超小波稀疏表示与相干体技术研究

发布时间：2018-08-03 21:10

【摘要】：人们对自然能源的开发和利用大大推动了社会的发展，提高了社会生产力，为人们生活提供了极大的便利，为了保持能源，尤其是煤炭、石油、天然气这些地下能源的可持续利用，人们必须尽可能最大程度的合理开发和利用这些地下的天然宝藏。随着能源需求度的越来越大，对地下能源的寻找勘探变得愈发重要，与此同时寻找新的油气储藏变得越来越困难，这就需要人们高效快速准确的寻找油气储藏区域，以便及时开发并合理利用自然能源。地震勘探是分析、构造地下地质结构的重要方法之一，地震数据包含了包括振幅、波形、频率、衰减、能量等多种地震属性在内的丰富的信息，对地震属性的解释能有效解决地质构造问题，尤其有利于寻找石油和天然气藏。其中断层解释是地震解释中最重要的问题之一。断层线的精确识别能够帮助研究人员对地下构造、油气储藏等做出有正确分析和判断，因此提高断层线识别的精度具有重要的现实意义。小波对一维信号的点奇异的捕捉是很有力的，但是对高维信号中线奇异和面奇异的捕捉便不是那么令人满意了，而超小波的诞生解决了这个问题。近年来在地震勘探领域，，超小波的使用越来越深入。Surfacelet变换是指先对信号进行多分辨率分解变换，然后利用多维方向滤波器组对相同方向上的变换系数进行合并，Surfacelet变换能有效的对高维信号中的曲面奇异进行刻画。Surfacelet能很方便的处理离散的三维信号，这一特性十分适合处理三维地震数据，提高地震资料信噪比，本文采用Surfacelet变换处理地震数据，对各向异性这一性质进行充分利用。在三维地震数据的预处理上，取得了比较好稀疏表示和降噪效果，对后续提取断层信息提供了更多帮助。相干体方法的的基本原理是通过在一定时窗内计算三维地震数据体中每一道上每一点所在地震道与相邻地震震道的相似程度，这个相似程度可以用一个具体的数值来表示，最后会形成一个新的三维数据体，数据体中的每一点都代表该点与邻域内点的相关程度。其中分析时窗中心某一道的某一点与相邻若干道的相似程度表明了该点所在的地震道与相邻地震道的相关程度，相关程度大则相干值大，相关程度小则相干值小，这个结果体现了地震道之间的不连续程度，有利于识别断层、裂缝和特殊岩体等。本文的主要研究内容包括一下几点： 1.研究了地震勘探领域的相关知识，包括地震数据的采集，地震数据处理等，并研究了图像稀疏表示领域中Surfacelet变换的主要理论知识，对Surfcelet变换进行了全面而深入的学习，包括方向滤波器（DFB）的起源和多维方向滤波器（NDFB）的具体概念及操作流程，然后对Surfacelet变换的过程进行了详细的研究和分析； 2．提出了基于Surfacelet变换的地震数据去噪方法。针对三维地震数据体，先对目标区域进行Surfacelet频域分解变换，得到地震数据不同尺度下不同方向的频率子带，这些子带都是由一系列系数构成的，由于地震数据中的噪声往往分布在数据的高频部分，在此基础上，对精细尺度层系数进行处理，然后对系数处理后的数据进行重构，达到了提高地震数据信噪比的目的； 3．对地震勘探领域断层识别用到的主要技术进行深入学习和研究，包括第一代基于互相关的相干体技术，第二代基于相似的相干体技术，第三代基于本征结构的相干体技术等，并对相干体技术在断层识别上的应用进行分析和总结； 4．提出了改进的多特征值相干体算法。在第三代相干体技术的基础上，对地震道内每一点地震数据的相干值计算的方法进行改进，使用多个特征值计算相干性，提高了对地震数据能量的保持程度。实验结果表明，与传统的第三代相干体算法比，改进的多特征值相干体算法更适于提取断层信息，此外当与Surfacelet变换结合时，效果更为明显。
[Abstract]:The development and utilization of natural energy has greatly promoted the development of society, improved the social productivity and provided great convenience for people's life. In order to keep energy, especially coal, oil and natural gas, the sustainable utilization of underground energy, such as coal, oil and natural gas, must be used to the maximum reasonable development and utilization of these underground days. As the energy demand is increasing, it is becoming more and more important for the exploration and exploration of underground energy. At the same time, it is becoming more and more difficult to find new oil and gas storage. This requires people to find oil and gas storage areas efficiently, quickly and accurately, so as to develop and utilize natural energy in time. One of the important methods of geological structure, seismic data contains abundant information including amplitude, waveform, frequency, attenuation, energy and many other seismic attributes. The interpretation of seismic attributes can effectively solve the problem of geological structure, especially in the search of oil and natural gas reservoirs. Fault interpretation is the most important problem in seismic interpretation. The accurate identification of fault lines can help the researchers to make correct analysis and judgment on the underground structure and oil and gas storage. Therefore, it is of great practical significance to improve the accuracy of fault line recognition.
The capture of point singularity of one-dimensional signals is very strong, but the capture of line singularities and surface singularity in high dimensional signals is not so satisfactory, and the birth of super wavelet solves this problem. In recent years, in the field of seismic exploration, the use of super wavelet is more and more deep into the.Surfacelet transform, which means that the signal is first divided into multiple points. Discrimination rate decomposition transformation, and then the use of multidimensional directional filter banks to merge the transformation coefficients in the same direction, Surfacelet transform can effectively describe the surface singularity in high dimensional