Prediction of Wellbore Liquid Loading in Gas Well Based on RBF Neural Network

In the later stage of gas field development, the number of gas wells with liquid loading gradually increases, and the influence of fluid loading on gas production is also increasing. How to accurately predict the liquid loading state in the gas production period is of great significance to alleviate the liquid loading in gas wells and improve the recovery rate of gas reservoirs. Based on the influencing factors of wellbore effusion in gas wells, this paper establishes a prediction model of wellbore effusion in gas wells based on the RBF neural network. Finally, the Y1 well in actual production is predicted. The results show that the inclination angle, oil pressure, tubing diameter, daily gas production, and daily water production are the main influencing factors of gas well effusion. The RBF neural network model has the characteristics of fast prediction speed, high precision, and convenient use. The coincidence rate between the predicted results and the actual results is more than 90 %, which can be used to guide the production.