改進遺傳算法及其應用.pdf

1、華北電力大學（保定）碩士學位論文改進遺傳算法及其應用姓名：申紅蓮申請學位級別：碩士專業(yè)：應用數(shù)學指導教師：張國立20071220華北電力大學碩士學位論文摘要摘要基于距離測度的實數(shù)編碼遺傳模擬退火算法是2004年蔡良偉提出的一種具有全局搜索能力的新算法。本文對該算法進行了改進，定義了基于距離測度和適應度的自適應變異概率，采用改進的交叉策略和模擬退火操作，并在群體趨于一致時保留最優(yōu)個體，重新產(chǎn)生其它新個體。對帶邊界約束函數(shù)優(yōu)化問題進行仿真計

2、算，結(jié)果表明改進后算法收斂概率較高，收斂速度較快。針對TSP問題，本文定義了基于距離測度和適應度的的自適應交叉概率，采用改進的交換啟發(fā)交叉算子，產(chǎn)生不差于父代的個體，根據(jù)最優(yōu)和次優(yōu)個體的差異，采用2變換法產(chǎn)生新個體或者進行模擬退火操作。仿真實驗結(jié)果表明新算法收斂概率較高，得到最優(yōu)解的平均代數(shù)要遠遠小于其他所比較的算法。關(guān)鍵詞：遺傳算法，模擬退火算法，距離測度，適應度，函數(shù)優(yōu)化ABSTRACTReal—codedgeneticanneal

3、ingalgorithmbasedondistancemeasurementisanewalgorithmfor哲obal—searchingwhichproposedbyCaiLiangWeiin2004Thealgorithmhasbeenimprovedinthispaperitdefinedthemutationprobabilityaccordingtodistancemeasurementandfitness，adopted

4、improvedcrossoverandsimulatedannealingoperationInaddition，whenpopulationtendedtouniform，itreservedthebestindividualandreproducedotherindividualsApplyingittothefunctionoptimizationproblemwithboundaryconstraints，thesimulat

5、edresultsshowthattheimprovedalgorithmismoreeffectiveinrealizingthehighconvergenceprobabilityandrapidevolvingspeedInordertosolveTSPproblemitdefinedthecrossingprobabilitybasedondistancemeasurementandfitness，adoptedimproved

6、exchangeheuristiccrossovcroperatorwhichthenewindividualisnoworsethantheoldone。Accordingtothedifferencebetweenthebestindividualandthebetterindividual，itadoptedthemethodthatusedoubleexchangeproducenewindividualsordosimulat

7、edannealingopermionThesimulatedresultsshowthatthenewalgorithmhashighconvergenceprobabilityandthegenerationnumbersisfarlessthanothercomparativealgorithmsShenHonglian(DepartmentofMathematicsandPhysics)DirectedbyprofZhangGu