基于远减近衰减梯度属性与Bi-LSTM神经网络联合的隐蔽河道砂体识别方法  

Concealed channel sand body identification method based on the combination of far minus near attenuation attributes and Bi-LSTM neural network

作　　者：周鹏[1,2] 曹俊兴 刘俊[1,2] 陈思远[1,2] 王俊 ZHOU Peng;CAO JunXing;LIU Jun;CHEN SiYuan;WANG Jun(Key Laboratory of Earth Exploration and Information Technology of Ministry of Education,Chengdu University of Technology,Chengdu 610059,China;State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation of Chengdu University of Technology,Chengdu 610059,China)

机构地区：[1]地球勘探与信息技术教育部重点实验室,成都610059 [2]油气藏地质及开发工程国家重点实验室(成都理工大学),成都610059

出　　处：《地球物理学进展》2022年第5期2129-2137,共9页Progress in Geophysics

基　　金：国家自然科学基金项目(41974160,42030812)资助。

摘　　要：岩性作为一种重要的储层物性,其包含了重要的流体信息及储层特征,对其准确的识别可以为储层预测工作提供重要基础.我国一些河流相油气藏沉积环境复杂,多期叠置等问题造成储层砂体隐蔽的现象.为有效的识别隐蔽河道砂体,本文发展了一种基于远减近衰减梯度属性与Bi-LSTM神经网络联合的隐蔽河道砂体识别方法.该方法以远减近地震衰减梯度属性凸显隐蔽河道砂体为基础,以此为输入,河道砂体作为标签,借助Bi-LSTM深度神经网络构建输入数据与砂体岩性间的非线性关系.应用该方法进行川西中江气田某工区隐蔽河道砂体识别,识别出了此前未曾发现的隐蔽河道砂体,并刻画出了2条隐蔽河道.As an important reservoir physical property,lithology contains important fluid information and reservoir characteristics,and its accurate identification can provide an important foundation for reservoir prediction.The depositional environment of some fluvial oil and gas reservoirs in my country is complicated,and multi-stage superposition and other problems have caused the concealment of the reservoir sand body.In order to effectively identify concealed channel sand bodies,this paper develops a concealed channel sand body identification method based on the combination of far minus near attenuation attributes and Bi-LSTM neural network.This method is based on the pre-stack remote-channel seismic attenuation attribute to highlight the concealed channel sand body,and uses this as the input,the channel sand body is used as the label,and Bi-LSTM deep neural network was used to construct the nonlinear relationship between the input data and the lithology of the sand body.This method was used to identify the concealed channel sand body in an area of Zhongjiang gas field in western Sichuan province,identifying the concealed channel sand bodies that had not been discovered before,and two hidden channels were depicted.

关 键 词：双向长短期记忆神经网络 岩性预测 地震属性 

分 类 号：P631[天文地球—地质矿产勘探]