欢迎来到天天文库
浏览记录
ID:46899153
大小:118.00 KB
页数:8页
时间:2019-11-29
《自适应遗传算法的改进与研究》由会员上传分享,免费在线阅读,更多相关内容在工程资料-天天文库。
1、南京信息工程人学硕士学位论文自适应遗传算法的改进与研究姓名:李欣申请学位级别:硕士专业:系统分析与集成指导教师:罗琦20080501摘要简单遗传算法作为一种启发式搜索算法,寻优理论还不完善。因此,在应用中常出现收敛过慢、稳定性差及早熟现象等问题,而现有的一些口适应遗传算法容易产生局•部最优解。因此,对自适应遗传算法的进一步研究和探讨是很必要的。针对简单遗传算法和现有的一些自适应遗传算法的缺陷,本文分析了种群“早熟‘‘性能指标和计算量,并且判断种群当前适应度最大的那些个体是否重复或相互趋同,由此发展了一种新的种群“早熟
2、,,程度评价指标,结合自适应调整遗传算法的控制参数的思想,提出了一种改进的自适应遗传算法。作者希望本论文提出的新的自适应遗传算法,不仅能加快遗传进化速度,而且能增强遗传算法的全局收敛性能,从而得到满意的全局最优解。木文首先介绍了遗传算法的背景、发展历程和应用,国内外研究现状,说明了研究的背景、日的和预期结果;其次介绍了简单遗传算法和儿种改进白适应遗传算法,分析了现有的一些自适应遗传算法存在的缺陷,为下一步T作奠定基础;最后本文提岀了一种新的判主种样“早熟”程度的方法,对算法的交义概率和变异概率进行改进,设计实现了本文
3、提出的新算法。实验结果说明新算法具有计算稳定性高、收敛速度快等特点,是一种性能良好的改进的自适应遗传算法。关键词:遗传算法,早熟,1自适应AbstractSimP1egeneticalgor•1thmisaheuristicsearchingalgorithm.Itsschemeofsearch•1ngforthebestresu1tisn<0tperfect.Ithassomeshortagessuchass14、p1icat■10n.Ex■1st■1ngadaptivegeneticalgorithmhas10ca1o]ptimizat•10nsolution.Theref0re,it•1sessentia1t'0■1mp1ementafurtherresearchanddiscussionontheauto——adaPt1edgenetica1g(0rithm.Inordertos01vethed•1sadvantagesofsimplegenet1ca1g0r■1thmsandeXis;tingadaptivegenet5、ica1g0r•1thms,webeganwithana1yzingthePerf0rmanceindeXandthecomputat•10n10ad0fthepopu1ation^premature^,andthende!V(elopedanewdegreeevaluat•1ngind•1cat0rofnewP0pu1ation“prematiareIt■1seXpectedthatthepresentpaperwi11pr0p0seanewse1f—adaptedgenet■1ca1g0r■1thmtoreac'6、hthesatisfactoryg1oba11y0Ptima1so1ution,whichCann0t0n1yspeedupthegeneticevolutionspeedbuta1s0strengthenthecorrespond•1ngg10ba1c7、presentresearchhbackground,theedomesticandsituation,theresearchgoalandtheedexpthmandresult.Nextthisthesisanticip1ainedthesimplehereditseveralkindalgoadaptedgeneticlyzedsaptedgnaadundationsarticlenationposofimprovrithm,moreflawinthethm,1ayingwork.Fina1anewkindya8、lgoedautoveraauto—thefo1ythiomeexistingenetica1goriforthelaterproposedanewkindofdetermipulation^precociousndegreemethod.Theinvolvedideaismakingtheimprovementtothealgorithmov
4、p1icat■10n.Ex■1st■1ngadaptivegeneticalgorithmhas10ca1o]ptimizat•10nsolution.Theref0re,it•1sessentia1t'0■1mp1ementafurtherresearchanddiscussionontheauto——adaPt1edgenetica1g(0rithm.Inordertos01vethed•1sadvantagesofsimplegenet1ca1g0r■1thmsandeXis;tingadaptivegenet
5、ica1g0r•1thms,webeganwithana1yzingthePerf0rmanceindeXandthecomputat•10n10ad0fthepopu1ation^premature^,andthende!V(elopedanewdegreeevaluat•1ngind•1cat0rofnewP0pu1ation“prematiareIt■1seXpectedthatthepresentpaperwi11pr0p0seanewse1f—adaptedgenet■1ca1g0r■1thmtoreac'
6、hthesatisfactoryg1oba11y0Ptima1so1ution,whichCann0t0n1yspeedupthegeneticevolutionspeedbuta1s0strengthenthecorrespond•1ngg10ba1c7、presentresearchhbackground,theedomesticandsituation,theresearchgoalandtheedexpthmandresult.Nextthisthesisanticip1ainedthesimplehereditseveralkindalgoadaptedgeneticlyzedsaptedgnaadundationsarticlenationposofimprovrithm,moreflawinthethm,1ayingwork.Fina1anewkindya8、lgoedautoveraauto—thefo1ythiomeexistingenetica1goriforthelaterproposedanewkindofdetermipulation^precociousndegreemethod.Theinvolvedideaismakingtheimprovementtothealgorithmov
7、presentresearchhbackground,theedomesticandsituation,theresearchgoalandtheedexpthmandresult.Nextthisthesisanticip1ainedthesimplehereditseveralkindalgoadaptedgeneticlyzedsaptedgnaadundationsarticlenationposofimprovrithm,moreflawinthethm,1ayingwork.Fina1anewkindya
8、lgoedautoveraauto—thefo1ythiomeexistingenetica1goriforthelaterproposedanewkindofdetermipulation^precociousndegreemethod.Theinvolvedideaismakingtheimprovementtothealgorithmov
此文档下载收益归作者所有