A Review of Deformations Prediction for Oil and Gas Pipelines Using Machine and Deep Learning

Pipelines are fundamental for conveying oil and gas, but deteriorations can have many effects, such as pollution and property deformities. Therefore, preserving pipeline integrity is important for a secure and sustainable energy provider. The fast development of Machine Learning (ML) methods gives a beneficial possibility to build predictive models that can efficiently resolve these complex problems. This review paper principally emphasizes applying deep learning (DL) and ML methods for estimating pipeline deformations in oil and gas production. The paper analyzes studies in this area, proposing a consistent discussion and determining the reasons and difficulties related to employing ML and DL for estimating deformations in pipelines. This review describes characteristics of ML algorithms often applied, presenting arguments. Based on a review, it is found that ML and DL procedures can precisely predict oil and gas pipeline deformations compared to typical techniques. Assessing ML predictive methods employed on reservoir or synthetic data for pipeline deformations can assist in developing monitoring tools for pipelines, reducing cost and time. It is expected that the development of this review will help comprehend the existing research gaps and give alternatives for other people who want to predict oil and gas pipeline deformations.