一种含最优变异的多微粒群优化算法  

A multi-swarm particle swarm optimization algorithm characteristic of optimum mutation

作　　者：陈彧[1] 邹秀芬[2] 张建军[1] 

机构地区：[1]海军工程大学理学院,武汉430033 [2]武汉大学数学与统计学院,武汉430072

出　　处：《海军工程大学学报》2007年第2期103-106,共4页Journal of Naval University of Engineering

基　　金：海军工程大学科学研究基金资助项目(HGDJJ2005005)

摘　　要：针对微粒群优化算法中的固有缺点,提出了带有最优变异算子的多微粒群优化算法,采用多个微粒群对目标函数进行寻优,并在寻优过程中对子群中最优微粒引入了最优变异算子。通过这样的处理,算法可以预防早熟收敛并具有更快的收敛速度和更好的局部开发能力。对一组测试函数的模拟实验结果表明,带最优变异的多微粒群优化算法可以摆脱局部最优解对微粒的吸引,在较少的代数内就能够获得好的优化结果。With respect to the inherent deficiency in the particle swarm optimization algorithm, a novel multl-swarm particle swarm optimization algorithm with optimum mutation is presented in this paper. By applying the mutation operator to the best particle in the sub-warm, the objective function can be optimized by the algorithm, which can prevent premature convergence and has better convergence and local exploitation ability. Experiments are conducted on a set of benchmark functions and the results show that it can break away from the attraction of the local optimal solution and thus achieve excellent results in fewer generations.

关 键 词：演化计算 多微粒群优化算法 最优变异 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]