基于无迹卡尔曼滤波的井筒压力实时校正模型  

Real-Time Correction Model of Wellbore Pressure based on Unscented Kalman Filter

作　　者：夏顺雷 李军[1,2] 柳贡慧 杨宏伟[1] 罗鸣[3] 李文拓 Xia Shunlei;Li Jun;Liu Gonghui;Yang Hongwei;Luo Ming;Li Wentuo(College of Petroleum Engineering,China University of Petroleum(Beijing);College of Petroleum,China University of Petroleum(Beijing)at Karamay;Zhanjiang Branch of CNOOC(China)Limited)

机构地区：[1]中国石油大学(北京)石油工程学院,北京市昌平区102249 [2]中国石油大学(北京)克拉玛依校区石油学院 [3]中海石油(中国)有限公司湛江分公司

出　　处：《石油机械》2022年第9期10-18,共9页China Petroleum Machinery

基　　金：国家自然科学基金重点项目“深水油气钻采井筒压力控制基础研究”(51734010);中国博士后科学基金项目“深井复杂地层智能井控实时决策与自适应协同控制方法”(2021M693494);国家自然科学基金青年科学基金项目“深井复杂地层智能井控井筒压力预测模型与优化控制方法”(52104012)。

摘　　要：为提高井筒压力预测精度和实时性,将无迹卡尔曼反演算法引入井筒压力预测模型中,建立了基于无迹卡尔曼滤波的井筒压力实时校正模型。该模型以无迹卡尔曼滤波为反演算法基础,以实测数据为输入参数,实时反演计算井筒摩阻校正系数,进而提高井筒压力预测精度和实时性。再分别使用井底压力模拟数据和井底压力实测数据对该模型进行研究。研究结果表明:建立的井筒压力校正模型不仅能够反演得到井筒摩阻校正系数,提高井筒压力预测精度,而且能够根据井下情况实时调整反演结果;利用摩阻校正系数能够提高井筒任意点压力和任意时刻井筒压力剖面预测精度,弥补井下工具单点测量的不足;模型预测精度受反演次数影响较大,模拟条件下,反演初始阶段,精度随反演次数增加而增加,反演次数达到200次后,反演精度趋于稳定。研究结果可预测井筒压力提供理论支撑。In order to improve the accuracy and real-time performance of wellbore pressure prediction,the unscented Kalman inversion algorithm was introduced into the wellbore pressure prediction model and a real-time correction model of wellbore pressure based on unscented Kalman filter was established.In this model,unscented Kalman filter is used as the inversion algorithm and the measured data is used as the input parameters to invert the wellbore friction correction coefficient in real time.Thus,the prediction accuracy and real-time performance of wellbore pressure prediction are improved.Bottom hole pressure simulation data and measured bottom hole pressure data were used respectively to study this model.The results show that,by using the wellbore pressure correction model,not only the wellbore friction correction coefficient can be calculated by inversion so as to improve the prediction accuracy of wellbore pressure,but also the inversion results can be adjusted in real time according to the downhole conditions.By means of friction correction coefficient,the prediction accuracy of wellbore pressure at any point and wellbore pressure profile at any time can be improved,which makes up for the deficiency of single-point measurement of downhole tools.The prediction accuracy of the model is greatly affected by the number of inversions.Under simulation conditions,the accuracy increases with the number of inversions in the early stage of inversion;while after the number of inversions reaches 200,the inversion accuracy tends to be stable.The research results provide theoretical support for wellbore pressure prediction.

关 键 词：井筒压力 无迹卡尔曼滤波 反演校正 实时更新 摩阻校正系数 

分 类 号：TE934[石油与天然气工程—石油机械设备]