Basic concepts of neural networks and deep learning and their applications for pipeline damage detection

Sina Razvarz, Raheleh Jafari, Alexander Gegov

Research output: Chapter in Book/Report/Conference proceedingChapter (peer-reviewed)peer-review


Pipelines have been extensively implemented for the transport of natural gas and liquid petroleum over large distances as they are safe, convenient, and more economical. However, there are several kind of damages that might happen to the pipeline, for instance erosion, breaking, and dent. Thus, if these faults are not correctly refit will cause significant pipeline demolitions which have a tremendous impact on the environment. Deep learning approaches assist operators to recognize the damages in pipelines in the earliest phases so that they can have a good time for discussion on how to solve the problem and gathering information. In this chapter, some types of threats which usually occur in pipelines are introduced. Moreover, early detection of pipeline threats using deep learning methods combined with classification methods are studied for pipeline safety and damage prevention.

Original languageEnglish
Title of host publicationFlow Modelling and Control in Pipeline Systems
Subtitle of host publicationA Formal Systematic Approach
Number of pages19
ISBN (Electronic)9783030592462
ISBN (Print)9783030592455, 9783030592486
Publication statusPublished - 2 Oct 2020

Publication series

NameStudies in Systems, Decision and Control
ISSN (Print)2198-4182
ISSN (Electronic)2198-4190

Cite this