引用本文:秦娜,金炜东,黄进,李智敏.高速列车转向架故障信号的聚合经验模态分解和 模糊熵特征分析[J].控制理论与应用,2014,31(9):1245~1251.[点击复制]
QIN Na,JIN Wei-dong,HUANG Jin,LI Zhi-min.Ensemble empirical mode decomposition and fuzzy entropy in fault feature analysis for high-speed train bogie[J].Control Theory and Technology,2014,31(9):1245~1251.[点击复制]
高速列车转向架故障信号的聚合经验模态分解和 模糊熵特征分析
Ensemble empirical mode decomposition and fuzzy entropy in fault feature analysis for high-speed train bogie
摘要点击 2547  全文点击 1452  投稿时间：2013-09-23  修订日期：2014-04-20
查看全文  查看/发表评论  下载PDF阅读器
DOI编号  10.7641/CTA.2014.31002
  2014,31(9):1245-1251
中文关键词  高速列车转向架  特征提取  聚合经验模态分解  模糊熵  最小二乘支持向量机
英文关键词  high speed train bogie  feature extraction  ensemble empirical mode decomposition (EEMD)  fuzzy en- tropy  least squares support vector machine (LSSVM)
基金项目  国家自然科学基金重点资助项目(61134002).
作者单位E-mail
秦娜* 西南交通大学 电气工程学院 qinna@swjtu.cn 
金炜东 西南交通大学 电气工程学院  
黄进 西南交通大学 电气工程学院  
李智敏 西南交通大学 材料科学与工程学院  
中文摘要
      为了对高速列车转向架关键部件进行状态监测, 利用转向架故障振动信号的特点, 提出了一种结合聚合经 验模态分解和模糊熵的特征提取方法. 对故障信号进行聚合经验模态分解, 得到一系列具有不同物理意义的简单 成分信号, 采用相关分析法选取最能反映原信号特征的本征模态函数. 对这些本征模态函数和原信号分别计算模 糊熵值构成多尺度复杂性度量的特征向量, 输入最小二乘支持向量机中进行分类识别, 与模糊熵特征相比得到了更 好的识别效果, 证明了算法的有效性.
英文摘要
      To monitor the working condition of key components in a high-speed train bogie, we make use of the char- acteristics of the fault signal to propose a new approach for feature extraction based on ensemble empirical mode de- composition and fuzzy entropy. Firstly, we decompose the fault vibration signal by using the ensemble empirical mode decomposition to obtain a series of simple composition signal with different physical significance. Then, we employ the correlation analysis to sift out intrinsic mode functions (IMF) that have largest correlation coefficients with the original signal and use it as the data source.The fuzzy entropy of IMF and the fuzzy entropy of the initial signal are calculated to form a multi-scale complexity measure feature vector. Finally, the feature vector is put into the least squares support vector machine for classification and identification. Comparing with the fuzzy entropy, we find the proposed approach gives better fault identification results. The efficacy of this method is validated.