变分框架下多尺度熵相关优化的模态分解在故障诊断中的应用

Application of multi-scale entropy correlation optimization to mode decomposition in fault diagnosis under variational framework

针对变分框架下,一种新的模态分解——变分模态分解(Variational Mode Decomposition,VMD)的最优模态分量选择和关键参数辨识问题,借鉴折半查找的思想,提出应用多尺度熵相关系数和频域相关系数来改进VMD的上述关键环节,并通过轴承故障信号仿真研究其频域分解的数据特点,揭示其滤波本质;轴承故障信号仿真及工程应用的结果表明,相对于经验模态分解(Empirical Mode Decomposition,EMD)和总体平均经验模态分解(Ensemble Empirical Mode Decomposition,EEMD),改进后的VMD(IVMD)去噪效果更为明显,是一种有效的自适应频域模态分解方法,可更为准确地提取出微弱特征频率信息,实现轴承故障的正确识别。

According to optimal mode selection and key parameter identification of a new adaptive mode decomposition under variational framework Variational Mode Decomposition （ VMD ）, from the idea of binary search, the multi-scale entropy correlation and correlation coefficient in Fourier domain are presented to solve the problem above for VMD, and its filtering essence is revealed through decomposition characteristics of bearing fault simulation signal in Fourier domain. With analysis for simulation signal and engineering application of bearing fault, the results show that, compared with Empirical Mode Decomposition（EMD） and Ensemble Empirical Mode Decomposition（EEMD） , de-noising effect of the Improved VMD （IVMD） is more obvious ,which is an effective adaptive mode decomposition method in Fourier domain, and can extract the weak feature frequency of fault signal more accurately, achieve correct recognition of bearing fault.