基于多尺度基本熵的CFS聚类滚动轴承故障诊断  

Fault Diagnosis Method of Roller Bearings Based on MBSE and CFS Clustering

作　　者：海波[1] 姚海龙[1] 孟丛丛[1] 

机构地区：[1]兰州城市学院电子与信息工程学院,兰州730070

出　　处：《轴承》2017年第4期54-60,共7页Bearing

摘　　要：提出了一种基于多尺度基本熵的CFS聚类滚动轴承故障诊断方法,首先使用多尺度基本熵对滚动轴承振动信号进行多尺度分解,然后使用PCA模型对得到的多尺度熵值特征向量进行降维操作,最后选择第1~2主成分作为CFS聚类算法的输入进行滚动轴承的故障诊断。试验结果表明:在聚类效果相同的情况下,多尺度基本熵的计算效率高于多尺度排列熵模型,并且CFS聚类算法的故障识别效果较好。A method based on MBSE （ Muhiscale Base - scale Entropy ） and CFS （ Clustering by Fast Search） cluste- ring algorithm is proposed for roller bearings diagnosis recognition. Firstly, the roller bearings vibration signals under different conditions are decomposed into feature information under different scales by using MBSE. Secondly, the PCA （Principal Component Analysis） model is used to reduce the dimension of multiscale entropy eigenvectors, and finally the first two PCs （Principal Components ） are regarded as the input of CFS clustering algorithm for fulfill the roller bearings fault recognition. The result shows that the computational efficiency of MBSE model is higher than MPE and the fault recognition for roller bearings is good by using CFS clustering algorithm under the same condition.

关 键 词：滚动轴承 故障诊断 CFS聚类算法 主成分分析 多尺度基本熵 

分 类 号：TH133.33[机械工程—机械制造及自动化]