基于经验模态分解法的层序地层划分及对比研究

Study on sequence stratigraphic division and correlation based on method of empirical mode decomposition

(中国石油大学（北京）油气资源与探测国家重点实验室，北京102249)

State Key Laboratory of Petroleum Resources and Processing,China University of Petroleum (Beijing),Beijing 102249,China

测井资料是地层信息的集合体，测井信号的奇异性一般反映了地层边界信息。经验模态分解能检测信号的奇异性。为此，针对目前层序地层划分及对比缺乏统一的标准及存在人为性和多解性等问题，提出了利用基于测井资料的经验模态分解法进行地层划分及对比。方法的基本原理是：①利用经验模态分解法对测井资料进行分解，得到不同频率成分的本征模态函数;②根据多级分解下的本征模态函数表现出的周期性振荡特征,将其与地质上划分的各级层序界面建立对应关系，即不同频率的周期对应不同规模的层序、各个频率段之间的突变点对应着层序界面，进而实现对各级层序的划分;③在各级层序格架控制下，进行等时地层对比。在长庆油田延长组的应用表明，该方法得到的结果比地质分层结果更精细，对比效果更好，为利用测井资料进行高分辨率层序地层研究提供了一种新的手段。

Logging data is an aggregate of stratigraphic information,the singularity of logging data generally reflects the stratigraphic boundary information. Empirical Mode Decomposition (EMD) can detect the singularity of signal. Therefore,aiming at the problems that sequence stratigraphic division and correlation lack unified standard and exist factitiousness & multiplicity,the method of Empirical Mode Decomposition based on logging data was proposed for stratigraphic division and correlation. The basic principles of this method are:①using EMD to decompose logging data,and obtaining intrinsic mode function (IMF) with different frequency components;②according to periodic oscillation characteristics of IMF by multistage decomposition,establishing the corresponding relationship between IMF and sequence boundaries of all levels (that is,different frequency cycle is corresponding to different scale sequence,and catastrophe points among frequency ranges are corresponding to sequence boundaries),then dividing sequences;③under the control of sequence framework at all levels,carrying out isochronous stratigraphic correlation. The application at Yanchang group in Changqing oilfield shows that the results are more accurate and better than geologic layer classification. The method provides a new means for studying high-resolution sequence by using well-logging data.