基于人工免疫粒子群优化算法的动态聚类分析  

A Dynamic Clustering Analysis Based on Artificial Immune Particle Swarm Optimization Algorithm

作　　者：王磊[1] 吉欢[1] 徐庆征[1] 

机构地区：[1]西安理工大学计算机科学与工程学院,陕西西安710048

出　　处：《西安理工大学学报》2008年第4期390-394,共5页Journal of Xi'an University of Technology

基　　金：国家自然科学基金资助项目(60603026)

摘　　要：模糊C-均值聚类算法受初始化影响较大,在迭代时容易陷入局部极小值。将粒子群优化算法与模糊C-均值聚类算法相结合,提出一种新颖的动态聚类算法。该算法利用人工免疫思想改进粒子群优化过程,在很大程度上避免了粒子群算法和聚类算法早熟现象的发生,全局搜索能力和局部搜索能力优于同类算法。利用聚类理论中的经验规则kmax≤n来确定聚类数k的搜索范围,在最优粒子基础上进化新一级种群,该方案可有效提高算法的收敛速度。两组数据的仿真实验表明,新算法优于传统模糊C-均值聚类算法,具有收敛速度快和解的精度高的特点。The fuzzy C-means （FCM） clustering algorithm is sensitive to the situation of the initialization and easy to fall into the local minimum when iterating. A novel dynamic clustering algorithm based on the combination of FCM algorithm with particle swarm optimization （PSO） algorithm is proposed in this paper. The artificial immune mechanism is introduced in improving the process of particle swarm optimization, so that the premature convergence of PSO and FCM algorithm is avoided, which makes the algorithm＇ s global search and local search capability appear better than normal ones. The search range is determined according to the experiential rule kmax≤√n and the new population is evolved based on the optimal particles, which makes the convergence speed increased. The experiments on two data sets show that when compared with the classical clustering method, the new algorithm is capable of improving the clustering performance significantly in convergence ability and solution precision.

关 键 词：人工免疫系统 粒子群优化算法 动态聚类 收敛性 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]