 2 were unsu essfull. With the advent of the �rst results in the early 70-s [4, 5, 6℄ in omputational omplexity theory that fa t had obtained an explanation. N P-hardnessin the strong sense was proved in [7℄. Sin e then resear hers' e�orts have been fo used onapproximation and heuristi algorithms design.Some time after that it has be ome lear that some problems are hard not only forsolving to optimality but are hard for approximating. It was proved [8℄ for onsideredproblem there is no (1 + �)-approximation algorithm for 1 + � < 5=4. Furthermore, so farthe problem whether there exist polynomial approximation s heme for FmjjCmax wherem 48 Bondarenko O.S., Kozin I.V.is �xed is open. Hen e, the design of algorithms with no theoreti al bounds and worst- ase ratios, whi h solve omplex pra ti al problems, is now urgent. As su h we onsidersto hasti lo al sear h [9℄ algorithms: evolutionary algorithms, taboo sear h, simulatedannealing, ant olony systems, variable neighbourhood sear h.The survey of lo al sear h algorithms for �ow shop problem an be found in [10℄. The omparative analysis of the most e� ient lo al sear h methods from the literature, alledalso metaheuristi s, was ondu ted in [11℄.The goal of the paper is the inquiry of the new solution approa h to ombinatorialproblems based on the ombination of the evolutionary and the fragmentary approa hes.1. The Fragmentary Stru tureDe�nition 1. Fragmentary stru ture [12℄ is tuple (X ;Y ;R), where X � a �nite set ofthe fragments, Y � a family of subsets of X , R � a ombining rule, i.e. the de ision rulefor re ognizing whether a union of subsets of X is a feasible set of Y .F jjCmax an be onsidered as the fragmentary stru ture. By fragmments we will haveany tuples of jobs, respe tively by the elementary fragments we will mean jobs. The ruleof ombination � no job repetition in the ombined tuples. Maximal by in lusion fragmentwill be a feasible solution the problem under onsideration. So the feasible solution anbe onsidered as ertain �xed permutation of jobs.In the papers [12, 13℄ it has been showed that the sear h of feasible solutions an onsidered in the fragmentary stru ture framework. For the ostru tion of feasiblesolutions the fragmentary algorithm F is applied. In the general ase, the input of Falgorithm are a tuple I = (i1; : : : ; in) and the empty solution set S . The input is lookedthrough by F in the ma hine indexing order. On every step in the tuple I the �rstelement is looked for whi h the ombining rule is ful�lled. The found elementary fragmentis added into S . The pro edure is repeated until the maximal by in lusion fragment isbuilt. 2. The Evolutionary Fragmentary AlgorithmThe fragmentary algorithm allows to obtain only some feasible solutions for theoptimization problem. For �nding optimal solution of F jprmujCmax the ombinationevolutionary and fragmentary approa hes is proposed. A few analogous approa hesgrounded on the evolutionary me hanism and performan e evaluation of su h approa hesare onsidered in [14, 15℄.The des ription of evolutionary fragmentary algorithm (EVF algorithm), exploredin the work, is the following. The feasible solutions of the problem are presented by jobnumber permutations, whi h form the set of all permutations I1; : : : ; In!. Ea h permutationwill be treated as hromosome. The sear h an be des ribed as the following.Stage 1. Initialization.The initial population, onsisting of N hromosomes, is generated at random.Stage 2. Sele