一种基于KPCA的绕射信息分离方法

Separation method of diffraction on post-stack record based on kernel principal component analysis

在油气勘探的过程中,需要对叠后地震数据中的绕射波信息进行提取,从而得到地下小尺度地质体的信息.本文在前人提出的使用主成分分析法(PCA)提取绕射波信息的基础上,进一步发展使用了基于核的主成分分析技术(KPCA)的提取绕射波信息方法,即通过构建数据模型,选取核函数进行KPCA运算,并利用不同的主分量进行信号重构,从而达到将绕射波和反射波分离的目的.相对于传统PCA,这种方法对于地震数据中弯曲的同相轴,以及倾斜的界面或者倾斜的同相轴有着更好的识别能力,在提取过程中能够提供更加精细的绕射波信息.

In oil and gas exploration, small geologic features manifest themselves in seismic data in the form of diffracted waves, so it＇s significant to separate seismic diffraction from post-stack seismic data and apply the imaged results to elucidate underground minor structure.In this paper, we develop an new approach for extracting and imaging of diffracted events based on KPCA which is a nonlinear extension of PCA. The method proposed can be divided into three steps. First, we build the seismic data model which can be processed by the method of KPCA. Second, principal component is obtained after the KPCA processing using the appropriate kernel function. Third, the signal reconstruction is gained using different principal components. Compared with the traditional method of PCA, the novelty and advantage of our approach is that it has a better ability to identify the crooked seismic events in the seismic data, and it can also provide more precise diffraction wave information during the processing.