改进CEEMDAN-小波包阈值降噪在旋转机械轴承中的应用

Improved CEEMDAN-wavelet packet thresholding for noise reduction in rotating machinery bearings

针对旋转机械滚动轴承振动加速度信号在采集过程中存在大量噪声导致振动信号表征不明显的问题,提出了一种在对自适应噪声完备集合经验模态分解(CEEMDAN)进行改进的基础上与小波包阈值降噪联合的一种数据去噪方法。首先,通过计算分解后得到的各分量的均方根和互相关系数对与原始信号相关性较大的分量进行提取;然后,分别计算提取后每个分量对应的阈值进行小波包阈值降噪处理;最后,将处理后的信号与分量中不存在噪声的分量进行叠加,即为降噪后的信号。经过对比实验验证,改进CEEMDAN-小波包阈值降噪方法的信噪比最大提高6.41,均方根误差降低0.12,证明了本方法的有效性。

Aiming at the problem of vibration acceleration signals of rolling bearings in rotating machinery are not characterised by a large amount of noise in the acquisition process,data denoising method based on the improvement of the empirical modal decomposition of adaptive noise-complete sets(CEEMDAN)combined with wavelet packet thresholding noise reduction is proposed.Firstly,the components that are more correlated with the original signal are extracted by calculating the root mean square and the number of interrelationships of each component obtained after decomposition;then,the threshold value corresponding to each component after extraction is calculated separately for wavelet packet threshold noise reduction;finally,the processed signal is superimposed with the component that does not have noise,which is the noise reduced signal.After comparison and verification,the signal-to-noise ratio of the improved CEEMDAN-wavelet packet thresholding noise reduction method is improved by 6.41 and the root-mean-square error is reduced by 0.12,which proves the effectiveness of the method in this paper.