基于Kriging法的造斜率预测精度提升研究  

Research on Improving the Prediction Accuracy of Build-up Rate Based on Kriging Method

作　　者：潘勇博 张红[1] 冯定[1] 冯云 Pan Yongbo;Zhang Hong;Feng Ding;Feng Yun(School of Mechanical Engineering,Yangtze University;Hubei Engineering Research Center for Oil&Gas Drilling and Completion Tools)

机构地区：[1]长江大学机械工程学院,湖北省油气钻完井工具工程技术研究中心

出　　处：《石油机械》2023年第10期41-49,共9页China Petroleum Machinery

基　　金：国家采油装备工程技术研究中心开放基金资助项目“抽油机井杆管偏磨系统动力学行为研究”(ZBKJ2021-A-02);湖北省技术创新专项(重大项目)“智能油气钻采井眼轨迹控制工具研究”(2019AAA010);长江大学大学研究生院优质课程建设项目(YYZ202212)。

摘　　要：钻探行业中传统的造斜率预测方法存在些许不足,如几何法预测精度较低、力学法计算复杂度过高等,间接提高了钻探成本。为此,将相关性分析法运用于前期研究的Kriging代理模型,并以四川省某个井段的实钻数据为例,对各个影响因素的实钻数据进行归一化处理,使其分布于同一区间内;通过相关性分析法分别求出5个实测因素与造斜率之间的相关系数,确定与造斜率相关性较高的几种因素,并将其作为造斜率的敏感因子;根据相关系数的大小对敏感因子进行加权变换,用加权变换后的敏感因子作为训练样本,重新构建造斜率预测模型并进行造斜率预测,用模型预测误差间接反映模型的预测精度。研究结果表明,相较于单纯用初始数据进行拟合,该方法剔除无关因素后仅用敏感因子做拟合,预测值的均方根误差、最大绝对误差和平均绝对误差分别降低了13.26%、11.53%和17.37%,3种误差平均降低了13.39%,即经过初始数据处理后建立的Kriging代理模型能使预测精度有效提升。所得结论可为造斜率的预测提供一种新的方法。The traditional build-up rate prediction methods in the drilling industry have some shortcomings,such as low prediction accuracy using geometric method and high computational complexity using mechanical method,which indirectly increase drilling costs.Therefore,the correlation analysis method was applied to the Kriging surrogate model studied in the previous stage.Taking the actual drilling data of a hole interval in the Sichuan Basin for example,the actual drilling data of various influential factors were normalized to make them distributed in the same interval.The correlation analysis method was used to calculate the correlation coefficients between the 5 measured factors and the build-up rate respectively,determine the factors having high correlation with the build-up rate,and use them as the sensitive factors of the build-up rate.Base on the size of the correlation coefficient,the sensitive factors were weighted and transformed,which were then used as training samples to reconstruct a build-up rate prediction model to conduct build-up rate prediction,and the prediction error of the model indirectly reflected the prediction accuracy of the model.The research results show that compared with fitting only using initial data,this method eliminates irrelevant factors and only uses the sensitive factors to conduct fitting,the root mean square error,the maximum absolute error and the mean absolute error of the predicted values are reduced by 13.26%,11.53%and 17.37%respectively,and the 3 types of errors are reduced by 13.39%on average,that is,the Kriging surrogate model built after processing of the initial data can effectively improve the prediction accuracy.The conclusions provide a new method for the prediction of the build-up rate.

关 键 词：造斜率 Kriging代理模型 相关性分析 归一化 预测精度 

分 类 号：TE21[石油与天然气工程—油气井工程]