Siamese network based noise elimination of artificial seismic data recorded by distributed fiber-optic acoustic sensing
文献类型:期刊论文
作者 | Shao Jie1,3; Wang YiBo1,3; Liang Xing4; Xue QingFeng1,3; Liang EnMao2; Shi ShuYou4 |
刊名 | CHINESE JOURNAL OF GEOPHYSICS-CHINESE EDITION |
出版日期 | 2022-09-01 |
卷号 | 65期号:9页码:3599-3609 |
ISSN号 | 0001-5733 |
关键词 | Siamese network Self-supervised deep learning Distributed fiber-optic acoustic sensing Random noise suppression |
DOI | 10.6038/cjg2022P0919 |
英文摘要 | The signal-to-noise ratio (SNR) of field distributed fiber-optic acoustic sensing (DAS) data is usually relatively low due to the poor coupling and many other factors. Therefore, it is necessary to denoise the field DAS seismic data before the subsequent processing. The traditional artificial intelligence supervised learning-based denoising method can effectively eliminate noise in the DAS data. However, it needs to label the data using large groups of noisy and noise-free data, which turns out to be a huge effort. This paper proposes a noise suppression method for DAS seismic data using a self-supervised siamese network. Based on the theory of siamese network in self-supervised learning, the network structure is built using a U-Net architecture. The optimal solution of the denoising objective function is obtained iteratively by learning characteristics of the input noisy data from itself. Then a self-supervised high-precision deep learning-based denoising network is established. Results of synthetic data and field data show that the proposed method can effectively suppress the random noise of artificial seismic data recorded by DAS, and improve the SNR and continuity of seismic events. In addition, the proposed method avoids manual labeling in the traditional supervised learning methods and significantly improves the efficiency of seismic data denoising. |
WOS关键词 | SUPPRESSION |
WOS研究方向 | Geochemistry & Geophysics |
语种 | 英语 |
出版者 | SCIENCE PRESS |
WOS记录号 | WOS:000851307600024 |
源URL | [http://ir.iggcas.ac.cn/handle/132A11/108353] |
专题 | 地质与地球物理研究所_中国科学院油气资源研究重点实验室 |
通讯作者 | Wang YiBo |
作者单位 | 1.Chinese Acad Sci, Inst Geol & Geophys, Beijing 100029, Peoples R China 2.Beijing Global Safety Technol Co Ltd, Beijing 100094, Peoples R China 3.Chinese Acad Sci, Innovat Acad Earth Sci, Beijing 100029, Peoples R China 4.China Natl Petr Corp, Zhejiang Oilfield Branch, Hangzhou 311100, Peoples R China |
推荐引用方式 GB/T 7714 | Shao Jie,Wang YiBo,Liang Xing,et al. Siamese network based noise elimination of artificial seismic data recorded by distributed fiber-optic acoustic sensing[J]. CHINESE JOURNAL OF GEOPHYSICS-CHINESE EDITION,2022,65(9):3599-3609. |
APA | Shao Jie,Wang YiBo,Liang Xing,Xue QingFeng,Liang EnMao,&Shi ShuYou.(2022).Siamese network based noise elimination of artificial seismic data recorded by distributed fiber-optic acoustic sensing.CHINESE JOURNAL OF GEOPHYSICS-CHINESE EDITION,65(9),3599-3609. |
MLA | Shao Jie,et al."Siamese network based noise elimination of artificial seismic data recorded by distributed fiber-optic acoustic sensing".CHINESE JOURNAL OF GEOPHYSICS-CHINESE EDITION 65.9(2022):3599-3609. |
入库方式: OAI收割
来源:地质与地球物理研究所
浏览0
下载0
收藏0
其他版本
除非特别说明,本系统中所有内容都受版权保护,并保留所有权利。