基于改进小波包阈值降噪的滚动轴承故障分析

具有非平稳特性的滚动轴承振动信号易受到外界噪声干扰，且传统的小波包硬、软阈值函数降噪方法无法根据信号中的噪声干扰情况自适应调节。因此，提出一种基于排列熵的改进小波包阈值降噪方法，并与自适应噪声的完整集成经验模态分解(CEEMDAN)相结合进行故障信号分析。首先，对采集的滚动轴承故障信号进行改进小波包阈值降噪处理，然后将降噪信号进行CEEMDAN处理，分解得到一系列固有模态分量(IMF)，根据相关系数选择IMF，并作包络谱分析。最后对滚动轴承实际振动信号的故障分析，证明了此方法的有效性。

Fault Analysis of Rolling Bearing Based on Improved Wavelet Packet Threshold Denoising

The vibration signals of rolling bearings with non-stationary characteristics are susceptible to external noise interference, and the traditional wavelet packet denoising methods based on hard and soft threshold functions can not be adjusted according to the noise interference in the signals. Therefore, an improved wavelet packet threshold denoising method based on permutation entropy is proposed in this paper, which is combined with Complete Ensemble Empirical Mode Decomposition with Adaptive Noise(CEEMDAN) for fault signal analysis. Firstly, the fault signal of rolling bearing is processed by Improved Wavelet Packet Threshold denoising, then the noise signal is decomposed into several intrinsic mode functions (IMF) by CEEMDAN, and the IMF are selected according to the correlation coefficients in combination with envelope spectrum analysis. Finally, the analysis of the actual vibration signal of rolling bearing proves that the method is effective.