Significant advancements and application of Artificial Intelligence in various fields is becoming a reality. The 4th Industrial Revolution is taking place. However, oil and gas industry being conservative in terms of safety is not really looking forward to the changes. The content consists of three main applications:

1-Deep Reinforcement Learning for control of Nonlinear Distillation: The Distillation Column that serves as transforming crude oil to various petrochemical substances such as gasoline, naphtha, jet fuel, etc. which is related to process engineering is modeled; The system is non-linear with the correlation between equations that makes it good case for the application of deep reinforcement learning. The control goal is to follow the distillation composition output between 0.98-0.96 of the first tray i.e. produce the desired amount of specific petrochemical substance in the least cycle time and reach the set in the allowed control effort allowance. Typically, a PI or MPC controller is used for this control task in refinery plants, I will compare the deep reinforcement learning controller of type advantage actor critic(A2C) results with the typical PI and MPC controller’s outputs, the training process, and model evaluation.

2-Anamoly detection in a typical distillation tower’s numerous sensors read:

The dataset of several measurement units such as temperature, flow rate, and pressure are available in the dataset. Using principal component analysis I will predict anomalies in sensor reads, discussing the power of model by metrics.

3-Deep reinforcement Learning for operator’s vigilance evaluation in static and dynamic mode: The task is to train a model which can distinguish tiredness of facial expressions be it in an image or a video using Yolo and providing my own dataset for training from scratch.

By investigating the application of AI, specifically Deep Reinforcement Learning, in the oil and gas industry, this work aims to provide valuable insights into the potential benefits and challenges of integrating AI systems into critical processes such as distillation control, instrumentation safety, and operators' vigilance. The study outcomes are expected to contribute to the advancement of AI technology in the oil and gas sector

Abstract/11

Chapter 1: Introduction/13

Chapter 2: Literature Overview/21

Chapter 3: Theoretical Foundations/27

Chapter 4: System Architecture and Integration/37

Chapter 5:  Methodology/45

Chapter 6: A2C and PPO Deep Reinforcement Learning Controller of Distillation Column/51

Chapter 7: Anomaly Detection in oil & gas plant sensor reads/71

Chapter 8: Deep Reinforcement Learning of Operator’s Vigilance Evaluation/85

Chapter 9: Experimental Results, Simulations and Analysis/105