广义变分模态分解方法及其在变工况齿轮故障诊断中的应用

Generalized variational mode decomposition and its applications to gearbox fault diagnosis under variable conditions

变分模态分解(Variational Mode Decomposition,VMD)是近年来提出的非平稳信号分解方法,通过将信号分解问题转化为变分约束问题,从而实现多变量信号的模态分离。但VMD方法在分析时变多分量信号时存在模态混叠现象。对此,提出了一种适合分析时变模态的信号处理方法——广义变分模态分解(Generalized VMD,GVMD)。通过分析仿真信号,将GVMD与小波变换,原VMD和希尔伯特黄变换等方法进行了对比,结果表明,新提出的GVMD方法分解结果更精确,时频分辨率更高。最后,将GVMD方法应用于变转速齿轮振动信号故障特征的识别,结果表明了论文方法的有效性。

The variational mode decomposition （VMD） is a recently proposed non-stationary signal analysis method. However, the mode mixing will occur when the VMD is used to analyze the time-varying multi-component signal. In this paper, a new signal decomposition method called generalized variational mode decomposition （GVMD） is proposed for analyzing the time- varying multi-component signal. Also the GVMD method is compared with the continuous wavelet transform method and Hilbert- Huang transform by analyzing the simulation signal. The results show that the decomposition of GVMD is more accurate and having higher time-frequency resolution. Finally, the proposed method is applied to identify the time-varying fault identification under variable working conditions from the gear vibration signals and the analysis results verified the effectiveness of the proposed method.