基于压缩感知的多源地震交涉炮分离方法研究

With the detecting target of seismic exploration becoming more and more complex, the wide-azimuth, broadband and high-density acquisition technology has been widely used for seismic data acquisition, which also costs a significant rise in the cost of data acquisition. Due to its high efficiency and low cost, the multi-source acquisition technology becomes more and more popular in field data acquisition. However, it makes the standard pre-stack processing such as velocity analysis difficult to implement. Therefore, it is necessary to separate the blended data to obtain single-shot records. Based on compressive sensing theory, this project proposes to study the separation method for both data acquisition and data processing. For data acquisition, we propose to optimize the time delay sequence of sources to reduce the coherence of sources. For data processing, we propose to construct an over-complete dictionary to transform multi-source seismic data to sparse domain. A hybrid iterative algorithm of both L1 and L0 regularizations will be used to separate the blended data so that the single shot seismic records can be obtained. The proposed method is expected to solve the problems in the application of the multi-source acquisition technology, and we believe it has good application prospect.

随着地震勘探目标日趋复杂，“两宽一高”的采集方式被广泛应用，但同时也导致采集成本急剧升高。多源地震采集技术凭借高效率和低成本，越来越多的被应用于野外地震资料采集。由于该技术获得的是交涉炮地震记录，无法进行速度分析等叠前常规地震数据处理，因而需进行交涉炮数据分离，获得各个震源的单炮地震记录。基于压缩感知理论，本项目从采集和处理两个方面入手，进行交涉炮数据的分离方法研究。在资料采集方面，优化设计震源激发时间延迟序列，从采集上降低交涉炮能量的相干性。在资料处理方面，构建多源地震信号的超完备字典，获得稀疏域地震记录，利用L1正则化和L0正则化混合迭代算法，反演分离交涉炮数据，获得各震源单炮地震记录。本项目提出的方法能解决多源地震采集技术的应用难题，具有良好的应用前景。

随着勘探环境和勘探目标的日趋复杂，对勘探方法提出了更高勘探精度的要求。多源地震采集技术凭借高效率和低成本，越来越多的被应用于环境复杂地区的高精度勘探。项目进行多源地震交涉炮数据分离方法研究。首先，基于压缩感知思想，为震源激发时间序列添加稀疏约束，进行基于压缩感知的多源激发延迟时间的优化设计，获得各震源的最优激发延迟时间，在采集上降低交涉炮能量的相干性；然后，采用字典学习方法，自适应地按照稀疏约束条件训练出超完备字典，自适应地根据训练样本提取地震信号特征，采用K-SVD方法进行字典求取和更新，获得了地震资料的超完备字典，对多源地震信号进行稀疏表达；最后，基于超完备字典将地震信号变换到稀疏域，通过软阈值算法求解L1范数约束问题，获得迭代初值，并估算信号的稀疏水平，通过硬阈值算法求解L0范数约束问题，获得数据分离效果，将分离数据进行反变换，得到时空域的数据分离结果。通过项目的研究，进一步提高了多源地震数据分离的精度，从而提高了勘探效率，进一步推动了“两宽一高”采集方式的发展。