We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Deep Learning Approach for Non-destructive Radiography Testing of Piping Welds.
- Authors
EL HARRAK, Adil; RHAZZAF, Mohamed; MOUMENE, Imane
- Abstract
Radiographic testing is the most common method of non-destructive testing to detect discontinuities in industrial piping welds. Human interpretation of radiographic films is time-consuming and requires high degrees of expertise. Despite the significant advancements of deep learning techniques in related fields, such as medical radiography, prior research endeavors focused on weld discontinuities were constrained by the limitations stemming from a lack of training data and their inadequate representation of real-world conditions in the field. This paper introduces a comprehensive system that automatically detects welding zones, assesses film quality, and classifies weld discontinuities for the piping process. The proposed framework demonstrates superior generalization capabilities that bypass a single industry or piping size. The key advantages of our technique lie in its enhanced accuracy, rapid processing, and automatic interpretation of welding films across a wide range of image qualities. Consequently, it achieves remarkable detection and classification accuracy, offering substantial benefits for welding inspection and quality assessment.
- Subjects
DEEP learning; WELDED joints; MEDICAL radiography; RADIOGRAPHIC films; WELDING
- Publication
IAENG International Journal of Computer Science, 2024, Vol 51, Issue 4, p378
- ISSN
1819-656X
- Publication type
Article