新疆石油天然气 ›› 2024, Vol. 21 ›› Issue (2): 37-.DOI: 10.12388/j.issn.1673-2677.2024.02.005

• 油气开发 • 上一篇    下一篇

基于BP神经网络的低渗透底水油藏油井见水模式预测模型

  

  1. 1. 西南石油大学油气藏地质及开发工程国家重点实验室,四川成都 610500;
    2,天府永兴实验室,四川成都 610213
  • 出版日期:2024-06-06 发布日期:2024-06-06
  • 通讯作者: 靳星(1995-),西南石油大学石油与天然气工程学院在读博士,目前从事提高采收率技术研究。(Tel)17809212967(E-mail)jinxing19950214@163.com
  • 作者简介:蒲万芬(1961-),1983年毕业于西南石油学院开发系油田化学专业,教授,博士生导师,目前从事提高采收率理论教学和科研工作。(Tel)18180525412(E-mail)puwanfen214@126.com
  • 基金资助:

    国家自然科学基金重点项目“黏度可控的原位增黏体系构建及高效驱油机理研究”(U19B2010)

Prediction Model of Water Breakthrough Patterns of Low-Permeability Bottom Water Reservoirs Based on BP Neural Network

  1. 1. State Key Laboratory of Oil and Gas Reservoir Geology and Exploitation,Southwest Petroleum University,Chengdu 610500,Sichuan,China;
    2. Tianfu Yongxing Laboratory,Chengdu 610213,Sichuan,China
  • Online:2024-06-06 Published:2024-06-06

摘要:

注水开发使得低渗透底水油藏油井见水模式更加复杂,需要进一步明确及预测油井见水模式来针对性地指导水淹治理措施。神经网络模型具备处理多元回归问题和计算速度快等优势,可被用于分析地质工程多因素参数与油井见水模式的内在关系,构建见水模式预测模型。在油井见水模式划分的基础上,通过灰色关联理论和神经网络算法对BCL低渗透底水油藏油井见水模式的主控因素和预测模型进行了研究。发现水层厚度、隔夹层数、隔夹层长度和避水高度是该类油藏注水开发下影响油井见水模式的主控因素。基于主控因素结合神经网络算法建立了油井见水模式预测模型。通过对18组测试数据进行验证,平均预测误差1.4%,获得了较好的预测精度。通过易于获取的主控因素快速预测注水开发低渗透底水油藏油井的见水模式,为该类油藏的高含水针对性治理提供基础依据。

关键词:

神经网络, 预测模型, 见水模式, 主控因素, 低渗透底水油藏

Abstract:

Water injection development makes the water breakthrough patterns of low-permeability bottom water reservoirs more complex,requiring further clarification and prediction to guide targeted waterflooded treatment measures. With the advantages of handling multiple regression problems and fast computation,neural network models can be used to analyze the inherent relationship between engineering geological multi-factor parameters and water breakthrough patterns and establish prediction models of the patterns. Based on the classification of water breakthrough patterns,this paper studied the main controlling factors and prediction models of water breakthrough patterns of BLC low-permeability bottom water reservoirs through grey correlation theory and neural network algorithms. The results show that the thickness of the bottom water,the number of interbeds,the length of interbeds,and the height of the water-avoiding perforation section are the main controlling factors affecting the water breakthrough patterns under water injection development of such reservoirs. Based on the main controlling factors and neural network algorithms,a prediction model of water breakthrough patterns was established. By verifying 18 test data sets,the model achieved an average prediction error of 1.4%,with good prediction accuracy. The water break through patterns of low-permeability bottom water reservoirs under water injection development can be quickly predicted through easily obtainable main controlling factors,providing a primary basis for targeted treatment of high-water-containing reservoirs.

Key words:

neural network, prediction model, water breakthrough pattern, main controlling factor, low-permeability bottom water reservoir

中图分类号: