基于群体智能的思维进化算法设计  

Swarm intelligence based MEA design

作　　者：王芳[1] 谢克明[1] 刘建霞[1] 

机构地区：[1]太原理工大学信息工程学院,太原030024

出　　处：《控制与决策》2010年第1期145-148,共4页Control and Decision

基　　金：国家自然科学基金项目(60843006);高等学校博士学科点专项科研基金项目(2006112005);山西省自然科学基金项目(2008011027-4)

摘　　要：针对思维进化算法(MEA)没有充分利用公告板信息的问题,结合群体智能的优点,提出基于群体智能的思维进化算法,同时分析其算法的机制,设计利用群体信息共享进行子群体迁徙策略和拥挤浓度控制异化策略,提高了搜索速度,保证了种群的多样性.通过整个群体的总体优化特征体现了寻优方式的实现,使得收敛速度和全局收敛性均达到最好平衡.测试函数寻优及PID控制器参数整定实验,验证了算法的可行性和高效性.Mind evolutionary algorithm （MEA） is a kind of evolutionary algorithm of simulating the human＇s mind developments,which has two operators,similartaxis and dissimilation.Intrigued by the advantages of swarm intelligence（SI）,MEA based on SI is presented.The mechanisms of the improved MEA is analyzed.The group migration strategy and dissimilation strategy with density control are designed according to social information share.These two strategies enhance searching velocity,and ensure population diversity.The improved algorithm realizes optimization process by the total optimal characteristics of the whole population,and achieves the best balance between the searching velocity and global convergence.Optimum experiments by the testing functions and parameters of PID controller show the effectiveness and feasibility of the improved MEA.

关 键 词：思维进化算法 群体智能 子群体迁徙 拥挤浓度控制 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]