tion.Sele t K pairs of parents from the urrent population for reprodu tion. The sele tionis random without repla ement. ¾Òàâðè÷åñêèé âåñòíèê èí�îðìàòèêè è ìàòåìàòèêè¿, �2' 2009 EVF algorithm for permutation �ow shop problem 49Stage 3. Re ombination.The re ombination in the EVF algorithm is implemented by the following n-steppro edure. All the pairs of hromosome-parents' I-tuples are being looked through. Onevery step the minimal element from the �rst two elements of I-tuples' pair is hosen andis inserted into tuple-o�spring. The inserted element is deleted from both tuples-parents.For instan e, in the ase with m = 5 hromosome-parents [12543℄ and [53142℄ yield the hromosome-o�spring [12534℄.Stage 4. Mutaion.The mutation is implemented by transposition of jobs' indexes from two randomlysele ted positions in hromosome.Stage 5. Repla ement.The hromosomes-o�springs are added in the urrent population. The shrinking of the urrent population to prede�ened size is provided by sequential deletion of hromosomeswith maximal riterion's values, where the riterion's value is omputed by the followingre ursive formula: Ci;j = maxfCi�1;j; Ci;j�1g+ pi;jStage 6. Termination.If prede�ened termination riterion is a hieved, then stop the algorithm, otherwisego to the se ond stage. The termination riterion, used by authors, is the number ofgenerations in evolution, i.e. the number of the sear h stages repetitions.3. The Test ResultsThe ne essity of test ondu tion is explained by the di� ulties and, sometimes, byimposibility of theoreti al estimation of sto hasti lo al sear h algorithms performan ederivaton in general, and of evolutionary algorithms, in parti ular.For the EVF algorithm performan e veri� ation the test was being ondu ted onwell-known in the Taillard's [16℄ problem set. The problem set omprises twelve groups onsisting of ten instan es ea h with the numbers of jobs from 20 to 500 and the numberof ma hines from �ve to 20. It is the most used in the literature for F jprmujCmax testing.Thus, there appears an opportunity for the omparison of the proposed algorithm withalready existing and tested ones on the given set.EVF algorithm was tested against random sear h (RS) algorithm as well as NEHalgorithm [17℄, whi h is urently onsidered the best performing deterministi heuristi .All the three algorithms were implemented in Visual Basi for Appli ations andrun on omputer with Celeron pro esor 1800 Mhz and 256 MB of the main memory.The termination riterion was hosen to be the number of generations. a hieved by thealgorithm during its run.The algorithms' performan e was measured by relative devian e (RD) from optimum(OPT) or the least known upper bound [18℄ of the value (A) found by given algorithm:RD = A�OPTOPT � 100:In Table 1 ea h group of problems is denoted as n�m, where n is the number of jobs,m is the number of ma hines. In it RD for ea h instan e is averaged on the group size. ForEVF algorithm in the table in the parentheses the number of generations during whi h it¾Òàâðiéñüêèé âiñíèê ií�îðìàòèêè òà ìàòåìàòèêè¿, �2' 2009 50 Bondarenko O.S., Kozin I.V.Group of problems RS NEH EVF20� 5 6.95 2.46 1.25 (300)20� 10 10.29 4.88 2.86 (300)20� 20 8.52 3.72 3.28 (300)50� 5 5.18 0.81 0.92 (300)50� 10 14.39 5.23 4.51 (300)50� 20 16.71 6.21 5.85 (300)100� 5 4.14 0.67 0.73 (500)100� 10 10.74 2.45 2.4 (500)100� 20 16.85 5.72 6.08 (500)200� 10 8.74 1.8 1.52 (1000)200� 20 15.66 4.66 4.95 (1000)500� 20 11.79 2.41 3.74 (1000)Average 10.83 