一种基于聚簇的自适应改进粒子群优化算法  

An Adaptively Improved Particle Swarm Optimization Algorithm Based on Clustering

作　　者：张延年[1] 吴士力[1] 刘永[2] ZHANG Yannian;WU Shili;LIU Yong(Electronic Information Engineering College,Nanjing Vocational Institute of Transport Technology,Nanjing 211188,China;College of Computer Science and Technology,Nanjing University ofScience and Technology,Nanjing 211188,China)

机构地区：[1]南京交通职业技术学院电子信息工程学院,南京211188 [2]南京理工大学计算机科学与技术学院,南京211188

出　　处：《实验室研究与探索》2018年第11期22-27,共6页Research and Exploration In Laboratory

基　　金：国家自然科学基金项目(61473155)

摘　　要：为避免陷入局部最优解,提出了一种聚簇自适应粒子群优化算法(Heterogeneous Particle Swarm Optimization Based-Clustering,HPSOC)。HPSOC利用K-均值聚簇方法将种群动态划分为包含不同粒子数量的异构聚簇;利用Ring结构进行不同聚簇间的信息交换。通过动态聚簇,粒子可以在不同聚簇间移动,使得信息交换速度更快。提出一种新的聚簇搜索阶段发现方法,使得每个粒子可以根据其聚簇搜索阶段动态自适应地调整粒子参数,进而使得每个聚簇可以获得与其阶段相匹配的自我搜索能力。通过6种基准函数的测试,与两种同类型的粒子群优化算法进行性能比较。结果表明,HPSOC算法不仅可以改进粒子的多样性和积极性,还可以增加粒子的搜索能力和收敛性,从而降低对于参数初始选择的依赖性。For dealing with falling into local optimal solution, a novel adaptive particle swarm optimization algorithm （HPSOC） is proposed based on clustering. The HPSOC takes K -means clustering method to divide the swarm into different neighborhoods dynamically. These neighborhoods have different number of particles and are heterogeneous clusters. A Ring-structure is used to exchange information among clusters. With the dynamic clustering, particles can move from one cluster to another so that information exchange is much faster. Furthermore, a novel discriminating method is proposed to detect the exploring stage of a cluster. Every particle adjusts its parameters automatically according to the exploring stage of its cluster, it is an adaptive process. By these operations, each cluster can have its own exploring capability matching its stage. Finally, through the tests with six benchmark functions, comparisons are made with two same types of algorithms. Experimental results show that the HPSOC can improve diversity and energetic of the particles, increase exploring ability and convergence, and reduce the dependence of initial election of parameters.

关 键 词：粒子群优化 自适应 聚簇 异构性 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]