基于Lotka-Volterra模型的双群协同竞争粒子群优化算法  

Lotka-Volterra model based particle swarm optimization

作　　者：吴宪祥[1] 郭宝龙[1] 王娟[2] 

机构地区：[1]西安电子科技大学机电工程学院,西安710071 [2]空军工程大学电讯工程学院,西安710077

出　　处：《控制与决策》2010年第11期1619-1624,共6页Control and Decision

基　　金：国家863计划项目(2006AA01Z127);国家自然科学基金项目(60802077;61003196);中央高校基本科研业务费资助项目(K50510040007)

摘　　要：针对粒子群优化算法易出现早熟收敛的问题,提出了基于Lotka-Volterra模型的双群协同竞争粒子群优化算法(LVPSO).LVPSO算法借鉴种群生态学中著名的Lotka-Volterra双群协同竞争模型,讨论了两种种群协同竞争方案,通过群内和群间竞争增加粒子的多样性,提高了种群摆脱局部极值的能力.对5个典型基准测试函数进行优化实验表明,LVPSO在收敛速度和优化精度方面均有良好的表现.A particle swarm optimization algorithm inspired by Lotka-Volterra model（LVPSO） is proposed to avoid the premature convergence problem.The famous Lotka-Volterra model in ecology is introduced into basic particle swarm optimization algorithm,and two different cooperative-competitive schemes are discussed.The diversity of particles are increased by intraspecific and interspecific competition,and the ability of particles breaking away from the local extremum is improved remarkably.The experimental results show that the proposed LVPSO algorithm can converge in higher speed and higher precision by optimizing five typical benchmark functions.

关 键 词：粒子群优化 早熟收敛 LOTKA-VOLTERRA模型 多样性 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]