基于SVD-AVMD的液膜密封声发射特征提取  

Feature extraction of acoustic emission of liquid film sealing based on SVD-AVMD

作　　者：孙鑫晖[1] 刘怀顺 王明洋 李勇凡 王增丽[1] 郝木明[1] 力宁 任宝杰 SUN Xinhui;LIU Huaishun;WANG Mingyang;LI Yongfan;WANG Zengli;HAO Muming;LI Ning;REN Baojie(College of New Energy in China University of Petroleum(East China),Qingdao 266580,China;AECC Hunan Aviation Powerplant Research Institute,Zhuzhou 412002,China;Dongying Hiscien Sealing Technology Company Limited,Dongying 257000,China)

机构地区：[1]中国石油大学(华东)新能源学院,山东青岛266580 [2]中国航发湖南动力机械研究所,湖南株洲412002 [3]东营海森密封技术有限公司,山东东营257000

出　　处：《中国石油大学学报（自然科学版）》2021年第6期168-174,共7页Journal of China University of Petroleum(Edition of Natural Science)

基　　金：国家重点研发计划项目(2020YFB2010002);国家自然科学基金项目(51975585);中国航发湖南动力机械研究所实验室开放基金项目(KY-1003-2020-0040)。

摘　　要：将声发射技术应用于液膜密封端面状态监测时,声发射信号易受噪声影响、特征信号难以提取,为此提出一种基于奇异值分解和自适应变分模态分解(SVD-AVMD)的信号处理方法。该方法首先以奇异值分解消除信号中的随机强噪声影响,获取降噪信号,然后在不同模态数下对降噪信号作变分模态分解,并计算各模态分量与降噪信号之间的显著性水平,以显著性水平大于阈值作为分解的停止准则,最终达到获取最优模态分量的目的。结果表明:SVD-AVMD对各模态分量中心频率的捕捉能力以及对各模态分量的恢复效果均明显优于单纯的变分模态分解,能够滤除背景噪声的同时最大程度地保留有效信息;获得表征液膜密封端面状态的声发射信号,实现了对液膜密封端面状态的识别。When the acoustic emission technology is applied to the condition monitoring of the sealing surface of the liquid film,the acoustic emission signal is easy to be affected by noises,and the characteristic signal is difficult to extract.Therefore,a signal processing method based on the singular value decompositions and adaptive variational mode decomposition(SVD-AVMD)was proposed.Firstly,the singular value decomposition was used to eliminate the influence of random strong noises in the signal.Then,the denoised signal was decomposed by the variational mode decomposition under different modal numbers,and the significance level among modal components and the denoised signal was calculated.The stop criterion of decomposition was determined that the significance level was greater than the threshold value,for the purpose of obtaining the optimal modal component.The results show that the ability of SVD-AVMD to capture the center frequency and the recovery effect of each mode component are significantly better than that of the simple variational mode decomposition,which can filter out the background noise and retain the effective information to the greatest extent.The acoustic emission signal representing the state of the liquid film s sealing surface was obtained,and the recognition of the end state of the liquid film sealing was realized.

关 键 词：声发射 液膜密封 奇异值分解 自适应变分模态分解 显著性水平 

分 类 号：TH136[机械工程—机械制造及自动化] TN911.7[电子电信—通信与信息系统]