篇名 | Research on Noise Reduction Method Based on CEEMD-WT-SVD and Its Application in Acoustic Signal of Pipeline Blockage |
---|---|
卷期 | 30:2 |
作者 | Jing-zong Yang 、 Zao Feng 、 Xiao-dong Wang 、 Guo-yong Huang |
頁次 | 224-239 |
關鍵字 | complete ensemble empirical mode decomposition 、 noise reduction 、 singular value decomposition 、 wavelet transform 、 EI 、 MEDLINE 、 Scopus |
出刊日期 | 201904 |
DOI | 10.3966/199115992019043002021 |
Signal denoising is one of the most important tasks in the application of acoustic detection for pipeline blockage. Aiming at the effects of random noise and impulse noise on acoustic signal from pipeline blockage, a noise reduction method based on complete ensemble empirical mode decomposition (CEEMD), wavelet transform (WT) and singular value decomposition (SVD) is proposed. First, the continuous mean square error criterion is introduced to judge the dominant high-frequency intrinsic mode component obtained by CEEMD. Then wavelet soft threshold is used to denoise and reconstruct the signal, which avoids the loss of high frequency useful signals. Secondly, to achieve the further purpose of suppressing noise, the phase space reconstruction and SVD of the signal are carried out, and the reconstruction order of the signal is determined by using the larger peak position of the singular value energy difference spectrum. Through the analysis of the noise reduction effect of simulation signal and pipeline blockage signal, the results show that this method not only can improve the problem of the modal mixture well, but also can effectively extract the useful features of the signal. Meanwhile, the noise reduction performance is better than the noise reduction method based on ensemble empirical mode decomposition.