基于改进VMD与包络导数能量算子的滚动轴承早期故障诊断

Rolling bearing early fault diagnosis based on improved VMD and envelope derivative operator

针对滚动轴承早期故障比较微弱,特征信息难以提取且变分模态分解(VMD)中分解层数k的大小需要使用者反复尝试而不能有效确定的问题,提出了改进的VMD方法,以能量差作为评价参数自适应地确定分解层数k。在此基础上,将改进的VMD与包络导数能量算子结合,提出了VMD与包络导数能量算子的轴承早期故障诊断方法。采用VMD对轴承故障振动信号进行分解,根据能量差曲线确定最佳的分解层数k;依据峭度准则,从分解得到的k个本征模态分量中选取敏感分量进行重构;并用包络导数能量算子对重构信号进行解调分析,从其能量谱中便可准确地提取轴承的故障特征信息。通过仿真信号和实验数据的分析,验证了该方法的有效性与可行性。

Aiming at rolling bearings ＇ early fault features being weaker and difficult to extract, and the decomposition layer number k in VMD being too difficult to determine,the improved VMD method was proposed. Energy difference was taken as an evaluation parameter to adaptively determine the decomposing layer number k. Then,the improved VMD was combined with the envelope derivative operator,a rolling bearing early fault feature diagnosis method was proposed. Firstly,a rolling bearing＇s original fault vibration signal was decomposed with the VMD. According to the energy difference curve,the optimal value of k was determined. Secondly,according to the kurtosis criterion,sensitive components were selected from k IMFs obtained with decomposition to reconstruct a signal. The reconstructed signal was demodulated and analyzed with the envelope derivative operator. The rolling bearing ＇s fault feature information was extracted correctly from the energy spectrum of the reconstructed signal. Through analyzing simulated signals and test data,the validity and feasibility of the proposed method was verified.