欢迎来到天天文库
浏览记录
ID:12584354
大小:555.15 KB
页数:57页
时间:2018-07-17
《硕士论文-多宇宙并行量子多目标进化算法及其应用研究》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、湖南大学硕士学位论文多宇宙并行量子多目标进化算法及其应用研究姓名:李絮申请学位级别:硕士专业:计算机科学与技术指导教师:李智勇20090511多宇宙并行量子多目标进化算法及j£戍用研究AbstractQuantum—inspiredEvolutionaryAlgorithm(QEA)isanovelkindofprobabilityseachalgorithmbycombiningquantumtheorywithevolutionaryalgorithm.Basedontheprinciples
2、ofQuantumComputing,QEAusesQ・bitstringastheprobabilisticrepresentationofsolutionsandQ-gatesasvariationoperatorstodriveevolutionarysearching.ComparedwithtraditionalEAs,QEAcanbalancebetweenexplorationandexploitationbetter.Additionallyitischaracterizedbys
3、mallpopulationsize,rapidconvergenceandstrongglobalsearchcapability.Aswhattheresearchresultsshow,QEAhasbetterperformancesthanconventionalEAsonmanyproblems,butittendstoranintolocaloptimainsolvingsomecomplexoptimizationproblems,andfundamentallytheproblem
4、ofQEA’Sprematureconvergencehasstillnotbeensolved.Meanwhile,thegeneralQEAstillevolvesbasedonthesinglegroupatpresentanddoesn’tmakefulluseofthecharacteristicisofquantum’Smulti-universe.ToimproveQEAwell,theideathatmanyuniverseshavebeenusedandcooperatetoge
5、therinalgorithmmaybethepossibleresolvent.Consequently,anovelQuantum・inspiredMulti—objectiveEvolutionaryAlgorithmisproposedinspiredbyquantumcomputing,whichisnamedmulti--universeparallelquantum--inspiredmultiobjectiveevolutionaryalgorithm(MPQMEA).Inorde
6、rtogetmoreefficientconvergence,thispaperdividesallindividualsintosomeindependentsub—colonies,calleduniverses,accordingtotheirdefinitetopologicalstructure.Theuniformassignmentprincipleoftargetindividualsanddynamicadjustingrotationanglemechanismareappli
7、edtoupdateeachuniversalindividual.Informationamongtheuniversesisexchangedbyadoptingthebestemigration.Tomakegooduseofglobalinformation,thebestreservationschemeisdesignedfortheimprovementofsearchefficiency.Intheory,weprovetheconvergenceofthealgorithmbas
8、edonthepartialsettheoryandprobabilitytheory.Meanwhile,Multi—objective0-1knapsackproblemisacomplexNPhardproblem.Itcanbeusedtotestwhetherthemulti—objectiveevolutionaryalgorithmisgoodornot.Inthispaper.thealgorithmweproposedisappliedtonineO-1knaps
此文档下载收益归作者所有