粒子群与差异演化的混合进化算法研究  

Particle Swarm Optimization Algorithm Optimal Design

作　　者：洪少春[1] 毛恒[2] 王永初[2] 

机构地区：[1]泉州师范学院应用科技学院,福建泉州362000 [2]华侨大学机电及自动化学院,福建泉州362021

出　　处：《泉州师范学院学报》2008年第4期9-15,共7页Journal of Quanzhou Normal University

基　　金：福建省自然基金资助项目(2006J0042)

摘　　要：将标准粒子群(PSO)算法和差异演化(DE)算法混合进行进化,把差异演化算法的优势带入到粒子群算法中,利用差异演化算法其本身具备的对粒子个体的交叉和变异操作使标准粒子群算法种群保持寻优所需的多样性.并将这种算法与标准粒子群算法、G-PSO算法等进行了对比研究,通过混合进化的方法使粒子群种群保持多样性,实验结果表明新算法避免陷入"早熟"或"停滞"的能力大大提高.Carry out evolution of mix discrepancy evolution （ DE ） algorithm and standard PSO algorithm, bring the advantage of DE algorithm into the algorithm of the Particle Swarm, make use of DE algorithm＇s own faction, that is overlapping and variation operating, so as to maitain required diversity when seeking Optimization in PSO algorithm. Successfully introdunce this algorithm to the optimization of nerve network,and consquently it can guarantee particle speed being updated at greater level,maitain better search ability,and greatly increase the ability to avoid falling into prematurity. All these are sensitive and rich for the diversity of population.

关 键 词：粒子群优化算法 早熟 混合进化算法 差异演化 

分 类 号：O442[理学—电磁学] O125.8[理学—物理]