期刊文献+

基于判别式的核独立元分析

Kernel independent component analysis for discrimination
下载PDF
导出
摘要 核独立元分析(KICA)法是近年来发展起来的核化算法,但难以将其用于故障诊断问题。为了解决该问题,对两种核独立元分析算法——基于受限协方差测度的方法KICA2和基于核互信息测度的方法KICA3进行变形得到适用于分类或故障诊断的形式。进一步分析了KICA2与一种核偏最小二乘(KPLS)方法的等价性以及KICA3与核主元分析(KPCA)的等价性。最后对Tennessee Eastman过程进行仿真,说明了方法的有效性。 Kernel independent component analysis(KICA) is a kernel method which has been rapidly developed in recent years and has been widely used in various kinds of fields.However,it was difficult to apply it to fault diagnosis.The aim of this paper was to improve KICA method to adapt the question.First,two kinds KICA,one was based on constrained cova-riance(COCO) measure which called KICA2,the other was based on kernel mutual information(KMI) measure which called KICA3,reformed to suit for classification or fault diagnosis.Then analysed the equivalence between KICA2 and kernel partial least squares(KPLS) and the equivalence between KICA3 and kernel principal component analysis(KPCA).Simulations on Tennessee Eastman(TE) process data show validity and effectivity of the proposed kernel algorithm.
作者 于春梅
出处 《计算机应用研究》 CSCD 北大核心 2011年第9期3330-3331,3343,共3页 Application Research of Computers
基金 西南科技大学博士基金资助项目
关键词 核独立元分析 故障诊断 判别 kernel independent component analysis fault diagnosis discrimination
  • 相关文献

参考文献9

二级参考文献40

共引文献32

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部