一种多尺度协同变异的粒子群优化算法

《一种多尺度协同变异的粒子群优化算法》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、软件学报ISSN1000-9825,CODENRUXUEWE-mail:jos@iscas.ac.cnJournalofSoftware,2012,23(7):18051815[doi:10.3724/SP.J.1001.2012.04128]http://www.jos.org.cn©中国科学院软件研究所版权所有.Tel/Fax:+86-10-62562563一种多尺度协同变异的粒子群优化算法1+211陶新民,刘福荣,刘玉,童智靖1(哈尔滨工程大学信息与通信工程学院,黑龙江哈尔滨150001)2(黑龙江省电力有限公司,黑龙江哈尔滨15

2、0090)Multi-ScaleCooperativeMutationParticleSwarmOptimizationAlgorithm1+211TAOXin-Min,LIUFu-Rong,LIUYu,TONGZhi-Jing1(CollegeofInformationandCommunicationEngineering,HarbinEngineeringUniversity,Harbin150001,China)2(HeilongjiangElectricPowerCompanyLimited,Harbin150090,China)+

3、Correspondingauthor:E-mail:taoxinmin@hrbeu.edu.cn,http://www.hrbeu.edu.cnTaoXM,LiuFR,LiuY,TongZJ.Multi-Scalecooperativemutationparticleswarmoptimizationalgorithm.JournalofSoftware,2012,23(7):18051815(inChinese).http://www.jos.org.cn/1000-9825/4128.htmAbstract:Todealwithth

4、eproblemofprematureconvergenceandlowprecisionofthetraditionalparticleswarmoptimizationalgorithm,aparticleswarmoptimization(PSO)algorithmbasedonmulti-scalecooperativemutation,isproposed,whichisguaranteedtoconvergetotheglobaloptimalsolutionwithprobabilityone.Thespecialmulti-

5、scaleGaussianmutationoperatorsareintroducedtomaketheparticlesexplorethesearchspacemoreefficiently.Thelarge-scalemutationoperatorscanbeutilizedtoquicklylocatetheglobaloptimalspaceduringearlyevolution.Thesmall-scalemutationoperators,whicharegraduallyreducedaccordingtothechan

6、geofthefitnessvaluecanimplementtheaccuracyofthesolutionatthelateevolution.Theproposedmethodisappliedtosixtypicalcomplexfunctionoptimizationproblems,andthecomparisonoftheperformanceoftheproposedmethodwithotherPSOalgorithmsisexperimented.Theresultsshowthattheproposedmethodca

7、neffectivelyspeeduptheconvergenceandimprovethestability.Keywords:particleswarmoptimization;prematureconvergence;multi-scale;cooperativemutation;ftness摘要:为了改善粒子群算法易早熟收敛、精度低等缺点,提出一种多尺度协同变异的粒子群优化算法,并证明了该算法以概率1收敛到全局最优解.算法采用多尺度高斯变异机制实现局部解逃逸.在算法初期阶段,利用大尺度变异及均匀变异算子实现全局最优解空间的快速定位;

8、随着适应值的提升,变异尺度随之降低;最终在算法后期阶段,利用小尺度变异算子完成局部精确解空间的搜索.将算法应用6个典型复杂函数优化问题,并同其他带变异操作的PSO算法比较,结果表