压缩感知及其在地震勘探中的应用

发布时间：2018-04-13 20:06

本文选题：信号和图像的特征表达 + 稀疏与冗余表示　；参考：《石油物探》2016年04期

【摘要】：压缩感知思想在油气勘探领域已经受到了充分的关注。主要体现在两个方面,一个是与随机采样有关的随机地震数据采集系统的设计与评价、同时源激发的混叠数据分离、空间非规则数据的规则化、地震数据的去噪和地震数据压缩(传输)等;另一方面是地震数据的特征表达、模型参数(如速度模型参数)的特征表达、模型特征表达情况下的地震波正演模拟以及对应的稀疏反演方法(包括多尺度反演方法)。前者可以直接借鉴信号和图像处理领域发展的各种方法;后者需要勘探地球物理学家及相关领域研究人员提出合适的思想和方法。来自以沉积地层为主的地下介质中的反射(绕射)波场是有特征的或可压缩的;以层状地层为主的模型参数也是可压缩的。这是压缩感知方法能用于地震勘探的基础。首先简述了压缩感知的基本思想,指出它是一套随机采样理论。然后阐述了无论是叠前地震数据或是叠前偏移成像结果都是可以稀疏表达的,压缩感知理论可以在地震勘探中得到有效应用。接着从随机采样、同时源激发的混叠数据的分离、高维数据规则化方面展示了压缩感知思想的应用。关于随机采样,指出当前地震数据采集的技术重点应该是更高效地实现"两宽一高"的数据采集。最后列举了压缩感知思想在特征数据提取、编码炮集成像和特征波场成像中的应用。
[Abstract]:The idea of compressed perception has attracted much attention in the field of oil and gas exploration.There are two main aspects. One is the design and evaluation of random seismic data acquisition system related to random sampling, the separation of overlapped data generated by source, and the regularization of spatial irregular data.Seismic data denoising and seismic data compression (transmission), on the other hand, the characteristic expression of seismic data, model parameters (such as velocity model parameters),The forward modeling of seismic wave and the corresponding sparse inversion method (including multi-scale inversion method) in the case of model characteristic expression.The former can draw lessons from various methods in the field of signal and image processing, while the latter requires exploration geophysicists and researchers in related fields to put forward appropriate ideas and methods.The reflection (diffraction) wave field from the subsurface medium dominated by sedimentary strata is characteristic or compressible, and the model parameters mainly in stratiform strata are compressible.This is the basis on which compression sensing can be used in seismic exploration.Firstly, the basic idea of compressed sensing is briefly introduced, and it is pointed out that it is a set of random sampling theory.Then it is expounded that prestack seismic data and prestack migration imaging results can be expressed sparsely, and compression sensing theory can be effectively applied in seismic exploration.Then, the application of compression sensing is demonstrated from random sampling, separation of aliasing data excited by source and regularization of high-dimensional data.With regard to random sampling, it is pointed out that the key technology of seismic data acquisition at present should be to realize the data acquisition of "two widths and one height" more efficiently.Finally, the applications of compression sensing in feature data extraction, coding gun integrated image and feature wave field imaging are listed.
【作者单位】：同济大学海洋与地球科学学院波现象与反演成像研究组;
【基金】：国家自然科学基金项目(41374117) 国家科技重大专项(2011ZX05005-005-008HZ;2011ZX05006-002;2011ZX05023)联合资助~~
【分类号】：P631.4

【相似文献】