Leakage detection in underwater oil and natural gas pipelines using convolutional neural networks

Abstract

Underwater oil and natural gas pipelines are an underwater transport infrastructure known to be reliable, fast, and efficient, preferred for the transmission of energy to far distances. The rapid and continuous increase in demand for energy due to population growth, industrial developments, and global growth requires economic and environmental solutions for the safe transmission and control of energy sources such as oil and natural gas. These lines are damaged due to their work in corrosive ambient conditions, natural elements such as sudden change of air and water temperatures, tectonic activities, and external elements such as blows caused by fishing equipment and military exercises. Therefore, it is necessary to determine the damages without requiring more hardware, saving time, and cost. In this study, underwater oil and natural gas pipelines were detected using convolutional neural networks and the detection performance of artificial neural network was analyzed. Underwater pipelines are detected using convolutional neural networks with 97.63% accuracy. A reliable, fast, efficient, controlled, and sustainable model is established to prevent potential damage to underwater pipelines from becoming an environmental threat to water and air pollution and living creatures in the underwater ecosystem with this study.