计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (15): 56-57.

• 理论研究 • 上一篇    下一篇

基于ART2改进算法的故障聚类研究

段霞霞,刘彦明,李小平,杨一展   

  1. 西安电子科技大学 通信工程学院,西安 710071
  • 收稿日期:2007-08-29 修回日期:2007-10-29 出版日期:2008-05-21 发布日期:2008-05-21
  • 通讯作者: 段霞霞

Study of fault cluster based on ART2 progressed algorithm

DUAN Xia-xia,LIU Yan-ming,LI Xiao-ping,YANG Yi-zhan   

  1. School of Communication Engineering,Xidian University,Xi’an 710071,China
  • Received:2007-08-29 Revised:2007-10-29 Online:2008-05-21 Published:2008-05-21
  • Contact: DUAN Xia-xia

摘要: ART2(自适应谐振理论2)算法是神经网络中一种可以对模拟输入信号或二值信号进行无监督聚类的算法,所以ART2算法能够降低数据挖掘中原始数据的预处理的复杂度,提高挖掘效率。针对ART2算法中出现的聚类中心偏移的缺点,采用ART2算法与K-均值算法相结合的方法来抑制ART2中聚类中心偏移的现象。通过仿真对该方法进行了验证。

关键词: 神经网络, 聚类, ART2, K-均值算法

Abstract: The Adaptive Resonance Theory 2(ART2),processing the analog signals of input patterns,is one of unsupervised clustering algorithms in neural networks.ART2 can reduce the pretreatment complexity of original data in data mining and improve the efficiency of mining.A progressed algorithm of ART2 which is composed of ART2 algorithm and K-Means algorithm can restrain cluster centers drift which is one of disadvantages of ART2 is developed.This method has been proved by simulation.

Key words: neural networks, cluster, Adaptive Resonance Theory2(ART2), K-means