利用经验模态分解改进的CEEMD故障诊断方法

Improved CEEMD fault diagnosis method by using empirical mode decomposition

针对互补集合经验模态分解(CEEMD)方法在分解过程中会产生模态分裂的现象,提出了一种利用经验模态分解改进的CEEMD方法。由于经传统CEEMD方法分解得到的IMF分量并不能满足IMF分量的严格定义,将这些分量定义为预分解IMF分量,然后利用经验模态分解对这些预分解IMF分量重新分解,得到正确的IMF分量。为了验证改进CEEMD方法的有效性,将它用于仿真信号分解中。仿真结果表明,该方法可以有效消除传统CEEMD方法出现的模态分裂现象,分解结果更符合实际情况。将改进的CEEMD方法对真实轴承故障信号进行分解,结合包络谱分析,可以准确提取故障特征频率,从而实现对轴承故障的有效诊断。

In view of the modal splitting existing during decomposition process of the complementary ensemble empirical mode decomposition（CEEMD） method, an improved CEEMD method improved by empirical mode decomposition was presented. Since the IMF components obtained by traditional CEEMD method fails to satisfy the strict definition of IMF components, the components were tentatively denominated as pre-decomposed IMF components. And then, the empirical mode decomposition method was applied to re-decompose these predecomposed IMF components, thus the correct IMF components were obtained. In order to check the effectiveness of the improved CEEMD method, it was applied to the decomposition of a simulated signal. The simulation results showed the method effectively eliminated the modal splitting of traditional CEEMD method, and the decomposition results were more reasonable. In addition, the improved CEEMD method was applied to decompose the real bearing faulty signal, and the characteristic frequency of the fault was accurately extracted in combination with envelope spectrum analysis, thus the bearing faults were effectively identified.