VMD和ICA联合降噪方法在轴承故障诊断中的应用

Application of VMD-ICA combined method in fault diagnosis of rolling bearings

针对振动信号易受噪声干扰的影响、故障特征提取困难的问题,提出一种基于变分模态分解(Variational Mode Decomposition,VMD)和独立分量分析(Independent Component Analysis,ICA)相结合的去噪方法。该方法首先利用VMD算法将振动信号分解成若干不同频率的本征模态分量(Intrinsic Mode Function,IMF),有效的抑制了LMD分解中存在的模态混叠现象和端点效应等问题,然后依据峭度准则选取相应分量进行重构,引入虚拟噪声通道;最后利用Fast ICA将重构后信号再次进行去噪处理,分离出有效的故障特征分量,从而识别故障类型。将该方法应用到滚动轴承故障数据中,并与LMD-ICA方法作对比,结果表明,提出方法不仅能够有效的解决去噪过程中丢失故障信息以及由于模态混叠导致噪声不能完全去除的问题,还能更清晰、准确地提取出故障特征频率。

In order to solve the problem of fault features of rolling bearings being difficult to extract, the signal analysis method of variational mode decomposition （ VMD ） combined with independent component analysis （ ICA ） was proposed here. At first, VMD was used to decompose a multi-component vibration signal into a number of quasi-orthogonal intrinsic mode functions （IMFs） to effectively suppress problems, such as, mode mixing, end effect and so on existing in LMD algorithm. Then the kurtosis criterion was used to choose the corresponding IMFs to be reconstructed, and to induce a virtual noise channel. Finally, the reconstructed signal was denoised again with the fastICA, the effective fault feature components were extracted and the fault types were identified. In order to verify the effectiveness of the proposed method, this method was applied in fault diagnosis of rolling bearings, its effect was compared with that of LMD-ICA algorithm. The results demonstrated that the proposed method can not only solve problems of losing fault information and not enough eliminating noise due to mode mixing in denoising process, but also extract fault feature frequencies more clearly and correctly.