signals,.Surfacelet can easily handle discrete three-dimensional signals. This characteristic is suitable to deal with 3D seismic data and improve seismic data letter. In this paper, the Surfacelet transform is used to deal with seismic data and make full use of the properties of anisotropy. In the preprocessing of 3D seismic data, a better sparse representation and noise reduction effect is obtained, and more help is provided for the subsequent extraction of fault information. The basic principle of the coherent body method is to be calculated in a certain time window. The similarity degree of the seismic channel and the adjacent seismic channel in each point in the 3D seismic data body can be represented by a specific value. Finally, a new three-dimensional data body is formed, and each point in the data body represents the degree of correlation between the point and the neighborhood point. The similarity between one point of the road and the adjacent main road shows the degree of correlation between the seismic channel and the adjacent seismic channel in which the point is located. The correlation is large and the correlation is small, the coherence is small. This result shows the discontinuity between the seismic channels, and is beneficial to the identification of faults, cracks and special rock masses.
The main contents of this paper include a few points:
1. study the related knowledge in the field of seismic exploration, including seismic data acquisition, seismic data processing, and study the main theoretical knowledge of Surfacelet transform in the field of image sparse representation, and make a comprehensive and in-depth study of Surfcelet transform, including the origin of directional filter (DFB) and the concrete of multidimensional directional filter (NDFB). Concept and operation process, and then the process of Surfacelet transformation is studied and analyzed in detail.
2. the method of seismic data de-noising based on Surfacelet transform is proposed. For 3D seismic data, the Surfacelet frequency domain decomposition transformation of the target area is first carried out to obtain the frequency subbands of different directions under different scales of seismic data. All these subbands are made up of a series of coefficients, because the noise in the seismic data is often distributed in the number of data. According to the high frequency part, the fine scale layer coefficient is processed on this basis, and then the data after the coefficient processing are reconstructed to improve the signal to noise ratio of seismic data.
3. the main techniques used in fault recognition in the field of seismic exploration are studied and studied in depth, including the first generation of coherent coherent technology based on cross correlation, the second generation based on the similar coherent body technology and the third generation of the coherent body based on the eigenstructure, and the application of the coherent body technology in fault recognition;
4. an improved multi eigenvalue coherent body algorithm is proposed. On the basis of the third generation coherent body technology, the method of calculating the coherence value of each seismic data in the seismic channel is improved. The coherence of the seismic data is calculated by using multiple eigenvalues. The experimental results show that the third generation coherency body is with the traditional third generation coherence body. Compared with the improved multi-eigenvalue coherence cube algorithm, the improved multi-eigenvalue coherence cube algorithm is more suitable for extracting fault information, and the effect is more obvious when combined with Surfacelet transform.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.4

【参考文献】