摘要
受无标度网络结构特性的启发,将BA模型的"择优连接"机制进行扩展,引入微粒群群体组织方式的构造过程,提出基于高聚集性的无标度网络模型的微粒群算法。算法初期微粒被随机分布在环形结构中,随着搜索的进行不断增加新的微粒,并依据节点度和节点间的距离增加新的连接,最终形成具有高聚集性的无标度网络模型。这样,群体中多数微粒进行局部范围的搜索,而少量微粒按照全局模式搜索,两种方式相互制衡。仿真实验表明,改进后的算法能获得更好的收敛精度和进化速度。
Enlightened by the properties of scale-free network model,"preferential attachment" mechanism of the BA model is extended and introduced into particle swarm optimization,and a novel particle swarm optimization with highly-clustered scale-free network model(PSO-HCSF) is proposed.At the early stage of the algorithm,particles were randomly distributed in a ring,new particles are continuously added into the population with searching,and based on the node degree and the distance between nodes new connections are produced,and a high aggregation degree of scale-free network model are formed in the end.In this way,the majority of particles search in local scope and a small amount of particles search with the overall pattern,two ways check and balance.Experimental simulations show that the new method obtains better evolution speed and convergence performance.
出处
《复杂系统与复杂性科学》
EI
CSCD
2010年第1期82-87,共6页
Complex Systems and Complexity Science
关键词
微粒群算法
无标度网络模型
择优连接
高聚集性
particle swarm optimization
scale-free network model
preferential attachment
high cluster