混合蛙跳算法的Markov模型及其收敛性分析  

The Markov Model of Shuffled Frog Leaping Algorithm and Its Convergence Analysis

作　　者：骆剑平[1] 李霞[1] 陈泯融[1] 

机构地区：[1]深圳大学信息工程学院,广东深圳518060

出　　处：《电子学报》2010年第12期2875-2880,共6页Acta Electronica Sinica

基　　金：国家自然科学基金资助项目(No.60772148);高等学校博士点基金资助项目(No.200805900001)

摘　　要：本文就混合蛙跳算法(Shuffled Frog Leaping Algorithm,SFLA)建立了Markov链数学分析模型,详细分析了该Markov链的性质,证明青蛙族群状态序列是齐次Markov链.在此基础上,通过分析族群状态序列的转移过程,指出序列必将进入最优状态集.同时证明混合蛙跳算法满足随机搜索算法全局收敛的两个条件,能够保证全局收敛.The Markov chain model for the shuffled frog leaping algorithm（SFLA） was established.It was shown that the frog memeplex state sequence containing both the frog states and the current local and the global optimal frog states constructs a homogeneous Markov chain.The transition process of the frog memeplex state sequence was analyzed,and the conclusion that sequence will eventually converges to the optimal state set was drawn.Furthermore,it was proved that the shuffled frog leaping algorithm ensures global convergence as it meets the global convergence criterions of random search algorithms.

关 键 词：智能优化 混合蛙跳算法 MARKOV链 全局收敛 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TP183[自动化与计算机技术—控制科学与工程]