地震属性神经网络油气模式识别技术及其在东海的应用
|
摘要
神经网络油气模式识别技术是综合利用地震属性进行油气预测的技术之一 ,它首先通过计算获得多种地震属性 ,综合分析找出对储层油气比较敏感的地震属性组合 ;然后收集油气井与非油气井的井旁道地震属性 ,组成学习样本并进行神经网络学习 ;最后利用学习结果对储层进行油气预测。该技术在东海某工区的应用结果表明 ,振幅统计类和复数道统计类地震属性是对该地区储层油气最敏感的地震属性组合。神经网络油气模式识别技术可以作为东海地区储层油气预测的一种手段。
The seismic hydrocarbon pattern recognition of neural network is one of the techniques to forecast oil and gas by seismic attributes. At first, multiple seismic attributes have to be obtained by calculations, and an integrated analysis is conducted to look for a seismic attribute combination relatively sensitive to reservoir hydrocarbon. Then, the attributes from the seismic traces near oil/gas wells and non_oil/gas wells are collected respectively, and are used to make the learning samples of neural network. Finally, the learning results can be used to forecast oil and gas in reservoirs. The application results of this technique in an area, East China Sea, have suggested that the amplitude and complex trace seismic attributes are a combination most sensitive to reservoir hydrocarbon. Therefore, the hydrocarbon pattern recognition of neural network may be considered as an effective tool to forecast reservoir oil and gas in East China Sea.
The seismic hydrocarbon pattern recognition of neural network is one of the techniques to forecast oil and gas by seismic attributes. At first, multiple seismic attributes have to be obtained by calculations, and an integrated analysis is conducted to look for a seismic attribute combination relatively sensitive to reservoir hydrocarbon. Then, the attributes from the seismic traces near oil/gas wells and non_oil/gas wells are collected respectively, and are used to make the learning samples of neural network. Finally, the learning results can be used to forecast oil and gas in reservoirs. The application results of this technique in an area, East China Sea, have suggested that the amplitude and complex trace seismic attributes are a combination most sensitive to reservoir hydrocarbon. Therefore, the hydrocarbon pattern recognition of neural network may be considered as an effective tool to forecast reservoir oil and gas in East China Sea.
引文
1 QuincyChen,SteveSidney.Seismicattributetechnologyforreservoirforecastingandmonitoring.TheLeadingEdge,1997,16(5):445～456
2 杜江.模糊神经网络及其在油气预测中的应用.石油物探,1999,38(4):25～30
3 CctherineLewis.Seismicattributesforreservoirmonitoring:Afeasibilitystudyusingforwardmodeling.TheLeadingEdge,1997,16(5):459～468
4 陈遵德著.储层地震属性优化方法.北京:石油工业出版社,1996
5 倪逸,杨慧珠,郭玲萱等.储层油气预测中地震属性优选问题探讨.石油地球物理勘探,1999,34(6):614～626
6 赵林明,胡浩云著.多层前向人工神经网络.北京:黄河水利出版社,1999
7 张娥,高书琴,侯成福等.利用地震属性预测砂岩储集层厚度及含油饱和度.石油勘探与开发,2000,27(1):92～94
2 杜江.模糊神经网络及其在油气预测中的应用.石油物探,1999,38(4):25～30
3 CctherineLewis.Seismicattributesforreservoirmonitoring:Afeasibilitystudyusingforwardmodeling.TheLeadingEdge,1997,16(5):459～468
4 陈遵德著.储层地震属性优化方法.北京:石油工业出版社,1996
5 倪逸,杨慧珠,郭玲萱等.储层油气预测中地震属性优选问题探讨.石油地球物理勘探,1999,34(6):614～626
6 赵林明,胡浩云著.多层前向人工神经网络.北京:黄河水利出版社,1999
7 张娥,高书琴,侯成福等.利用地震属性预测砂岩储集层厚度及含油饱和度.石油勘探与开发,2000,27(1):92～94
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心