李晋^1,2,,
张贤^1,2,
蔡锦^1,2
1. 湖南师范大学智能计算与语言信息处理湖南省重点实验室, 长沙 410081
2. 湖南师范大学信息科学与工程学院, 长沙 410081

基金项目: 国家高技术研究发展计划（863计划）重大项目"深部矿产资源勘探技术"第三课题（2014AA06A602）、国家重点研发计划"深地资源勘查开采"重点专项（2018YFC0603202）、国家自然科学基金（41404111）、湖南省自然科学基金（2018JJ2258）和湖南省研究生科研创新项目（CX20190414）联合资助


详细信息

作者简介: 李晋, 男, 1981年生, 博士, 副教授, 主要从事矿集区大地电磁强干扰压制及信噪辨识研究.E-mail:geologylj@163.com

中图分类号: P631

收稿日期:2018-07-05
修回日期:2019-02-18
上线日期:2019-10-05

Suppression of strong interference for AMT using VMD and MP

LI Jin^1,2,,
ZHANG Xian^1,2,
CAI Jin^1,2
1. Hunan Provincial Key Laboratory of Intelligent Computing and Language Information Processing, Hunan Normal University, Changsha 410081, China
2. College of Information Science and Engineering, Hunan Normal University, Changsha 410081, China


MSC: P631

--> Received Date: 05 July 2018
Revised Date: 18 February 2019
Available Online: 05 October 2019


摘要
摘要:为了有效分离矿集区音频大地电磁（AMT）信号中的大尺度强干扰、抑制近源效应，本文提出利用变分模态分解（VMD）和匹配追踪（MP）联合压制AMT强干扰的方法.首先，对比了VMD与经验模态分解（EMD）、固有时间尺度分解（ITD）的处理效果，验证了VMD在避免模态混叠和端点效应方面的优势；讨论了VMD中模态个数对典型大尺度强干扰的去噪性能，并选择合适的模态初步获取待处理信号的重构信息.然后，运用MP对VMD重构信号做二次信噪分离处理，进一步滤除残余的尖脉冲干扰.通过对模拟和实测数据的分析处理，以及与远参考法结果对比，本研究能有效剔除时间域序列中的大尺度强干扰，且重构信号中保留了更多的低频缓变化信息和细节成分，近源干扰得到有效压制；视电阻率-相位曲线更为光滑、连续，低频段的数据质量得到明显改善，其结果能更为真实、可靠地反映地下电性结构信息.
关键词: 音频大地电磁/
强干扰/
噪声压制/
变分模态分解/
匹配追踪

Abstract:In order to effectively separate the large-scale strong interference in the audio magnetotelluric (AMT) signal and suppress the near-source effect, this paper proposes a method of suppressing strong AMT by using variational mode decomposition (VMD) and matching pursuit (MP). First, we compare the performance of signal decomposition with empirical mode decomposition (EMD) and intrinsic time scale decomposition (ITD), and verify the advantage of VMD algorithm in avoiding model aliasing and endpoint effects. At the same time, we discuss the de-noising performance of the typical large-scale strong interference in a model number of VMD, and the appropriate model is selected to obtain the reconstructed information of the processing signal. Then, the reconstructed signal is further processed by MP, which filters the residual sharp pulse interference. Through analyzing and processing the simulated and measured data, and comparing with the remote reference method, the experiment results show that the proposed method can effectively eliminate the large-scale strong interference in the time domain sequence, and retain more low-frequency of slow change information and detail components. The near-source interference is effectively suppressed, and the apparent resistivity-phase curve is more smooth and continuous. Moreover, the data quality of the low-frequency band is improved obviously and the result can reflect the underground electrical structure information more truly and reliably.
Key words:Audio magnetotelluric/
Strong interference/
Noise suppression/
Variational mode decomposition/
Matching pursuit



PDF全文下载地址:

http://www.geophy.cn/data/article/export-pdf?id=dqwlxb_15195