Optimal Control Method of Oil Well Production Based on Cropped Well Group Samples and Machine Learning

Most traditional injection-production optimization methods that treat the entire oil reservoir as a whole require re-optimization when facing new reservoirs, which is not only time-consuming but also does make full use of historical experience information. This study decomposes into independent basic production units to increase sample size and diversity utilizes image enhancement techniques augment number samples. Two frameworks based on convolutional neural networks (CNNs) are employed recommend optimal control strategies for inputted well groups. Framework 1 uses bottom hole pressure (BHP) variable trains CNN with BHP obtained by reinforcement learning algorithms labels. 2 saves corresponding revenue (NPV) during groups features NPV The in this framework capable directly outputting according strategies. particle swarm algorithm (PSO) used generate call predict development effects until PSO converges strategy. experimental results demonstrate CNN-based outperform PSO-based terms accuracy computational efficiency. achieves an output 87% predicting groups, while 78%. Both exhibit fast running times, each iteration taking less than s. provides more effective accurate method optimizing reservoirs decomposing using construct framework, great significance real-time wells fields.