王琦^1,,
田宝凤^2,3,
张健^2,3,
蒋川东^2,3,,
1. 吉林大学通信工程学院, 长春 130012
2. 地球信息探测仪器教育部重点实验室(吉林大学), 长春 130061
3. 吉林大学仪器科学与电气工程学院, 长春 130061

基金项目: 国家自然科学基金青年科学基金（41704103，41604083）资助


详细信息

作者简介: 王琦, 女, 1988年生, 吉林大学工程师.主要从事磁共振及瞬变电磁数据处理方法研究.E-mail:wangqi0115@jlu.edu.cn

通讯作者: 蒋川东, 男, 1984年生, 吉林大学副教授.主要从事地球物理探测仪器及数据处理方法.E-mail:chuandongjiang@gmail.com

中图分类号: P631

收稿日期:2017-09-16
修回日期:2017-12-27
上线日期:2018-08-05

Surface nuclear magnetic resonance signal extraction based on the sparse representation

WANG Qi^1,,
TIAN BaoFeng^2,3,
ZHANG Jian^2,3,
JIANG ChuanDong^2,3,,
1. College of Communication Engineering, Jilin University, Changchun 130012, China
2. Key Laboratory of Geophysical Exploration Equipment, Ministry of Education(Jilin University), Changchun 130061, China
3. College of Instrumentation & Electrical Engineering, Jilin University, Changchun 130061, China


More Information

Corresponding author: JIANG ChuanDong,E-mail:chuandongjiang@gmail.com

MSC: P631

--> Received Date: 16 September 2017
Revised Date: 27 December 2017
Available Online: 05 August 2018


摘要
摘要:在多孔隙含水层中地面磁共振（surface nuclear magnetic resonance，SNMR）信号呈现多弛豫衰减特性，常规盲源分离方法和单指数拟合方法引起信号严重失真和信息缺失等问题.本文提出了基于稀疏表示的随机噪声背景下多弛豫SNMR信号的提取方法.根据SNMR信号的衰减特征，设计了精确刻画SNMR信号且与随机噪声不相关的离散衰减余弦冗余字典.其次，针对多弛豫SNMR信号稀疏度未知的问题，通过设置合理的残差比阈值控制迭代次数，改进了广义正交匹配追踪（generalized orthogonal matching pursuit，gOMP）算法，使得该方法应用于SNMR信号的提取时，具有更好的自适应性和普适性.再次，鉴于SNMR测量数据为多次独立重复采集的结果，提出了基于数据流的SNMR信号提取策略，在提高算法鲁棒性的同时，保证了信号提取结果的唯一性.最后，通过仿真和实测数据证明了基于gOMP算法的稀疏表示方法可以显著地提升多弛豫SNMR信号的提取质量，降低随机噪声对含水层反演结果的影响，提高SNMR探测能力.
关键词: 核磁共振/
稀疏表示/
广义正交匹配追踪/
冗余字典

Abstract:As the surface nuclear magnetic resonance signal (SNMR) is a multi-exponential decay wave in multi-porous aquifer, there are serious problems of signal distortion and information loss by conventional blind source signal extraction and mono-exponential fitting method. In this paper, we propose a method to extract multi-exponential SNMR signals from random noise background using the sparse representation based on generalized orthogonal matching pursuit (gOMP) algorithm. Firstly, the sparse decomposition and reconstruction model of multi-exponential SNMR signal is established by sparse representation theory. According to the known decay characteristics of SNMR signal, a discrete decay cosine (DDC) redundancy dictionary which can describe SNMR signal and not related to random noise is designed, and the influence of discretization scheme on reconstruction result is analyzed. Secondly, aiming at the problem that the sparseness of multi-exponential SNMR signal is unknown, the gOMP algorithm is improved by setting a reasonable number of iterative ratio thresholds, which makes the method has better adaptability and universality when applying to the SNMR signal extraction in random noise environment. Since the SNMR data is a series of repeated independent acquisition, a SNMR signal extraction strategy based on data flow is proposed in order to improve the robustness of the algorithm and ensure the uniqueness of the results. Finally, the field data results indicate that the sparse representation method based on gOMP algorithm can be applied to the signal extraction to improve the quality of multi-exponential SNMR signal, reduce the influence of random noise on the inversion result of aquifer, and improve the ability of SNMR detection.
Key words:Nuclear magnetic resonance/
Sparse representation/
Generalized orthogonal maturing pursuit/
Redundancy dictionary



PDF全文下载地址:

http://www.geophy.cn/data/article/export-pdf?id=dqwlxb_14644