摘要
针对如何从非线性、多分量、强背景噪声的滚动轴承早期故障振动信号中有效提取出微弱故障特征并准确判断故障类型,提出基于非局部均值去噪和快速谱相关的故障诊断方法。首先利用非局部均值去噪算法对原始振动信号进行降噪预处理,提高信号信噪比。然后,对降噪信号进行快速谱相关分析,增强信号中的周期成分,获得快速谱相关谱及其对应的增强包络谱。最后,将增强包络谱中幅值突出的频率成分与故障特征频率进行对比,判定故障类型并实现故障诊断。使用本文提出方法对仿真故障信号、实验故障信号进行分析。研究结果表明:相较于快速谱相关方法、谱峭度结合非局部均值去噪方法以及非局部均值去噪结合经验模态分解方法,本文提出方法可以抑制轴承早期故障振动信号中的背景噪声,有效提取出微弱故障特征,准确判断故障类型,避免出现误诊。
Aiming at how to effectively extract weak fault features from early fault vibration signals of rolling bearing with nonlinear,multi-component and strong background noise and accurately judge fault types,a fault diagnosis method based on nonlocal mean denoising and fast spectral correlation was proposed.Firstly,the nonlocal mean denoising algorithm was used to pre-process the original vibration signal to improve the signal-tonoise ratio.Then,the fast spectral correlation analysis was performed to enhance the periodic components of the signal,and the fast spectral correlation spectrum and its corresponding enhanced envelope spectrum were obtained.Finally,the frequency components with prominent amplitude in the envelope spectrum were compared with the fault characteristic frequencies to determine the fault type and realize fault diagnosis.The proposed method was used to analyze the simulated fault signal and the experimental fault signal.The results show that compared with the method based on nonlocal mean denoising and spectral kurtosis and the method based on nonlocal mean denoising and empirical mode decomposition,the proposed method can suppress the background noise of bearing early fault vibration signal,effectively extract weak fault features,and accurately judge the fault type and avoid misdiagnosis.
作者
万书亭
彭勃
WAN Shuting;PENG Bo(Department of Mechanical Engineering,North China Electric Power University,Baoding 071003,China)
出处
《中南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2020年第1期76-85,共10页
Journal of Central South University:Science and Technology
基金
国家自然科学基金资助项目(51777075)
河北省自然科学基金资助项目(E2019502064)
中央高校基本科研业务费专项资金资助项目(2019QN131)~~
关键词
滚动轴承
特征提取
微弱故障诊断
非局部均值去噪
快速谱相关
rolling bearing
feature extraction
weak fault diagnosis
nonlocal mean denoising
fast spectral correlation
