大庆石油地质与开发

2008, No.127(03) 113-116

[打印本页] [关闭]
本期目录(Current Issue) | 过刊浏览(Archive) | 高级检索(Advanced Search)

基于人工神经网络的聚合物驱提高采收率预测——人工神经网络与二次多项式逐步回归方法的对比
Predication for EOR by polymer flooding based on artificial neural network——Comparison between ANN and quadratic polynomial stepwise regression method

周丛丛,李洁,张晓光,罗峰,李亚,曾嘉
ZHOU Cong-cong1,LI Jie2,ZHANG Xiao-guang3,LUO Feng2,LI Ya2,ZENG Jia2(1.China University of Geosciences

摘要(Abstract):

准确评价和预测聚合物驱开发效果对油田开发和管理具有重要指导意义。储层微观孔隙结构参数是影响聚合物驱开发效果的重要因素,二者之间是一种非线性、不确定的复杂关系。用相关性分析优选出对聚驱提高采收率值影响较大的参数,采用多项式回归分析和BP神经网络的方法对于这种非线性、不确定的多变量系统进行预测,结果表明人工神经网络方法具有更好地自适应性,能较好的反映影响聚驱效果的各种微观参数与提高采收率值的内在联系,而且预测精度较高。因此认为应用BP神经网络方法预测聚驱效果是可行的、有效的。
Correctly evaluating and predicting development effects of polymer flooding have an important guide meaning.Microscopic pore structural parameters in reservoir significantly effect on polymer flooding development.There is a nonlinearity and uncertainty complex relationship between them.The parameters greatly affect EOR is optimized by using correlation analysis.Then this kind of nonlinearity and uncertainty multivariable system is predicted by applying quadratic polynomial stepwise regression method and BP neural network.Results show that BP neural network is a better method which can automatically adapt and reflect internal relations between various microscopic parameters and enhanced recovery factor values and have high prediction accuracy.

关键词(KeyWords): 聚合物驱;多项式回归;人工神经网络;孔隙结构;预测
polymer flooding;quadratic polynomial regression;artificial neural network;pore structure;prediction

Abstract:

Keywords:

基金项目(Foundation):

作者(Author): 周丛丛,李洁,张晓光,罗峰,李亚,曾嘉
ZHOU Cong-cong1,LI Jie2,ZHANG Xiao-guang3,LUO Feng2,LI Ya2,ZENG Jia2(1.China University of Geosciences

参考文献(References):

扩展功能
本文信息
服务与反馈
本文关键词相关文章
本文作者相关文章
中国知网
分享