基于EEMD的滚动轴承故障诊断方法

Rolling element bearing feature extraction based on EEMD

针对经验模态分解(empirical mode decomposition,EMD)方法处理轴承振动信号时存在的缺点,指出极值点的选择是产生模态混叠现象的原因。分析了集成经验模态分解(ensemble empirical mode decomposition,EEMD)方法抑制模态混叠现象的原理,讨论了加入高斯白噪声的次数和大小对EEMD方法分解结果的影响,并通过仿真和实测信号对EMD和EEMD方法的性能进行了比较测试。结果表明:EEMD方法不仅能够有效地抑制模态混叠现象,而且能更好地反映出轴承振动信号中的故障信息。同时,对通过EEMD方法得到的重构信号进行平方包络分析,结果证明:该方法能够有效地提取出滚动轴承的故障特征。

Aiming at the mode mixing problem in the sifting process of EMD, this paper points out the cause of mode mixing in respect of the variation of extremum series, analyzes why EEMD method can effectively suppress the mode mixing, and discusses the effect of the number of times and level of adding Gaussian white noise to each iteration. The comparison of the performances of EMD and EEMD methods through simulation and experiment signals shows that EEMD method cannot only suppress the mode mixing but also reflect the fault information of rolling bearing vibration signals better than EMD method. In the mean time, the reconstructed signal is obtained by choosing the appro- priate IMFs and then it is demodulated by square envelope. The rolling bearing＇s characteristic fault frequency is identified by the enveloped normalized amplitude-frequency spectrum. The results show that EEMD method combined with the envelope analysis can identify the fault features of rolling bearing effectively.