基于多尺度模糊熵和主成分分析的轴承故障特征提取  

Fault features extract of rolling bearing based on multiscale fuzzy entropy and principal component analysis

作　　者：李生鹏[1] 韦朋余[1] 丁峰 竺一峰[1] 姜朝文 周舒豪 LI Sheng-peng;WEI Peng-yu;DING Feng;ZHU Yi-feng;JIANG Chao-wen;ZHOU Shu-hao(China Ship Scientific Research Center,Wuxi 214082,China)

机构地区：[1]中国船舶科学研究中心,江苏无锡214082

出　　处：《船舶力学》2018年第10期1277-1285,共9页Journal of Ship Mechanics

基　　金：国家重点基础研究发展计划项目(973项目;2014CB046706);江苏省绿色船舶技术重点实验室资助

摘　　要：针对滚动轴承故障特征难以提取的问题,文章提出了基于多尺度模糊熵(MFE)和主成分分析(PCA)相结合的滚动轴承故障特征提取方法。首先利用经验模态分解(EMD)将原始振动信号分解成若干个本征模态函数(IMF),并根据相关系数和峭度值准则剔除虚假IMF分量;然后在不同尺度下求取真实IMF分量的模糊熵值,利用PCA对其进行降维处理,形成能表征不同轴承故障的特征向量,最后借用支持向量机对其进行诊断验证。实验表明,该方法可以有效地提取轴承故障信息,对4种轴承状态的识别率为95%,实现了对轴承故障的精确诊断。Because rolling bearing fault features are difficult to be extracted,in this paper,a rolling bearing fault features extract method is proposed based on multiscale fuzzy entropy(MFE)and principal component analysis(PCA).Firstly,the empirical mode decomposition(EMD)method is used to decomposing vibration data set into several intrinsic mode functions(IMF),and the false IMF can be eliminated based on correlation coefficient and kurtosis value.Secondly,the MFE of true IMF would be calculated in the different scale and dimensions can be reduced through PCA;as a result,the feature vector representing different rolling bearing fault could be obtained.Finally,the rolling bearing fault can be recognized based on support vector machine.The result shows that this method can effectively extract the fault informations,and the recognition rate is up to 95%.

关 键 词：故障诊断 经验模态分解 多尺度模糊熵 主成分分析 

分 类 号：U664.21[交通运输工程—船舶及航道工程]