基于改进HHT的奇异值分解和马氏距离的滚动轴承故障诊断  

Rolling Bearing Fault Diagnosis Based on Improved HHT Singular Value Decomposition and Mahalanobis Distance

作　　者：杨恭勇 周小龙[1] 梁秀霞 李家飞[2] 

机构地区：[1]东北电力大学工程训练教学中心,吉林吉林132012 [2]河南信宇石油机械制造股份有限公司,河南濮阳457001

出　　处：《东北电力大学学报》2017年第4期56-60,共5页Journal of Northeast Electric Power University

摘　　要：针对滚动轴承振动信号的非平稳性以及故障诊断样本总量少的特点,提出一种基于改进希尔伯特-黄变换的奇异值分解和马氏距离相结合的故障诊断方法。首先,采用改进经验模态分解方法将所测不同工况下的滚动轴承信号分解成多阶固有模态函数,并根据各模态函数的特性选取对各工况信号敏感的模态分量;其次,对敏感模态函数分量组成的特征向量进行奇异值分解,并以分解结果的期望值作为诊断的特征值;最后,将马氏距离判别算法应用于滚动轴承的工况和类型判别。试验结果表明,本文所提方法能有效识别出滚动轴承的工作状态,具有一定的应用价值。Aiming at the non-stationary feature of the rolling bearing vibration signal and the fault samples are always in a small number, a rolling bearing fault diagnosis method based on improved Hilbert-Huang transform singular value decomposition and Mahalanobis distance is proposed. Firstly, the vibration signal in different condition is decomposed by improved empirical mode decomposition, and the intrinsic mode functions can be obtained and sensitive mode functions are selected by the sensitivity selection method. Then, using the singular value decomposition technique to decompose the feature vector matrix based on sensitive mode functions, and the expected value of the matrix is regarded as the state feature value of the roller bearing vibration signal. Fi- nally, Mahalanobis distance is used to identify the roiling bearing fault pattern and condition. The experiment results show that this method can identify roiling bearing fault patterns effectively and offer a practical method for its fault diagnosis.

关 键 词：希尔伯特-黄变换 奇异值分解 马氏距离 滚动轴承 故障诊断 

分 类 号：TH17[机械工程—机械制造及自动化]