资源描述:
《几种现代优化算法的比较研究》由会员上传分享,免费在线阅读,更多相关内容在行业资料-天天文库。
1、几种现代优化算法的比较研究李益芳,黄仙(华北电力大学自动化系,北京,102206)StudiesonSomeModernOptimizationAlgorithmsYIFANG-liHUANGXian(AutomationDepartmentofNorthChinaElectricPowerUniversity,Beijing102206)摘要:现代最优化算法比较常见的有遗传算法、粒子群算法、群体复合形进化算法、鱼群算法、模拟退火算法和蚁群算法。文章主要是对遗传算法、粒子群算法和鱼群算法三个算法的优化性能进行比较。首先介绍了三个算法的基本思想和算法优化过程,以此可以了解三种算法有着自身的特点
2、和优势,促进理解后面不同的优化结果和改进方向。文章中,将三种算法分别对这三个函数用VC编出程序,得出优化结果,再针对结果分析算法。三个典型函数特点各不同,但对算法的优化能力要求都比较高,在不同方面考验了算法的收敛和爬山功能。最后,通过分析三个函数的九个优化结果,提出这三种算法的优点和不足,并列出改进措施。从分析结果可以看出遗传算法要优于另两种算法,并且其改进的余地也是最大的,粒子群算法的优化结果次之,鱼群算法的优化结果相对来说是最差的,但三种算法都可以进行改进以达到更好的优化结果。关键词:优化;遗传算法;粒子群算法;鱼群算法;比较Abstract:Modernoptimizationincl
3、udesgeneticalgorithm,particleswarmalgorithm,multi-complexalgorithm,fishschoolalgorithm,SimulatedAnnealingalgorithmandantcolonyalgorithm.Thepapermainlycomparestheoptimizationabilitiesofgeneticalgorithm,particleswarmalgorithmandfishschoolalgorithm.Firstly,thearticleintroducesthebasicideasandtheoptimi
4、zationprocessesofthethreealgorithms,fromwhichthecharacteristicsandadvantagesofthethreealgorithmswillbefoundout,afterthat,theoptimizationresultsandthewaysofimprovementsbehindwillbeunderstoodeasily.Secondly,thethreealgorithmsprogramwithVCforthethreefunctions,sogettheresultsofoptimizationandanalyzethe
5、m.Thethreerepresentativefunctionshavespecialtiesfromeachother,buttheyhaveonesamepointwhichishavingmuchmoredemandsonthealgorithms,whichteststheabilitiesofastringencyandmountainclimbing.Atlast,throughanalyzingthenineoptimizationresultsofthreefunctions,thepaperexplainstheadvantagesandthedisadvantageso
6、fthethreealgorithms,andputsforwardtheimprovementmeans.Fromtheconclusion,geneticalgorithmismuchbetterthantheothertwooptimizationalgorithms,anditsroomofimprovementisthemostmaximuminthethreealgorithmstoo.Thearticlealsogetstheconclusionthatgeneticalgorithmismuchbetterthantheparticleswarmalgorithmwhichi
7、smuchbetterthanthelastone.Andthethreealgorithmscanbeimprovedtogetbetteroptimizationresults.Keywords:Optimization;Geneticalgorithm;Particleswarmalgorithm;Fishschoolalgorithm;Comparison.1.引言传统的优化算法在优化可以解决一些比较