摘要
针对渤海P油田储层纵向非均质性较强的特点,提出基于主成分分析和优化神经网络(PCA-PSO-BP)的注水井吸水能力预测方法,以提高预测准确度,进而改善纵向吸水不均的问题。根据动静态生产资料,通过皮尔逊线性回归对渗透率、泥质含量、含油饱和度、电阻率、孔隙度、注采井距、射孔厚度共7个因素进行相关性分析,结果显示渗透率与泥质含量、电阻率、孔隙度之间存在一定的线性关系。通过主成分分析对这7个因素进行降维处理,以重新生成的3个主成分作为模型数据集,并对其神经网络模型进行优化,最终建立PCA-PSO-BP预测模型。通过注水井生产实例,验证了此模型的良好效果。
In view of the strong vertical heterogeneity of reservoirs in Bohai P oilfield,a method for predicting the water absorption capacity of water injection wells based on principal component analysis and optimized neural network(PCA-PSO-BP)is proposed to improve the accuracy of prediction and thus solve the problem of uneven vertical water absorption.According to the dynamic and static data of production,Pearson linear regression is used to analyze the correlation of seven factors,including permeability,shale content,oil saturation,resistivity,porosity,injection production well spacing,and perforation thickness.It is found that there is a certain linear relationship between permeability and shale content,resistivity and porosity.By using principal component analysis to reduce the dimensionality of these 7 factors,the three principal components are re-generated as the model data set,and the neural network model is optimized to ultimately establish the PCA-PSO-BP model.The good effect of this model has been verified through production examples of water injection wells.
作者
暴赫
侯亚伟
王刚
安玉华
范佳乐
BAO He;HOU Yawei;WANG Gang;AN Yuhua;FAN Jiale(Bohai Petroleum Research Institute,Tianjin Branch of CNOOC(China),Tianjin 300459,China)
出处
《重庆科技学院学报（自然科学版）》
CAS
2023年第5期22-27,共6页
Journal of Chongqing University of Science and Technology：Natural Sciences Edition
基金
“十四五”国家科技重大专项“海上油田大幅度提高采收率关键技术”子课题“海上双高—双特高水驱油田提高采收率油藏关键技术”(KJGG2021-0501)
中国海油重大科技专项“渤海油田强化水驱及增产挖潜技术”(CNOOC-KJ135 ZDXM 36 TJ 01 TJ)。
关键词
注水井
吸水能力
主成分分析
粒子群算法
BP神经网络
water injection well
water absorption capacity
principal component analysis
particle swarm optimization algorithm
BP neural network
