EMD与cICA方法在多级齿轮传动微弱故障特征提取中的应用

摘要

为提取多级齿轮传动单通道测量信号中隐含的微弱低频故障特征信息,提出了一种基于经验模态分解(Empirical mode decomposition,EMD)与约束独立分量分析(Constrained independent component analysis,cICA)相结合的故障特征提取方法.首先对实测的齿轮箱单通道测量信号进行EMD分解;然后计算各个本征模态函数(Intrinsic mode function,IMF)的峭度及其与原信号的互相关系数,并选择合适的IMFs分量与原信号组成新的虚拟观测向量;最后,通过构建合适的参考信号进行cICA分析,提取出了理想的微弱低频故障特征.通过多级齿轮传动中的低速级断齿故障特征提取试验分析,验证了该方法的有效性和适用性.%In order to extract the weak and low-frequency fault feature hidden in the single-channel measured signal from multi-stage gear transmissions,a joint approach of fault feature extraction based on empirical mode decomposition (EMD) and constrained independent component analysis (cICA) is proposed in this paper.Firstly,the single-channel measured signal is decomposed into several IMFs with EMD.Then,the kurtosis and crosscorrelation coefficient of each IMF are computed,and the suitable IMFs for constructing the new measured virtual vector are selected.Finally,the proper reference signal including gear fault feature frequency is constructed,and the desired low-frequency slight feature is extracted with cICA method.Through the experiment analysis of fault feature extraction on the low-speed gears with a missing tooth,the effectiveness and applicability of the proposed method is verified.