基于相似性传播聚类与主成分分析的断层识别方法

发布时间：2018-06-12 04:21

本文选题：断层识别方法 + 连通区域　；参考：《石油地球物理勘探》2017年04期

【摘要】：针对现有自动地震断层识别方法所存在的精度低、耗时较长及不能获得量化的断层识别结果等问题,提出了一种基于相似性传播聚类与主成分分析(Principal Component Analysis,PCA)的地震断层识别方法。首先,采用连通区域标注方法确定地震层位的不连续点。然后,利用相似性传播聚类算法对层位不连续点进行聚类,每一类不连续点可以确定一条断层,以此可获得断层的数量和每个类别的聚类中心。最后,基于PCA方法计算出每一类层位不连续点的主方向,将沿着主方向且经过相应聚类中心的线段作为断层。基于模型数据和实际地震数据将文中方法与现有方法进行了对比,在峰值信噪比、均方误差、时间消耗和断层条数符合率等方面彰显了文中方法的合理性,并可对断层进行"量化"解释,在地震勘探等领域具有较高的实际意义。
[Abstract]:Aiming at the problems existing in the existing automatic seismic fault recognition methods, such as low accuracy, long time consuming and unable to obtain quantitative fault identification results, a fault recognition method based on similarity propagation clustering and principal component analysis (PCA) is proposed. First, the discontinuous points of seismic horizon are determined by means of the connected region annotation method. Then, the similarity propagation clustering algorithm is used to cluster the discontinuous points of the horizon. Each discontinuous point can determine a fault, and then the number of faults and the cluster center of each class can be obtained. Finally, based on PCA method, the main directions of discontinuous points of each kind of layers are calculated, and the lines along the main direction and passing through the corresponding clustering center are regarded as faults. Based on the model data and the actual seismic data, the method is compared with the existing methods. The method is reasonable in the aspects of peak signal-to-noise ratio (PSNR), mean square error (MSE), time consumption and fault number coincidence rate, etc. It is of great practical significance in seismic exploration and so on.
【作者单位】：中国科学院地质与地球物理研究所中国科学院油气资源研究重点实验室;北京工业大学信息学部计算机学院;
【分类号】：P631.4

【参考文献】