3.42 3.17Òàáëèöà 1. The Test Results for F jprmujCmaxwas run is pointed out. From Table 1 it an be on luded that in the average evolutionaryfragmentary algorithm has showed the higher results than other two tested algorithms.Con lusionThe basi result of the paper is the new approa h to s heduling problems solution. Thefragmentary stru ture has been showed whi h allows the appli ation of EVF algorithm.The testing of the algorithm on the standard problem set is ful�lled.From the test results, represented in Table 1, it appears promising to apply theproposed aproa h to s heduling problems solution.Ñïèñîê ëèòåðàòóðû1. Beasley J. E. ORLib - Operations Resear h Library.http://people.brunel.a .uk/�mastjjb/jeb/orlib/�les/�owshop2.txt, 2008.2. Johnson S. M. Optimal Two and Three-Stage Produ tion S hedules with Setup Times In luded //Naval Resear h Logisti s Quarterly. � 1954. � Vol. 1, No. 1. � P. 61�68.3. Graham R. L. Optimization and approximation in deterministi sequen ing and s heduling:A survey / R. L. Graham, E. L. Lawler, J. K. Lenstra, A. H. G. Rinnooy Kan // Dis reteOptimization II / P.L. Hammer, E.L. Johnson and B.H. Korte (eds.). � Amsterdam : North-Holland,1979. � (Annals of Dis rete Mathemati s ; Vol. 5.). � P. 287�326.4. Cook S. A. The omplexity of theorem-proving pro edures / Stephen A. Cook // STOC '71:Pro eedings of the third annual ACM symposium on Theory of omputing. � New York : ACM,1971. � P. 151�158.5. Karp R. Redu ibility among Combinatorial Problems // Complexity of Computer Computations /R. E. Miller and J. W. That her (eds.). � New York : Plenum Press, 1972. � P. 85�104.6. Ëåâèí Ë. À. Óíèâåðñàëüíûå çàäà÷è ïåðåáîðà / Ë. À. Ëåâèí // Ïðîáëåìû ïåðåäà÷è èí�îðìà-öèè. � 1973. � Ò. 9, � 3. � Ñ. 115-116. ¾Òàâðè÷åñêèé âåñòíèê èí�îðìàòèêè è ìàòåìàòèêè¿, �2' 2009 EVF algorithm for permutation �ow shop problem 517. Garey M. R. The Complexity of Flow Shop and Job Shop S heduling / M. R. Garey, D. S. Johnson,R. Sethi // Mathemati s of Operations Resear h. � 1976. � Vol. 1, No. 2. � P. 117�129.8. Williamson D. P. Short shop s hedules / D. P. Williamson, L. A. Hall, J. A. Hoogeveen, C. A. J.Hurkens, J. K. Lenstra, S. V. Sevastianov, D. B. Shmoys // Operations Resear h. � 1997. � Vol. 45,No. 2. � P. 288�294.9. Hoos H. H. Sto hasti Lo al Sear h : Foundations and Appli ations / Hoos H. H., Stuetzle T. � SanFran is o : Morgan Kaufmann Publishers, 2004. � 660 p. � ISBN 1-55860-872-9.10. Ruiz R. A omprehensive review and evaluation of permutation �owshop heuristi s / R. Ruiz, C.Maroto // European Journal of Operational Resear h. � 2005. � Vol. 165. � P. 479�494.11. Ruiz R. A simple and e�e tive iterated greedy algorithm for the permutation �owshop s hedulingproblem / R. Ruiz, T. Stuetzle// European Journal of Operational Resear h. � 2007. � Vol. 177. �P. 2033�2049.12. Êîçèí È. Â. Ôðàãìåíòàðíûå àëãîðèòìû â ñèñòåìàõ ïîääåðæêè ïðèíÿòèÿ ðåøåíèé // Ïèòàííÿïðèêëàäíî¨ ìàòåìàòèêè i ìàòåìàòè÷íîãî ìîäåëþâàííÿ, çáiðíèê íàóêîâèõ ïðàöü. - Äíiïðîïåòðî-â üê. � 2006. � Ñ. 131�137.13. Êîçèí È. Â. Ôðàãìåíòàðíûé àëãîðèòì äëÿ çàäà÷è ñèììåòðè÷íîãî ðàçìåùåíèÿ // �àäèîýëåê-òðîíèêà, èí�îðìàòèêà, óïðàâëåíèå. � 2005. � � 1. � Ñ. 76�83.14. Ruiz R. Two new robust geneti algorithms for the �owshop s heduling problem / R. Ruiz, C. Maroto,J. Al araz // OMEGA, the International Journal of Management S ien e. � 2006. � Vol. 34. � P.461�476.15. Nagano M. S. A Constru tive Geneti Algorithm for permutation �owshop s heduling / M. S. Nagano,R. Ruiz, L. A. N. Lorena // Computers & Industrial Engineering. � 2008. � Vol. 55. � P. 195�207.16. Taillard E. Ben hmarks for basi s heduling problems // European Journal of Operational Resear h. �1993. � Vol. 64, No. 2. � P. 278-285.17. Nawaz M. A heuristi algorithm for the m-ma hine, n-job �ow-shop sequen ing problem / M. Nawaz,E. E. Ens ore Jr., I. Ham // OMEGA, The International Journal of Management S ien e. � 1983. �Vol. 11, No. 1. � P. 91�95.18. Taillard E. Summary of best known lower and upper bounds of Taillard's instan es.http://ina2.eivd. h/Collaborateurs/etd/problemes.dir/ordonnan ement.dir/�owshop.dir/best_lb_-up.txt, 2008. Ñòàòüÿ ïîñòóïèëà â ðåäàêöèþ 15.08.2009 ¾Òàâðiéñüêèé âiñíèê ií�îðìàòèêè òà ìàòåìàòèêè¿, �2' 2009
id nasplib_isofts_kiev_ua-123456789-18229
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
issn 1729-3901
language English
last_indexed 2025-11-30T09:41:29Z
publishDate 2009
publisher Кримський науковий центр НАН України і МОН України
record_format dspace
spelling Bondarenko, O.S.,
Kozin, I.V.
2011-03-18T23:43:13Z
2011-03-18T23:43:13Z
2009
Evolutionary fragmentary algorithm for permutation flow shop problem / O.S. Bondarenko, I.V. Kozin // Таврический вестник информатики и математики. — 2009. — № 2. — С. 47-51. — Бібліогр.: 18 назв. — англ.
1729-3901
https://nasplib.isofts.kiev.ua/handle/123456789/18229
519.8
Рассматривается NP - трудная в сильном смысле задача Джонсона. Установлена фрагментарная структура задачи. Предложен эволюционно-фрагментарный подход для поиска оптимального решения. Ппроведено тестирование эволюционно-фрагментарного алгоритма на наборе тестовых задач из библиотеки ORLib. Ключевые слова: задача Джонсона, фрагментарная структура, еволюционно-фрагментарный подход, NP - трудность.
Розглядається NP - важка в сильному сенсі задача Джонсона. Встановлено фрагментарну структуру задачі. Запропоновано еволюційно-фрагментарний підхід для пошуку оптимального розв'язку. Проведено тестування еволюційно-фрагментарного алгоритму на наборі тестових задач з бібліотеки ORLib. Ключові слова: задача Джонсона, фрагментарна структура, еволюційно-фрагментарний підхід, NP - важкість.
The article tackles the strongly NP - hard permutation flow shop problem. The flagmentary structure of the problem is pointed out. The evolutionary fragmentary approach for optimal solution is proposed. The testing of evolutionary fragmentary algorithm on instances' set from the ORLib library is conducted.
en
Кримський науковий центр НАН України і МОН України
Таврический вестник информатики и математики
Evolutionary fragmentary algorithm for permutation flow shop problem
Article
published earlier
spellingShingle Evolutionary fragmentary algorithm for permutation flow shop problem
Bondarenko, O.S.,
Kozin, I.V.
title Evolutionary fragmentary algorithm for permutation flow shop problem
title_full Evolutionary fragmentary algorithm for permutation flow shop problem
title_fullStr Evolutionary fragmentary algorithm for permutation flow shop problem
title_full_unstemmed Evolutionary fragmentary algorithm for permutation flow shop problem
title_short Evolutionary fragmentary algorithm for permutation flow shop problem
title_sort evolutionary fragmentary algorithm for permutation flow shop problem
url https://nasplib.isofts.kiev.ua/handle/123456789/18229
work_keys_str_mv AT bondarenkoos evolutionaryfragmentaryalgorithmforpermutationflowshopproblem
AT koziniv evolutionaryfragmentaryalgorithmforpermutationflowshopproblem