The application of deep learning in pipeline inspection: current status and challenges

ABSTRACT

Pipelines are a primary route of transportation for essential energy sources such as oil and gas, and the inspection and assessment of their operational status is vital to the health of the industry. In addition to traditional manual inspection techniques, emerging Deep Learning (DL) methods have promoted the development of intelligent pipeline inspection. Using DL techniques, oil and gas pipelines can be automatically, efficiently and accurately inspected and evaluated, which is important for improving pipeline safety and reducing accident risks. This paper reviews the application of DL to damage detection, identification and classification of pipelines. Firstly, a review of commonly used DL methods is given, and then the main application scenarios of current DL in pipeline inspection are discussed. Finally, the advantages and limitations of the existing detection methods are given.

KEYWORDS:

• Pipeline; oil and gas
• deep learning (DL); inspection; monitoring

Data availability statement

Data openly available in public repositories. The data that support the findings of this review are openly available in Google Scholar (scholar.google.com), Scopus (scopus.com), Web of Science (webofknowledge.com), Cnki (cnki.net).

Disclosure statement

No potential conflict of interest was reported by the author(s).