資源描述:
《求解置換流水車間調(diào)度問題的改進遺傳算法》由會員上傳分享,免費在線閱讀,更多相關(guān)內(nèi)容在行業(yè)資料-天天文庫。
1、502009,45(36)ComputerEngineeringandApplications計算機工程與應(yīng)用求解置換流水車間調(diào)度問題的改進遺傳算法涂雪平,施燦濤,李鐵克TUXue-ping,SHICan-tao,LITie-ke北京科技大學(xué)經(jīng)濟管理學(xué)院,北京100083SchoolofEconomyandManagement,UniversityofScienceandTechnologyBeijing,Beijing100083,ChinaE-mail:tuxueping1@126.comTUXue-ping,SHICan-tao,LITie-ke.Impro
2、vedgeneticalgorithmforpermutationflow-shopproblem.ComputerEngineeringandApplications,2009,45(36):50-53.Abstract:Accordingtothefeaturesofpermutationflow-shopproblemandtheprematuredefectofGA,animprovedGAforthisproblemisproposed.IntheprocessofproposedGA,theNEHandthePalmerheuristicsareuse
3、dtoinitializethepopulationtoimprovethequalityoftheinitialsolutions,theMetropolisruleisemployedinchromosomeselectionforavoidingfallintolocaloptimum,andthetabu-searchalgorithmisembeddedtogetawayfromcircuitoussearch.Inordertosavethegeneticinformationofexcellentchromosomes,an“elitemechani
4、sm”ispresentedtoremembergoodgenes,andthebestsolutionswillbesavedineachrun.Theauto-adaptiveterminationruleissuggestedtofurtherimprovesolutionquality.Atlast,theeffectivenessoftheimprovedGAisverifiedbasedonsomebenchmarkproblems.Theresultsshowthatthesolutionqualityandtheconvergencespeedar
5、ebet-terthantheNEHandoriginalGAinitializedbyheuristicalgorithm.Keywords:permutationflow-shopproblem;GeneticAlgorithm(GA);Metropolisrule;tabusearch;elitemechanism摘要:針對置換流水車間調(diào)度問題的基本特征和傳統(tǒng)遺傳算法易早熟的缺陷,設(shè)計了改進遺傳算法來求解此問題。采用NEH和Palmer啟發(fā)式算法進行種群初始化,以提高初始解的質(zhì)量;根據(jù)Metropolis準(zhǔn)則對染色體進行選擇操作,避免陷入局部最優(yōu);在變異過程
6、中引入禁忌算法,避免迂回搜索;在算法迭代過程中引入了保優(yōu)機制,避免丟失優(yōu)秀染色體的基因信息;采用自適應(yīng)終止準(zhǔn)則,以保證解的質(zhì)量?;诘湫虰enchmark算例的仿真實驗結(jié)果表明,算法在求解質(zhì)量和收斂速度方面明顯優(yōu)于NEH算法和種群經(jīng)過初始優(yōu)化的傳統(tǒng)遺傳算法。關(guān)鍵詞:置換流水車間調(diào)度;遺傳算法;Metropolis準(zhǔn)則;禁忌搜索;保優(yōu)機制DOI:10.3778/j.issn.1002-8331.2009.36.016文章編號:1002-8331(2009)36-0050-04文獻標(biāo)識碼:A中圖分類號:TP181引言在現(xiàn)有研究中,文獻[5]將NEH算法和蟻群算法結(jié)合起
7、來進行置換流水車間調(diào)度問題廣泛存在于生產(chǎn)系統(tǒng)和服務(wù)系統(tǒng)求解,使用插入型局部搜索策略;文獻[6]介紹了一種基于分支中,是典型的組合優(yōu)化問題,也是典型的NP難問題[1],工件加界定法的算法,這種算法采用逐對比較的形式來獲得較優(yōu)解;工順序的多樣性和每道工序開工時間的差異性都將大大增加文獻[7]介紹了微粒群優(yōu)化算法來求解此問題,該算法引入交換求解此問題的難度[2]。當(dāng)只有2臺機器時,可以用一個多項式解子和交換序的概念,利用實數(shù)進行編碼;文獻[8]采用了基于回決此問題;但是當(dāng)機器數(shù)達(dá)到3臺時,其調(diào)度問題就屬于強NP溯策略和關(guān)鍵工序鄰域搜索的并行禁忌搜索算法來進行求解,難題[
8、3]。鑒于