Research on the Classification Model of Gas Well Water Production Driven by Production Data of Shale Gas Cage Type Throttle Valve

Water production from gas wells is an important factor affecting gas well productivity. The excessive fluctuation of pressure after the wellhead water flows interfere with the pressure control gas production process, making it difficult to accurately control the wellhead pressure as a control target and thus affecting productivity, this paper carried out research on shale gas cage type throttle valve production data-driven gas well water production classification model. Firstly, the pre valve temperature, pre valve pressure, post valve temperature, and post valve pressure data from 6 wells on site were collected and preprocessed. Secondly, 39 kinds of data classification algorithms such as precise tree, linear discrimination, logistic regression, and Gaussian naive Bayes were used to predict the water production of the cage type throttle valve. Finally, the judgment results are directly used in the cage type throttle valve control system to solve the problem of real-time synchronization of data caused by the wellhead metering mechanism. The research results show that due to the complex characteristics of gas-liquid two-phase throttling at the wellhead of shale gas wells, the adaptability of traditional statistical classification models to the judgment of shale gas well water production is different. Through experimental verification, the optimized KNN classification model is more suitable for shale gas cage throttle valve production data processing. Based on data-driven methods, this paper can directly determine the water production situation of the throttle valve, and thus achieve real-time adjustment, laying the foundation for remote control of the throttle valve. The research results have important engineering value for improving the production efficiency and automation level of shale gas wells in Sichuan.