摘要
针对现场实测声发射信号经常淹没在噪声中的问题,提出了一种基于字典学习的碰摩声发射信号降噪算法。首先在训练阶段采用干净碰摩声发射信号利用K-SVD算法训练自适应字典,然后对含噪碰摩声发射信号的噪声方差进行估计,最后利用正交匹配追踪算法对含噪碰摩声发射信号在训练后的字典上进行稀疏分解,从而达到对碰摩声发射信号进行降噪的目的。实验结果表明:基于字典学习的算法对碰摩声发射信号能取得较好的降噪效果,相比与基于固定字典的传统算法能够获得更高的信噪比。
Aiming at the problem of the collected in-site acoustic emission (AE) signals usually covered up by noises,a denoising algorithm based on dictionary learning is propose for the rubbing AE signal. Firstly,it adopts clean rubbing AE signal using the k-means singular value decomposition (K-SVD) algorithm to train adaptive dictionary.Then,the noise variance of the noisy rubbing AE signal is estimated. Finally,based on the trained dictionary,the orthogonal matching pursuit algorithm is utilized to decompose the noisy rubbing AE signal into sparse representation so as to achieve the purpose of the rubbing AE signal noise reduction. Experimental results show that the algorithm based on dictionary learning can achieve better noise reduction effect on rubbing AE signal,and higher SNR than the traditional algorithm based on fixed dictionary.
作者
彭威
张祺威
PENG Wei;ZHANG Qiwei(School of Information and Electric Engineering,China University of Mining and Technology,Xuzhou Jiangsu 221008,China;School of Information Science and Engineering,Southeast University,Nanjing 210096,China)
出处
《电子器件》
CAS
北大核心
2019年第1期116-119,共4页
Chinese Journal of Electron Devices
基金
国家自然科学基金(61375028)
关键词
声发射
降噪
K-SVD
正交匹配追踪
acoustic emission
denoising
K-SVD
orthogonal matching pursuit
