Comprehensive Control System for Gathering Pipe Network Operation Based on Reinforcement Learning

Abstract

In the transmission process of crude oil gathering system, water-assisted heat transfer is often used to avoid wax formation, and pipeline temperature and pressure control are often controlled manually. In order to improve control efficiency and save labor cost. In this paper, we propose a DQN-based algorithm. The intensive learning model completes the temperature and pressure control in the pipeline. At the same time, because these two parameters have strong coupling, which affects the global control, this paper focuses on the joint optimization of valve opening and heating furnace and pressure pump. Finally, in order to verify the effectiveness of the system, the simulation control experiment is adopted. The test results show that the system control effect is excellent and the robustness is good.