作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (14): 239-241. doi: 10.3969/j.issn.1000-3428.2011.14.081

• 工程应用技术与实现 • 上一篇    下一篇

基于相似度的核主元分析方法及其应用研究

张传标,倪建军,苗红霞,韩光洁   

  1. (河海大学计算机与信息学院,江苏 常州 213022)
  • 收稿日期:2011-01-27 出版日期:2011-07-20 发布日期:2011-07-20
  • 作者简介:张传标(1983-),男,硕士研究生,主研方向:智能算法,故障诊断;倪建军,副教授、博士;苗红霞,讲师、博士研究生;韩光洁,副教授、博士
  • 基金资助:
    河海大学常州校区创新基金资助项目(XZX/09B002-02);常州市输配电及节电技术重点实验室开放课题基金资助项目(CS09 04)

Study on Kernel Principal Component Analysis Method Based on Similarity and Its Application

ZHANG Chuan-biao, NI Jian-jun, MIAO Hong-xia, HAN Guang-jie   

  1. (College of Computer and Information, Hohai University, Changzhou 213022, China)
  • Received:2011-01-27 Online:2011-07-20 Published:2011-07-20

摘要: 常规核主元分析(KPCA)方法在对大样本数据分析建模时,存在运算复杂度高、建模时间长以及所需存储空间大等缺点。为此,提出一种基于相似度函数的快速核主元分析(SF-KPCA)方法。建立大样本数据间的相似度函数矩阵,分析数据样本间的相似程度,剔除冗余数据,再利用优化数据样本建立核主元分析模型,对数据样本进行分析。将SF-KPCA方法应用于高压断路器故障诊断中,实验结果证明了该方法的快速性和有效性。

关键词: 大样本数据, 相似度函数, 快速核主元分析, 高压断路器, 故障诊断

Abstract: In order to overcome the shortcomings of conventional Kernel Principal Component Analysis(KPCA) method in modeling and analyzing of large sample data(e.g., high computational complexity, long time modeling and large storage space etc.), a fast KPCA method based on Similarity Function(SF-KPCA) is proposed. The similarity function matrix of a large data samples is established to analyze the similarity between data samples, and the redundant data is eliminated. The KPCA model using the optimized data samples is built. The data samples are analyzed. The method is applied to the fault diagnosis of high voltage circuit breaker. Simulation results show the proposed method’s rapidity and effectivity.

Key words: large sample data, similarity function, fast Kernel Principal Component Analysis(KPCA), high voltage circuit breaker, fault diagnosis

中图分类号: