应用人工神经网络预测油田注采比

英文篇名：Oilfield injection-production ratio prediction by artificial neural network
中文刊名：大庆石油地质与开发
英文刊名：Petroleum Geology & Oilfield Development in Daqing
作者：袁爱武 ; 郑晓松 ; 于洪波
英文作者：YUAN Ai-wu~1 ; ZHENG Xiao-song~2 ; YU Hong-bo~3(1.Management College of Tianjin University ; Tianjin 300072 ; China ; 2.Petroleum Engineering College of Daqing Petroleum Institute ; Daqing 163318 ; China ; 3.No.2 Drilling Company of Liaohe Oilfield ; Panjin 121207 ; China)
中文关键词：注采比 ; 物质平衡法 ; 多元回归法 ; 神经网络 ; 预测
英文关键词：injection-production ratio ; material balance method ; multiple regression method ; neural network ; prediction
出版日期：2005-04-01
机构：天津大学管理学院,大庆石油学院石油工程学院,辽河油田钻井二公司天津300072 ,黑龙江大庆163318 ,辽宁盘锦121207
年：2005
期：02
出版单位：大庆石油地质与开发

摘要

在简单介绍了BP神经网络的基本原理的基础上, 以实例说明了BP神经网络方法在油田注采比预测中的应用, 并与注采比与水油比法、多元回归法和物质平衡法进行比较; 其次, 分析了各种预测法的计算结果, 并对神经网络预测法进行了检验。结果表明, BP神经网络预测方法具有更好的自适应性, 能够较好地反映影响注采比的各种因素与注采比的内在联系, 而且预测精度较高。因此认为,应用BP神经网络方法预测油田注采比是有效、可行的。
Based on the brief introduction of fundamental principles of BP neural network, this paper states the application of the above method in oilfield injection-production ratio prediction by history case and at the same time it is compared with water-oil-ratio method, multiple regression method and material balance method; in addition, it also analyzes the calculated results of each predicting method and checks the first method. The results show that the first method possesses much better self-feasibility and can not only thoroughly respects the inner links between various influencing factors of injection-production ratio and the ratio itself, but also have much higher prediction accuracy. Therefore BP neural network method can effectively and practically predict oilfield injection-production ratio.

引文

[1]钟德康注采比变化规律及矿场应用[J]石油勘探与开发,1997,24(6):6569
    [2]赵向宏水驱小油藏合理采液速度、合理井网密度及布井方式研究[J]石油勘探与开发,1994,21(1):7579
    [3]郑俊德,姜洪福,冯效树萨中地区合理注采比研究[J]油气地质与采收率,2001,8(2):5557
    [4]姜　岩,周再林,秦月霜地震属性分析的神经网络岩性识别技术及应用[J]大庆石油地质与开发,2000,19(5):3941
    [5]阎平凡,张长水人工神经网络与模拟进化计算[M]北京:清华大学出版社,2000