基于形态分量分析的二维地震信号重建算法研究

发布时间：2018-07-05 14:42

本文选题：地震信号重建 + 形态分量分析　；参考：《河北工业大学》2015年硕士论文

【摘要】：地震信号重建在石油地震勘探资料处理中具有重要意义。它能够用来重建缺失的地震道,给波动方程偏移等后续处理模块提供规整的输入数据,实现不同区块间的连片处理,有助于消除海上采集中的羽状漂移和多次波等。为了更好地表达地震信号中的不同内蕴特征,进一步提升重建算法性能,本文在形态分量分析理论框架下开展地震信号重建新算法研究。主要工作概述如下:(1)形态分量分析框架下基于DCT和曲波字典组合的地震信号重建研究针对单一型数学变换或字典不能有效刻画地震信号的形态特征多样性这一问题,本文在形态分量分析(MCA)框架下,提出了一种基于离散余弦变换(DCT)与曲波双字典组合的地震信号重建算法。该算法首先将信号分解成局部奇异形态分量以及平滑与线状形态分量。然后采用DCT字典表示局部奇异分量,采用曲波字典表示平滑与线状分量。再以迭代求解方式逐一重建各分量,最后将重建后的分量合并。人工合成地震信号、二维叠前及叠后实际地震信号重建实验表明,该算法能很好完成重建,重建精度不仅要高于非抽样小波变换(UDWT)与曲波字典、曲波与曲波字典、余弦调制滤波器组与曲波等其它双字典组合,而且更是要高于DCT、UDWT、或曲波等单一型字典。(2)形态分量分析框架下基于波原子变换字典的地震信号重建研究地震信号具有震荡特点,虽然DCT或UDWT等变换能捕获沿震荡方向的形态特征,但对穿过该方向的特征描述不够,而波原子变换作为一种新的多尺度几何分析工具,能够弥补以上不足,且对震荡信号具有最优稀疏表示。为此本文提出了MCA框架下基于波原子变换字典的二维地震信号重建算法。该算法首先对二维地震信号进行形态分量分解,分别获得局部奇异分量和平滑与线状分量。然后利用波原子字典重建局部奇异分量;仍然利用曲波字典重建平滑线状分量,最后将重建结果合并。实验表明,基于波原子变换字典的地震信号重建精度不仅要高于DCT和曲波字典组合,而且更是要高于非抽样小波变换(UDWT)与曲波字典、曲波与曲波字典、余弦调制滤波器组与曲波等其它双字典组合。(3)形态分量分析框架下基于多冗余字典线性组合的地震信号重建研究地震信号是地下复杂地质结构的响应,具有特征多样性特点,仅依靠单一型字典或者双字典组合未必能够有效刻画这种特征多样性。本文尝试在MCA框架下实现地震信号更精细的分解,并用多个过完备冗余字典的线性组合来重建地震信号。通过对体现不同特征的分量进行有针对性的重建,并在重建中采用硬阈值筛选和指数递减更新策略,获取更高的重建性能。实验表明基于多字典线性组合的重建算法不仅能够有效重建地震信号,而且能适应复杂地震信号重建情形。该算法与基于波原子变换字典的地震信号重建算法性能相当,但比其它双字典组合的重建效果要好,更要优于单一型字典的重建效果。
[Abstract]:Seismic signal reconstruction is of great significance in the processing of petroleum seismic data. It can be used to reconstruct the missing seismic channel, provide regular input data for the subsequent processing modules such as wave equation migration, and realize the continuous processing between different blocks. It is helpful to eliminate the pinnate drift and multiple waves in the sea collection. To further improve the performance of the seismic signal, this paper studies the new algorithm of seismic signal reconstruction under the framework of morphological component analysis. The main work is summarized as follows: (1) the research of seismic signal reconstruction based on the combination of DCT and Qu Bo dictionary under the framework of morphological component analysis is aimed at single mathematical transformation or Dictionaries can not effectively characterize the morphological diversity of seismic signals. Under the framework of morphological component analysis (MCA), an earthquake signal reconstruction algorithm based on discrete cosine transform (DCT) and Qu Bo's double dictionary is proposed. The algorithm first decomposes the signal into local singular component and smooth and linear shape. Then the DCT dictionary is used to represent the local singular component, and the Qu Bo dictionary is used to express the smooth and linear components. Then the components are reconstructed one by one by iterative solution. Finally, the reconstructed components are merged. The seismic signals are synthesized by artificial synthesis. The reconstruction and reconstruction of the two dimensional pre stack and post stack actual seismic signals show that the algorithm can complete the reconstruction and reconstruct the essence well. The degree is not only higher than the unsampled wavelet transform (UDWT) and Qu Bo dictionary, Qu Bo and Qu Bo dictionary, the cosine modulation filter bank and Qu Bo and other double dictionaries, but also higher than the single dictionary of DCT, UDWT, or Qu Bo. (2) seismic signal reconstruction based on the wave atomic transformation dictionary based on the form component analysis framework With the characteristics of concussion, although the transformation of DCT or UDWT can capture the morphological characteristics along the direction of the oscillation, it is not enough to describe the characteristics passing through this direction. As a new multi-scale geometric analysis tool, the wave atom transformation can make up the above shortcomings and have the optimal sparse representation for the shock signal. Therefore, this paper proposes a wave based on the MCA framework. The algorithm of two-dimensional seismic signal reconstruction in the atomic transformation dictionary is used to decompose the morphological components of the two-dimensional seismic signals. The local singular components and the smooth and linear components are obtained respectively. Then the local singular components are reconstructed by the wave atom dictionary, and the Qu Bo dictionary is still used to reconstruct the smooth linear components. Finally, the reconstruction results are merged. The experimental results show that the accuracy of seismic signal reconstruction based on the wave atom transform dictionary is not only higher than the combination of DCT and Qu Bo dictionary, but also higher than the non sampling wavelet transform (UDWT) and Qu Bo dictionary, Qu Bo and Qu Bo dictionary, the cosine modulation filter bank and Qu Bo and other double word codes. (3) the multi redundant word based on the form component analysis framework. The seismic signal reconstruction of the canonical linear combination is a response to the complex geological structure of the underground. It has the characteristics of characteristic diversity. Only single dictionary or double dictionary combination may not be able to effectively depict this characteristic diversity. In this paper, we try to realize the more detailed decomposition of seismic signals under the MCA framework and use multiple overcomplete redundancy. The linear combination of the remaining dictionaries is used to reconstruct the seismic signals. Through the targeted reconstruction of the components that reflect the different features, the hard threshold screening and exponential reduction updating strategy are used in the reconstruction to obtain higher reconstruction performance. It can adapt to the reconstruction of complex seismic signal. The algorithm is equivalent to the seismic signal reconstruction algorithm based on the wave atom transformation dictionary, but it is better than the other two dictionary combinations, and it is better than the single type dictionary.
【学位授予单位】：河北工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.4

【相似文献】