基于搜寻者优化算法的K-means聚类算法  

K-means clustering algorithm based on seeker optimization algorithm

作　　者：王盛慧[1] 夏永丰 WANG Shenghui;XIA Yongfeng(College of Electrical and Electronic Engineering,Changchun University of Technology,Changchun,Jilin 130012,China)

机构地区：[1]长春工业大学电气与电子工程学院,吉林长春130012

出　　处：《燕山大学学报》2018年第5期422-426,433,共6页Journal of Yanshan University

基　　金：吉林省科技发展计划项目(20150203003SF)

摘　　要：针对K-means聚类算法易陷入局部最优的问题,提出一种改进的K-means算法,将搜寻者优化算法(SOA)和K-means聚类算法结合起来,利用SOA鲁棒性好、全局搜索能力强的特点,通过确定搜寻者的搜索方向和搜索步长,更新搜寻者的位置,进行全局寻优,提高K-means聚类算法的聚类精确度。在仿真实验过程中,首先,选取具有代表性的处于三种燃烧状态的水泥回转窑窑内视频图像为研究对象,分别采用K-means算法和改进后的算法进行仿真实验,实验结果表明,改进算法所获得的图像聚类效果更加精确;然后,分别用上述两种算法对数据集Iris和Wine进行相关测试,结果表明,改进算法的聚类精确度和运行效率都得到了有效提高。Aiming at the problem that the K-means clustering algorithm is easy to fall into the local optimal problem,an improved K-means algorithm is proposed,which combines the seeker optimization algorithm(SOA)and the K-means clustering algorithm to make use of the SOA robustness,The global search ability is strong,and the clustering accuracy of K-means clustering algorithm is improved by determining the searching direction and searching step size of the seeker,updating the position of the seeker,and searching for the global optimization.In the simulation experiment,firstly,the representative video images of cement kiln in three combustion states are selected as the research objects.K-means algorithm and improved algorithm are used to simulate the experiment.The experimental results show that the image clustering effect obtained by the algorithm is more accurate.Then,we use the above two algorithms to test the data set Iris and Wine respectively.The results show that the clustering accuracy and running efficiency of the improved algorithm are improved effectively.

关 键 词：K-MEANS聚类算法 搜寻者优化算法 全局寻优 聚类精确度 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]