摘要
本文研究无线传感器网络数据的聚类分析问题.针对传统k-means对初始聚类中心敏感和易于陷入局部次优解的缺点,提出一种基于传感器网络的分布式免疫遗传k-means聚类算法.该算法将聚类中心作为染色体,通过遗传算法来优化传统k-means聚类算法的初始聚类中心,将免疫算法的选择操作引入染色体的遗传进化中,使染色体的浓度和适应度共同对其在进化中被选择产生影响,实现了染色体种群的多样性保持机制和自我调节功能,将搜索工作引向全局最优,较好地解决了k-means算法的早熟现象问题.实验结果证明,本文算法改进了数据的聚类划分效果,能够把聚类结果快速收敛至全局最优,聚类准确率较高.
The problem of wireless sensor network data clustering analysis is researched. In order to solve the shortcomings that traditional k-means is seesitive for the initial clustering center and easily fall into local optimal solution, a distributed immune genetic algorithm k-means based on sensor network is proposed. In the algorithm, cluster center has been as chromosomes, and through the genetic algorithm to optimize the initial cluster centers of traditional k-means clustering algorithm, meanwhile the selecting operation of immune algorithm is introduced into the genetic evolution of chromosome, causing the concentration and fitness of chromosome will impact whether the chromosome is selected in evolutionary, which can realize chromosome population diversity maintaining mechanism and self regulating function, and the searching achieves the global optimum, so the algorithm can solve the prematurity problem of k-means algorithm better. The experimental results show that, the algorithm has been improved data clustering effect, and made the clustering can converge to the global optimal clustering rapidly, and the clustering has higher accuracy.
出处
《微电子学与计算机》
CSCD
北大核心
2013年第3期156-159,共4页
Microelectronics & Computer
基金
广西自然科学基金青年项目(2012jjBAG0074)
广西教育厅科研项目(201010LX428
201106LX504
201106LX510
200103YB195)
关键词
聚类分析
免疫算法
遗传算法
无线传感器网络
clustering analysis
immune algorithm
genetic algorithm
wireless sensor network
