Session: 03-04-01 Inline Inspection Performance Session 1.1

Paper Number: 133038

133038 - Matching of Corrosion Features in Multiset Pipeline In-Line Inspection Data Utilizing Relative Point Positions 

Abstract: 

Pipelines utilized for the transmission of oil and gas inevitably face corrosion degradation over their operational lifespan, posing a significant threat to the integrity of the entire system. One effective approach to comprehensively understand the conditions of these pipelines is through pipeline in-line inspection (ILI). ILI allows for the observation of corrosion features across multiple inspections, enabling the calculation of corresponding corrosion growth rates. However, a major challenge arises due to the fact that each ILI employs its own coordinate system to reference the location of detected corrosion features. This discrepancy in coordinate systems introduces a misalignment problem when attempting to directly match corrosion features between two ILIs. Consequently, the matching of corrosion features detected in multiset pipeline ILIs becomes an essential prerequisite for determining corrosion growth. State-of-the-art methods for corrosion defect matching often involve the iterative process of finding the optimal affine transformation or the extraction of individual defect attributes for matching model training. Unfortunately, these processes are not only labor-intensive but also face challenges associated with the collection of substantial amounts of data. In response to this, an automated attribute-free point matching method for corrosion features using ILI data was proposed in this study. This algorithm relies solely on the positions of detected corrosion points and is applicable to complex defect scenarios with high corrosion density. Firstly, triangles are constructed based on corrosion features. Secondly, local matching is employed to identify matching triangle pairs within two ILIs. Subsequently, a global matching strategy is applied to refine the initially identified matches, filtering them based on predefined requirements. Finally, points in the final matched triangle pairs serve as correspondences to determine the registration transformation. Experiments were conducted to validate the efficacy of the proposed method. The results demonstrate its reliability in accurately matching features within pipelines, thereby supporting the integrity and risk assessment of pipeline systems.

Presenting Author: Jiatong Ling The University of British Columbia

Presenting Author Biography: Jiatong Ling is currently pursuing the Ph.D. degree in electrical engineering with The University of British Columbia, Kelowna, BC, Canada. Her current research interests include pipeline integrity assessment, nondestructive inspection and evaluation, and machine learning.

Authors:

Jiatong Ling The University of British Columbia
Zheng Liu The University of British Columbia