基于自适应分数阶循环平稳盲反卷积的滚动轴承故障诊断方法  

Fault diagnosis method for rolling bearing based on adaptive fractial-order cyclostationary blind deconvolution

作　　者：吴怡 王金海 杨建伟[1,2] 徐丹萍 WU Yi;WANG Jinhai;YANG Jianwei;XU Danping(School of Mechanical-Electronic and Vehicle Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Beijing Key Laboratory of Performance Guarantee on Urban Rail Transit Vehicles,Beijing University of Civil Engineering and Architecture,Beijing 100044,China)

机构地区：[1]北京建筑大学机电与车辆工程学院,北京100044 [2]北京建筑大学城市轨道交通车辆服役性能保障北京市重点实验室,北京100044

出　　处：《北京交通大学学报》2024年第5期162-170,共9页JOURNAL OF BEIJING JIAOTONG UNIVERSITY

基　　金：国家自然科学基金(52205083,52272385,51975038);北京市自然科学基金(L211008)。

摘　　要：在工业领域中,由于机械设备恶劣的运行环境和复杂的工作条件,轴承故障信号常受到强烈背景噪声的干扰,难以有效提取故障特征.针对此问题,提出一种基于自适应分数阶循环平稳盲反卷积(Adaptiveγ-order Cyclostationary Blind Deconvolution,ACYCBDγ)的滚动轴承故障诊断方法.首先,引入一种新的局部峭度比指标(Local peak ratio,Lpr),确定最优滤波器长度.然后,计算基于高斯平稳模型的估计分数阶,构建分数阶循环平稳盲反卷积.最后,通过公开数据集和实测数据集验证所提模型的性能.结果表明:ACYCBDγ与最小熵反卷积、最大相关峰度反卷积和最大二阶循环平稳盲反卷积(Maximum Second-order Cyclostationarity Blind Deconvolution,CYCBD)相比,在公开数据集上的抑制比分别提高了20.61%、17.85%和44.95%,在实测数据集上的抑制比分别提高了53.63%、60.27%和55.16%;在−10~−20 dB的强信噪比影响下,ACYCBDγ相较于CYCBD,Lpr提升了87.51%.ACYCBDγ能够减弱噪声和干扰信号的影响,实现强噪声背景下轴承故障特征的准确提取.In the industrial field,bearing fault signals are often subject to significant interference from strong background noise due to the harsh operating environment and complex working conditions of mechanical equipment,making it challenging to effectively extract fault characteristics.To address this,this paper proposes a fault diagnosis method for rolling bearing based on Adaptiveγ-order Cyclo⁃stationary Blind Deconvolution(ACYCBDγ).First,a novel metric,the Local peak ratio(Lpr),is in⁃troduced to determine the optimal filter length.Then,the estimated fractional order based on a Gaussian smooth model is calculated to construct the fractional-order cyclostationary blind deconvolution.Finally,the proposed model’s performance is validated using both public and real-world datasets.The results demonstrate that ACYCBDγachieves suppression ratios that are 20.61%,17.85%,and 44.95%higher than those of Minimum Entropy Deconvolution,Maximum Correlated Kurtosis Decon⁃volution,and Maximum Second-order Cyclostationarity Blind Deconvolution(CYCBD),respectively,on the public dataset.On the real dataset,the suppression ratios are improved by 53.63%,60.27%,and 55.16%,respectively.Under signal-to-noise ratios of−10 to−20 dB,ACYCBDγen⁃hances the Lpr by 87.51%compared to CYCBD.Therefore,ACYCBDγeffectively reduces the impact of noise and interference,enabling the accurate extraction of bearing fault features in strong noise environments.

关 键 词：轴承故障诊断 循环平稳 盲反卷积 特征提取 

分 类 号：TH133.33[机械工程—机械制造及自动化]