基于AdaBoost-SVM算法的油井反向压驱效果预测方法  

A Prediction Method of Oil Well Reverse Enhanced Oil Recovery by Fracturing and Percola Based on AdaBoost-SVM Algorithm

作　　者：王梅[1] 田野 杨二龙[2] 邓庆军[3] 阚春玲[3] WANG Mei;TIAN Ye;YANG Er-long;DENG Qing-jun;KAN Chun-ling(School of Computer and Information Technology,Northeast Petroleum University,Daqing 163318,China;Key Laboratory of Enhanced Oil Recovery in Ministry of Education,Northeast Petroleum University,Daqing 163318,China;Production Plant No.1,Petro China Daqing Oilfield,Daqing 163318,China)

机构地区：[1]东北石油大学计算机与信息技术学院,黑龙江大庆163318 [2]东北石油大学提高采收率教育部重点实验室,黑龙江大庆163318 [3]大庆油田有限责任公司第一采油厂,黑龙江大庆163318

出　　处：《数学的实践与认识》2023年第2期128-135,共8页Mathematics in Practice and Theory

基　　金：国家自然科学基金(51774090);黑龙江省自然科学基金(F2018003)。

摘　　要：预测油井的反向压驱增油效果是反向压驱选井选层工作的重要环节。针对反向压驱历史数据规模有限的问题,提出一种基于AdaBoost-SVM的油井反向压驱增油效果预测方法:应用SMOTE算法对压驱历史数据进行过采样,并建立AdaBoost-SVM反向压驱增油效果预测模型,以对油井的压驱效果进行预测.在油井上的实际应用结果表明,该预测方法能够有效预测油井应用反向压驱工艺的增油效果,为进一步的选井选层工作提供参考。Predicting the oil-increasing effect of reverse fracturing is an important part in the selection of wells and layers for reverse fracturng.Aiming at the limited scale of historical data of reverse fracturing,a method for predicting the effect of reverse fracturing in oil wells based on AdaBoost-SVM is proposed.The SMOTE algorithm was used to oversample the historical data of fracturing,and the AdaBoost-SvM reverse fracturing effect prediction model was established to predict the fracturing effect of oil wells.The actual application results in oil wells show that the prediction method can effectively predict the oil-increasing effect of fracturing technology in oil wells,and provide a reference for further well selection and layer selection.

关 键 词：支持向量机 ADABOOST算法 SMOTE 过采样 反向压驱 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TE357[自动化与计算机技术—控制科学与工程]