Reinforcement Learning-Based Process Control in the Oil and Gas Industry

This paper explores the application of reinforcement learning for process control in an oil plant, aiming to enhance operational efficiency and stability. The project develops a custom simulation environment, replicating a crude oil separation processing plant. The reinforcement learning algorithms were implemented to manage the entire plant, aiming to maintain stable operation and maximize production. A comparison between traditional control methods and reinforcement learning based control was conducted and the results demonstrate that reinforcement learning can achieve comparable or superior performance to traditional controllers in managing process variables across multiple interconnected components. The paper contributes to the growing body of work in applying reinforcement learning to industrial process control, highlighting its potential to automate complex real-time operations.