摘要
将归一化的地层系数、油水井连通状况、连通油井数、连通井距、量化的砂体类型和措施类型以及渗透级差倒数作为输入参数,相对吸水量作为目标参数,结合支持向量机(SVM)和BP神经网络方法建立隐式非线性吸水剖面预测模型,通过模型的样本学习,建立地质、开发参数和相对吸水量之间的隐式非线性关系。实例分析表明,隐式非线性方法预测吸水剖面精度高且易于扩充模型变量数,要求样本少,更适合矿场应用。
A prediction model of water injection profile using implicit nonlinear method was established combined support vector machine and BP neural network method.In this model,the normalized formation factor,oil and water well connectivity,number of connected wells,connected well space,types of sandstone and stimulations and reciprocal of permeability ratio were taken as input parameters,and the relative intake capacity was taken as target parameter.The implicit nonlinear relationship among geology and production parameters and the relative intake capacity was developed.The results show that this method has high prediction precision for water injection profile,and it is easy to extend the variable number.Under small sample data conditions,this method is more suitable for field application.
出处
《中国石油大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2010年第6期95-98,共4页
Journal of China University of Petroleum(Edition of Natural Science)
基金
国家科技重大专项课题(2008ZX05024-04-008)
关键词
隐式非线性方法
注水井
吸水剖面
预测
implicit nonlinear method
water injection well
water injection profile
prediction
