資源描述:
《基于改進(jìn)遺傳算法的調(diào)度問題研究》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、浙江大學(xué)碩士學(xué)位論文基于改進(jìn)遺傳算法的調(diào)度問題研究姓名:黃少鋒申請(qǐng)學(xué)位級(jí)別:碩士專業(yè):模式識(shí)別與智能系統(tǒng)指導(dǎo)教師:王寧20060501摘要遺傳算法作為一種新型優(yōu)化算法,由于具有簡單、易操作、并行信息處理等特點(diǎn),已經(jīng)在許多領(lǐng)域的優(yōu)化問題求解方面取得了成功的應(yīng)用。但是遺傳算法在理論上還不夠完善,例如存在容易產(chǎn)生早熟現(xiàn)象以及局部尋優(yōu)能力較差等問題,影響了其進(jìn)一步的應(yīng)用。本文針對(duì)常規(guī)遺傳算法的不足,提出利用分布種群遺傳算法求解Job-shop和Flow-shop問題,提出了一種基于啟發(fā)式的遺傳算法來求解單機(jī)加工時(shí)間可控問題,并進(jìn)行數(shù)字計(jì)算研究。本文主要內(nèi)容包括以下幾個(gè)方面:
2、1.針對(duì)常規(guī)遺傳算法的不足,給出了分布種群遺傳算法的抽象形式,并分析了其收斂性,將該算法運(yùn)用子Flow-shop和Job-shop問題計(jì)算,結(jié)果表明了該算法的有效性。2.針對(duì)單機(jī)加工時(shí)間可控調(diào)度問題,使用基于啟發(fā)式遺傳算法求解單機(jī)加工時(shí)間可控問題和單機(jī)加工時(shí)間離散問題。對(duì)具有多變量、非線性和不確定的此類問題的計(jì)算結(jié)果表明優(yōu)化整定后的算法在性能上有了明顯的提高·3.使用基于分布種群的遺傳算法對(duì)Job-shop問題的加工時(shí)間可控問題進(jìn)行了研究,完成了對(duì)加工可控Job-shop問題的智能算法求解研究,計(jì)算結(jié)果表明了所設(shè)計(jì)的算法具有優(yōu)良的品質(zhì)。關(guān)鍵詞:遺傳算法,分布種群,N
3、P-hard,單機(jī)加工時(shí)間可控,Job-shop,F(xiàn)low-shopⅡAbstractAsanewoptimizationmethod,GAwaswidelyusedintheopfmaizationsofmanyfieldsowingtothefeatttresofsimpfic毋,easilyhandingandparallelprocessing.HoweverGAtheoryisnotperfect,suchasthereexisttheproblemsofeasilycreatingearlinessandbadabilityinlocaloptimal,
4、etc.Enlightenedbydism'butienofcreatureliving缸nature,themathematicmodeloftheDistr/bufionPopu/at/onbasedG∞eticAlgorithm(DPGA)isproposedinthispaper,anditsconvergenceanalysisa∞alsogiven.DPGAisappliedtooptimizetheJob-shopproblem(JSP)andHow-shopproblems(FSP),andthemodifiedGAbasedOllhem'istic
5、mlesispresentedforthesinglemachineschedulingproblemswithcontrollableprocessingtimes.ThethreeproblemsaretypicallyNl'-hard,whichlneallsthatitisimpossibletofindtheglobaloptimuminpolynomialcomplexity.Goodalgorithmsforthisproblemcanpromoteproductivityofenterprises.Thesimulationtests搬madeand
6、theresultsdemonstratetheefficiencyoftheabovemethedr-1nbemaincontentofthisthesisincludesthefollowing:。1.EnlightenedbydisU-ibutionofcreaturelivinginnaturalecologyenvironment,themathematicmodeloftheDistributionPopulationbasedGeneticAlgorithm,andtheconvergenceanalysisofDPGAaregiven.2.Theal
7、gorithmsbasedonDPGAisdesignedfortheJSPandFSP.ThesimulationtestsforsomebenchmarksshowtheefficiencyoftheDPGA.3.ThemodifiedGAispresentedtogettheoptimalresultsoftheNP-hardsinglemachineproblemwithcontrolhbleprocessingtimesandtheNP-hardsinglemachineproblemwilhdiser礎(chǔ)controllableprocessingti