基于数据驱动的钻井过程气侵工况预测方法

钻井过程中异常工况的及时诊断是保证快速安全钻井的重要手段。笔者以提前预判气侵工况为目标，应用自适应观测器，以实际立压回压数据作为观测器输入，对未知的井底流量、井底压力数据进行估计。井底压力数据估计值与实际测得的井底压力数据吻合良好。将井底流量数据的估计值与井口实际出口流量数据做差，对该差值和实际立压、回压以及入口流量数据一起进行独立主元分析，提取该组数据的独立元、确定正常工况下的统计量控制限，并提取工作状态下的数据独立元的统计量，将该统计量数值与控制限进行对比，用于气侵工况诊断。经过实际钻井数据验证，该方法能够更准确判断正常工况及异常工况、并判断气侵的发生。

Data driven prediction method for gas cut in drilling process

Timely diagnosis of abnormal working conditions in drilling process is an important means of ensuring fast and secure drilling. Aiming at predicting the gas cut in advance, a self-adaptive observer is used to estimate the unknown bottom hole flow rate and pressure data by inputting the actual standpipe pressure and return pressure data into the observer. The estimated bottom hole pressure data is well consistent with the actual measured bottom hole pressure data; the difference between the estimated bottom hole flow data and the actual wellhead flow data can be obtained; an independent component analysis is performed on such a difference, the actual standpipe pressure and return pressure, and the wellhead flow data. This study extracts independent elements of the data set, determines the statistical control limit under normal working conditions, obtains the statistics of independent elements in the working state, and performs a comparison between the statistics and the statistical control limit, aiming to detect the work conditions for gas cut. Verified by the real drilling data, the above method can accurately judge the normal or abnormal working conditions; compared with traditional methods, it can more timely and precisely predict the occurrence of gas cut.