Determining the Place of Depressurization of Underground Pipelines (Gas Pipelines): New Solutions in Industry based on Thermal Image Analysis Using Computer Vision
Determining the Place of Depressurization of Underground Pipelines (Gas Pipelines): New Solutions in Industry based on Thermal Image Analysis Using Computer Vision
Author(s): Roman Mysiuk, Iryna Mysiuk, Volodymyr Yuzevych, Grzegorz PawłowskiSubject(s): Energy and Environmental Studies, ICT Information and Communications Technologies, Transport / Logistics
Published by: Altezoro, s. r. o. & Dialog
Keywords: depressurization of gas pipelines; gas; corrosion; computer vision; image mining;
Summary/Abstract: An analysis of the analytical ratios of the mathematical model, which characterizes the development processes of a corrosion cavern on the surface of an underground metal pipeline, which is placed in the environment of moist soil with an electrolyte solution, is performed. A neural network method for estimating the main informative parameters for determining the place of gas depressurization on the surface of an underground pipe and an expression for calculating the change in gas pressure around a crack after its formation have been developed. The principles of determining the limit values of the parameters of the “pipe-cathodic protection” system are formulated, considering the metal's quality and strength criteria at the top of the cavern. Depressurization causes fluid to flow from the pipeline to the surface. Thermal imaging devices make it possible to detect the place of damage to the pipeline based on the temperature properties of the surrounding objects. Thermal imaging can be used to analyze the location of a fluid leak or warn of it using computer vision. Thus, preventing an accident or even a catastrophe in the pipeline. In the work, the colour gamuts of the thermal image in the places of depressurization are considered, and the regularities of detecting damaged sections of the pipeline are established.
Journal: Traektoriâ Nauki
- Issue Year: 8/2022
- Issue No: 10
- Page Range: 1001-1010
- Page Count: 10
- Language: English