基于循环谱分析的鲁棒性滚动轴承故障特征提取方法  

Robust rolling bearing fault feature extraction method based on cyclic spectrum analysis

作　　者：晏云海 郭瑜[1] 伍星[1] YAN Yunhai;GUO Yu;WU Xing(Facllty of Mechanical and Electrical Engineering,Kunming University of Science and Technology,Kunming 650504,China)

机构地区：[1]昆明理工大学机电工程学院,昆明650504

出　　处：《振动与冲击》2022年第6期1-7,共7页Journal of Vibration and Shock

基　　金：国家自然科学基金(51675251);云南省科技重大专项(202002AC080001)。

摘　　要：循环平稳分析是滚动轴承故障特征提取的重要方法之一,但在用于滚动轴承故障特征提取时,存在因干扰成分较强而不能有效提取轴承故障特征的问题。为能在干扰环境中有效提取滚动轴承故障信息,基于循环谱分析提出一种鲁棒性滚动轴承故障特征提取方法。首先通过离散随机分离(discrete random separation,DRS)分析分离信号中的周期分量,提取其随机分量;随后用Teager能量算子(Teager energy operator,TEO)提取随机分量的振动能量序列;再对该序列进行快速谱相关(fast spectral correlation,Fast-SC)分析,采用基于能量熵的能量差异系数评价各循环频率(阶次)切片的能量强度;最终经熵加权降低无关干扰成分影响以有效提取故障特征。通过传统的快速谱峭度、快速谱相关和基于总变差去噪的快速谱相关分析方法与该方法对美国智能维护系统中心的滚动轴承振动数据以及实测齿轮箱复合故障试验信号进行对比分析,验证了该方法在滚动轴承故障诊断应用中的优势。Cyclic stationary analysis is one of the important methods for rolling bearing fault feature extraction,however,by which the bearing fault features sometimes cannot be effectively extracted due to the existense of excessive irrelevant interference components.A robust rolling bearing fault feature extraction method based on cyclic spectrum analysis was proposed to solve the problem.The random component of the signal was extracted by the discrete random separation(DRS),and then the vibration energy sequence of the random component was calculated by the Teager energy operator(TEO).With the fast spectral correlation(Fast-SC)analysis,the energy intensity of each cycle frequency(order)slice was characterized by an energy difference coefficient based on the energy entropy.The influence of the irrelevant interference component was reduced by entropy weighting.Then,the fault features of the rolling bearing were effectively extracted.The advantages of the method were verified by its application in the real data rolling bearing fault diagnosis through an experimental comparison with the methods of fast spectral kurtosis,fast spectral correlation and fast spectral correlation based on total variation de-noising.

关 键 词：滚动轴承 循环谱分析 振动能量 故障特征提取 

分 类 号：TH133.33[机械工程—机械制造及自动化] TH113.21