摘要
弹性阻抗反演是主要的储层参数预测方法之一。入射角范围有限和低地震资料质量等原因,会导致密度反演的准确率较低。此外,纵横波速度比在速度峰值处的反演稳定性较差,也不能直接反演孔隙度等储层物性参数。目前通过常规的弹性阻抗反演等方法无法有效解决上述问题。本文将深度学习技术与弹性阻抗反演技术相结合,基于全连接深度神经网络建立起三个角度弹性阻抗与弹性、物性储层参数之间的非线性映射关系,测井数据的密度、纵横波速度比预测的均方根误差均降低10%以上。并以弹性阻抗搭建起测井、地震数据之间的桥梁,通过标准化等数据处理技术,最终得到密度、纵横波速度比和孔隙度的三维预测结果。
Elastic impedance inversion is one of the main reservoir parameter prediction methods.Unfortunately,we may obtain an inaccurate estimation of density from EI due to limited range of incident angle of seismic data and the poor quality of seismic data.In addition,the estimation of V P/V S has low stability at the peak,and the physical parameters of reservoirs such as porosity can not be directly inverted.At present,conventional elastic impedance inversion methods can not effectively solve the above problems.This paper combined deep learning technique with elastic impedance inversion technique,we established a non-linear mapping relationship between the three-angle elastic impedances and the reservoir parameters based on the fully connected deep neural network.The root mean square error of the predictions of both density and V P/V S is reduced by more than 10%in logging data.Then the elastic impedance is used as a bridge between logging and seismic data,with the data standardization technic,the three-dimensional prediction results of density,V P/V S and porosity are finally obtained.
作者
安鹏
高健祎
曹丹平
牛洪彬
吴凡
An Peng;Gao Jian-yi;Cao Dan-ping;Niu Hong-bin;Wu Fan(Information Technology Center,China National Offshore Oil Corporation,Beijing 100010,China;College of Geosciences,China University of Petroleum(East China),Qingdao 266580,China;CNOOC Energy Development Limited by Share Ltd.Engineering Branch,Tianjin 300452,China)
出处
《海洋工程装备与技术》
2019年第S01期255-260,共6页
Ocean Engineering Equipment and Technology
关键词
深度学习
地震反演
深海勘探
储层参数预测
deep learning
seismic inversion
deepwater exploration
reservoir parameter prediction
