摘要
随着油田的不断开采,油田的产量预测也变得越来越重要。目前有许多基于机器学习的预测方法,但大多数都不能给出具体的预测模型。提出一种基于Lasso算法的预测方法,结合现场生产数据,选取一系列相关特征参数,通过对参数数据的分析,初步选取各个参数的函数形式,然后利用Lasso算法得到最终的预测模型,达到预测产量的目的。现场试验表明,该方法得到的预测模型比较准确,可解释性强,且预测精度高,可以应用于矿场产量预测。
As the continuous mining of oil fields,production forecast for oil fields is becoming more and more important.There are many prediction methods based on machine learning,but most of them cannot give a specific prediction model.A prediction method based on Lasso algorithm was proposed.Based on the field production data,a series of related feature parameters were selected.Through the preliminary analysis of the sample data,the best function form of each parameter was selected,and the prediction model was obtained by Lasso algorithm.Finally,the purpose of predicting production was achieved by the prediction model.The field tests show that the prediction model obtained by this method is accurate and is highly explanatory.Besides,prediction results with high accuracy and can be applied to mine production forecasting.
作者
谷建伟
周鑫
王硕亮
GU Jian-wei;ZHOU Xin;WANG Shuo-liang(School of Petroleum Engineering,China University of Petroleum(EastChina),Qindao 266580,China;School of Energy Resources,China University of Geosciences(Beijing),Beijing 100088,China)
出处
《科学技术与工程》
北大核心
2020年第26期10759-10763,共5页
Science Technology and Engineering
基金
国家科技重大专项(2016ZX05011-001)。
关键词
产量预测
机器学习
Lasso算法
函数选取
production forecast
machine learning
Lasso algorithm
function selection
