基于双群体进化规划的动态优化算法  

A Dynamic Optimization Algorithm Based on Bi-Population Evolutionary Programming

作　　者：蒲保兴[1] 杨路明[1] 肖潇[1] 

机构地区：[1]中南大学信息科学与工程学院

出　　处：《计算机仿真》2008年第9期182-186,共5页Computer Simulation

基　　金：湖南省教育厅科研基金(05C671)资助

摘　　要：动态优化问题的优化环境随时间变化导致了最优解随时间移动。为了有效地跟踪最优解,提出了一个基于双群体进化规划的动态优化算法。局部搜索群体运用高斯变异算子,并接受已有信息;全局搜索群体运用柯西变异算子,与已有信息隔离并传送较优个体至局部搜索群体。在进化过程中,它们的群体规模动态地变化。算法有效地利用了已有信息,实现了全局搜索与局部搜索的分离,适合于求解环境变化方式未知的动态优化问题。对三个动态优化模型进行了测试,并与随机初始化群体法进行了比较,仿真结果表明了提出的算法是有效的。The optimal solution of dynamic optimization problem moves with time due to environment changing.In order to efficiently track the optimal solution,this paper proposes a dynamic optimization algorithm based on bi-population evolutionary programming.Local search population receives previous information and adopts Gaussian mutation,while the global search population,which is insulated from previous information,uses Cauchy mutation and transports its elitists to local search population.In the process of evolution,their population size is dynamically altered.The algorithm efficiently makes use of previous information,separates local search process from global search process,and is fit for solving the dynamic optimization problems where environment change mode is unclear.Experiments for three type dynamic optimization models show that the proposed method is more effective than the reinitializing strategy.

关 键 词：动态优化 进化规划 多样性 双群体 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]