Application of Deep Learning in Prediction of Linkage Relationship between Wells

Abstract

Aiming at the prediction problem of interconnection relationship between injection and production wells, a prediction method based on deep learning is proposed. The network structure of injection production wells is established with the center of water injection well location, which is simplified as the flow distribution problem under the conditions of well number, well spacing, angle and physical properties. Through numerical simulation, the distribution amount of various water injection between wells varies with time, and the learning model is optimized by results and parameters optimization, it can achieve high calculation accuracy and has broad application potential.