基于PCA和LMD分解的滚动轴承故障特征提取方法

Feature Extraction Method of Rolling Bearing Fault Based on Principal Component Analysis and Local Mean Decomposition

局部均值分解(LMD)是一种自适应时频分析方法,并在轴承的故障诊断中成功应用,但是受噪声的影响比较大。为了最大程度地降低噪声的干扰,提出了主分量分析(PCA)与局部均值分解(LMD)相结合的故障诊断方法。该方法首先利用相空间重构将一维时间序列振动信号嵌入为等效的多维时间序列信号,然后利用主分量分析提取主要成分实现降噪,最后把降噪之后的信号进行LMD分解,分解成若干个乘积函数(PF)之和,对能量最高的PF1进行包络谱分析,提取出故障特征信息。通过仿真试验和轴承故障试验,结果表明该方法能够有效地提取出信号的故障特征,证明了该方法的有效性。

Local mean decomposition（LMD） is an adaptive time-frequency analysis method,which is successfully used in roll- ing bearing fault diagnosis but strongly influenced by noise.In order to reduce the noise interference to the greatest extent, a fault di- agnosis method based on principal component analysis （PCA） and local mean decomposition（LMD） was proposed.Firstly, one-di- mension time series vibration signals were embedded to equivalent multi-dimensions through the reconstructed phase space.Then, a- vailable component was extracted by PCA ,achieving the effect of noise reduction.Finally,the de-noised signal was decomposed by LMD and can be represented as the sum of the product functsons.The component PF1 which contains the highest power was select- ed to conduct the envelope spectrum analysis and the fault features were exacted.The results show that the method can effectively extract the fault features through the analysis of the simulation signal and the roiling bearing fault diagnosis data experiment, pro- ving the effectiveness of the proposed method.