【摘要】 针对齿轮故障信号易受噪声干扰导致故障特征难以提取的问题,提出一种基于变分模态分解(variational mode decomposition,VMD)和最大重叠离散小波包变换(maximal overlap discrete wavelet packet transform,MODWPT)相结合的信号去噪方法。采用VMD方法将齿轮振动信号分解成一系列不同中心频率的固有模态函数(intrinsic mode function,IMF),对VMD分解过程中影响其精度的主要参数选择方法进行了探究,提出相关参数的选取依据。结合能量熵增量-频域互相关系数准则以剔除分解出的高频噪声和虚假干扰成分;采用MODWPT方法对包含高频噪声的IMF分量进行去噪,以进一步提升信号的去噪效果和性能指标;最后将去噪后高频IMF分量同表征信号自身特征的敏感模态分量重构为去噪信号。通过仿真信号和齿轮断齿故障信号的分析,证明了所提方法的有效性和实用性。更多还原

【Abstract】 Aiming at the problem that gear vibration signal is easily affected by noise and it is difficult to extract the fault feature of it, a method for gear vibration signal de-noising based on variational mode decomposition(VMD) and maximal overlap discrete wavelet packet transform(MODWPT) was proposed. Firstly, VMD was used to decompose the gear vibration signal into a number of intrinsic mode functions(IMFs) in different center frequency scales. For this method, the parameter selection that affects the accuracy of VMD decomposition has been deeply studied, and the solution to this problem was given. Then, a joint de-noising algorithm based on the criterion of energy entropy increment and frequency domain cross correlation coefficient was used to eliminate high frequency noise components and false components, in order to improve the de-noising effect and performance index. Finally, the high frequency noise components were decomposed by MODWPT, the high frequency IMF component after de-noising and the IMF components representing the characteristics of the signal itself reconstructed the de-noising signal. This method was applied in fault diagnosis of simulation signal and measured gear breakage fault signal. The results proved the effectiveness and practicality of the proposed method.更多还原