EEMD和优化的频带熵应用于轴承故障特征提取

EEMD and optimized frequency band entropy for fault feature extraction of bearings

针对滚动轴承早期故障特征提取困难的问题,提出了将集合经验模态分解(Ensemble Empirical Mode Decomposition,EEMD)和优化的频带熵(OFBE)相结合的轴承故障特征提取方法。针对EEMD的多个本征模态分量(Intrinsic Mode Function,IMF),如何选出更能反映故障特征的敏感IMF的问题,提出一种基于频带熵的敏感IMF的选取方法。首先,对原始振动信号进行EEMD分解,获得一系列IMFs;然后,对原信号和各个IMF分量求频带熵,在熵值最小处设计带通滤波器带宽作为特征频带,比较各个IMF的特征频带与原信号熵最小值所处频带之间的从属关系,进而选出反映故障特征的敏感IMF。由于背景噪声的影响,从选取的IMF中难以准确地得到故障频率。因此,利用FBE在选取IMF的基础上设计的带通滤波器,并提出利用包络峭度最大值原则优化带宽,然后对其进行带通滤波,并进行包络功率谱分析以提取故障特征频率。将该方法应用到轴承仿真数据和实际数据中,能够实现轴承故障特征的精确诊断,证明了该方法的有效性和优势。

Aiming at the difficulty in extracting early fault features of rolling bearings,a bearing fault feature extraction method combining ensemble empirical mode decomposition(EEMD)and optimized frequency band entropy(OFBE)is proposed.Aiming at the problem that EEMD decomposition has multiple intrinsic mode functions(IMFs),and how to select the sensitive IMF that can better reflect fault characteristics,a method based on FBE for sensitive IMF is proposed.First,the original vibration signal is subjected to EEMD decomposition to obtain a series of IMFs.Then,the FBE is obtained for the original signal and each IMF component.The bandwidth of the band-pass filter is designed as the characteristic band at the minimum entropy value.The affiliation between the characteristic band of each IMF and the characteristic band of the original signal is compared to select a sensitive IMF that reflects the characteristics of the fault.Then,due to the influence of background noise,it is difficult to accurately obtain the fault frequency from the selected IMF.Therefore,the band-pass filter designed by FBE based on IMF is selected,and the bandwidth is optimized by the envelope kurtosis maximum principle.Then band-pass filtering is performed,and the envelope power spectrum analysis is carried out to extract the fault characteristic frequency.Applying the method to the simulation data and actual data of the bearing can accurately diagnose the bearing fault characteristics,which proves the effectiveness and advantages of the method.