摘要
利用逐步回归分析、神经网络、相关滤波、协克里金和非参数回归分析等方法 ,实现了由地震属性与测井资料联合应用对孔隙度参数的平面分布预测。通过实例分析 ,比较了各自的地质效果 ,归纳总结出各种方法的特点及应用条件
By means of stepwise regression analysis, neural network, correlation filtering, CoKrige, and nonparametric regression, this article realizes the prediction of porosity distribution from seismic attributes and logging data. The behaviors of different methods and their conditions of application are summarized through the comparison of geological effects resulted from case analysis.
引文
1 LawrenceP ,AramcoS ,Dhahran,etalSeismicattributesinthecharacterizationofsmall scalereservoirinAbqaiqField[J].TheLeadingEdge,1998,17(4):521~525
2 ChenQ ,SideneyS .Seismicattributetechnologyforreservoirforecastingandmonitoring[J]TheLeadingEdge,1997,16(5):445~456
3 杜世通,王永刚地震参数综合处理方法在储层横向预测中的应用[J]石油大学学报(自然科学版),1993,17(1):8~15
4 刘企英编利用地震信息进行油气预测[M]北京:石油工业出版社,1994172~219
5 王捷油藏描述技术—勘探阶段[M]北京:石油工业出版社,1996176~244
6 庄镇泉,王熙法,王东生神经网络与神经计算机[M],北京:科学出版社,199018~31
7 StoneCJ.Consistentnonparametricregression(withdiscus sion)[A]AnnMathStatist5,1997。595~620
8 陈希儒,王松桂近代实用回归分析[M],南宁:广西人民出版社,1984233~260