基于固有时间尺度分解与最大相关峭度解卷积的轴承故障特征提取

Feature Extraction of Rotating Bearing Fault Based on Intrinsic Time-scale Decomposition and Maximum Correlated Kurtosis Deconvolution

针对滚动轴承故障信号的非平稳性、非线性及复杂性特征以及在故障识别过程中存在噪声干扰、故障特征不清晰的问题,提出一种基于固有时间尺度分解(Intrinsic Time-scale Decomposition,ITD)与最大相关峭度解卷积(Maximum Correlated Kurtosis Deconvolution,MCKD)结合的轴承故障特征提取方法。利用固有时间尺度分解对故障信号进行分解,降低信号分解中的模态混叠;对分解的固有旋转分量进行选择,提取故障信号的有用成分,实现信号降噪;采用粒子群优化的MCKD提取故障特征信号的冲击成分。实验结果表明,该方法可以降低信号的模态混叠问题,增强故障特征,有利于噪声环境下轴承故障特征的提取。

Aiming at the non-stationarity,nonlinearity and complexity characteristics of rolling bearing fault signals,as well as the problems of noise interference and unclear fault characteristics in the process of fault identification,this paper proposed a feature extraction method for bearing faults combined with Intrinsic Time-scale Decomposition(ITD)and Maximum Correlated Kurtosis Deconvolution(MCKD).First,Intrinsic Time-scale Decomposition was used to decompose the fault signal to reduce the modal aliasing in the signal decomposition;then the intrinsic rotation component of the decomposition was selected to extract the useful components of the fault signal to achieve signal noise reduction;finally,particle swarm optimization MCKD was used to extract the impact components of fault characteristic signals.Experimental results show that this method can reduce the signal modal aliasing problem,enhance the fault characteristics,and is beneficial to the extraction of bearing fault characteristics in a noisy environment.