形态分量分析结合再分配尺度谱的碰摩特征提取

Rubbing Feature Extraction Using Morphological Component Analysis and Reassigned Scale Spectrum

针对转子系统碰摩故障问题,从信号的形态结构角度出发,分析了碰摩信号组成成分的形态差异性,将用于图像处理的形态分量分析方法引入到碰摩信号组分分离中,并与小波再分配尺度谱相结合,提出了形态分量分析结合再分配尺度谱的转子碰摩故障特征提取方法。利用形态分量分析方法对原信号进行预处理,将原信号分解成光滑分量和细节分量两个部分。对细节分量做进一步再分配尺度谱分析,从而提取出与故障相关的时频特征信息。通过仿真信号及实测信号进行验证,分析结果表明该方法可有效应用于转子碰摩故障诊断,具有一定的可行性。

Aiming at solving the problem of rubbing fault of rotor system,the morphological differences of the components of the rubbing signal were analyzed from the perspective of morphological structure of the signal,the morphological component analysis method which is used for processing images was introduced to separate the components of the rubbing signal and combined with reassigned wavelet scalogram,a feature extraction method for rubbing fault of rotor combined morphological component analysis and reassigned scalogram was proposed. Firstly,the original signal was processed by the morphological component analysis method and decomposed in two parts which are smooth component and detail component,then the detail component was analyzed by using the reassigned scalogram,and the time-frequency characteristic information related to the fault was acquired. Simulated signal and measured signal were used for examining the effectiveness of the method and the results show this method can be applied to diagnose the rubbing fault of rotor.