求解全局优化问题的正交协方差矩阵自适应进化策略算法.doc

求解全局优化问题的正交协方差矩阵自适应进化策略算法16————————————————————————————————作者：————————————————————————————————日期：个人收集整理勿做商业用途个人收集整理勿做商业用途个人收集整理勿做商业用途求解全局优化问题的正交协方差矩阵自适应进化策略算法摘要:针对协方差矩阵自适应进化策略（cmaes)求解高维多模态函数时存在早熟收敛及求解精度不高的缺陷，提出一种融合量化正交设计（od/q）思想的正交cmaes算法.首先利用小种群的cmaes进行快速搜索,当算法陷入局部极值时,依据当前最好解的位置动态选取基向量,接着利用od/q构造的试验向量探测包括极值附近区域在内的整个搜索空间,从而引导算法跳出局部最优。通过对6个高维多模态标准函数进行测试并与其他算法相比较,其结果表明,正交cmaes算法具有更好的搜索精度、收敛速度和全局寻优性能.关键词：协方差矩阵自适应进化策略;正交设计；高维多模态；进化策略;函数优化hybridorthogonalcmaesforsolvingglobaloptimizationproblemshuangya.fei1,2＊,liangxi。ming1，chenyi.xiong11。schoolofinformationscienceandengineering，centralsouthuniversity,changshahunan410083,china;2。schoolofelectricandinformationengineering，changshauniversityofscienceandtechnology，changshahunan410114,chinaabstract：inordertoovercometheshortcomingsofcovariancematrixadaptationevolutionstrategy（cmaes),suchasprematureconvergenceandlowprecision,whenitisusedinhigh-dimensionalmultimodaloptimization,anhybridalgorithmcombinedcmaeswithorthogonaldesignwithquantization（od/q)wasproposedinthisstudy。firstly,thesmallpopulationcmaeswasusedtorealizeafastsearching.whenorthogonalcmaesalgorithmtrappedinlocalextremum,basevectorsforod/qwereselecteddynamicallybasedonthepositionofcurrentbestsolution.thentheentiresolutionspace，includingthefieldaroundextremevalue,wasexploredbytrialvectorsgeneratedbyod/q.theproposedalgorithmwasguidedbythisprocessjumpingoutofthelocaloptimum。thenewapproachistestedonsixhigh—dimensionalmultimodalbenchmarkfunctions.comparedwithotheralgorithms,thenewalgorithmhasbettersearchprecision,convergentspeedandcapacityofglobalsearch.inordertoovercometheshortcomingsofcovariancematrixadaptationevolutionstrategy（cmaes），suchasprematureconvergenceandlowprecision,whenitisusedinhigh.dimensionalmultimodaloptimization,ahybridalgorithmcombinedcmaeswithorthogonaldesignwithquantization(od/q)wasproposed。firstly，thesmallpopulationcmaeswasusedtorealizeafastsearching。whenorthogonalcmaesalgorithmtrappedinlocalextremum,basevectorsforod/qwereselecteddynamical