Prediction system for water-producing gas wells using edge intelligence

Abstract

Fluid accumulation in gas wells lead to a sudden drop in production and is a problem. This paper proposes a prediction system for water-producing gas wells using edge intelligence. The method is characterized by use of edge intelligence to realize automatic collection of gas well data, effusion analysis, data backup, upload, production dynamic monitoring, and other functions. The Pearson correlation coefficient was employed to analyze the relevant parameters of liquid accumulation in gas wells. A collaborative discrimination approach utilizing convolutional neural networks and ensemble learning models was implemented to predict liquid accumulation state of gas wells. The model achieved an accuracy rate as high as 98.2%, which aligns well with actual liquid accumulation state of gas wells compared to traditional physical models. In addition, suggestions for extraction and drainage plans are provided based on the degree of liquid accumulation in gas wells, thereby realizing automated and refined management of liquid accumulation wells in the field. This method provides effective theoretical support and reasonable guidance for the efficient development and overall decision-making in gas fields.