Bibliographic Details
| Title: |
Data processing method for evaluating pipe wall thinning in nuclear secondary systems using SVM regression algorithm |
| Authors: |
Seongbin Mun, Young-jin Oh, Sanghoon Lee |
| Source: |
Nuclear Engineering and Technology, Vol 57, Iss 7, Pp 103517- (2025) |
| Publisher Information: |
Elsevier, 2025. |
| Publication Year: |
2025 |
| Collection: |
LCC:Nuclear engineering. Atomic power |
| Subject Terms: |
Pipe wall thickness, Pipe wall thinning, Measurement error, Data processing, Kernel regression, Support vector machine regression, Nuclear engineering. Atomic power, TK9001-9401 |
| More Details: |
Nuclear power plants have been experiencing continuous pipe wall thinning in carbon steel materials used in the secondary system, potentially leading to pipe rupture. To prevent the sudden pipe rupture caused by this thinning, pipe thinning management programs have been conducted which include periodic thickness measurement and remaining life assessment for sampled pipes and fittings. However, periodic thickness measurement data from ultrasonic testing (UT) have a significant range of uncertainty, which can significantly affect the assessed thinning values. Moreover, the uncertainty of the evaluated thinning value is intensified because the amount of thinning of a certain pipe or fitting is defined by its maximum thinning value. Therefore, a data processing method to minimize the effect of thickness measurement uncertainty is crucial to determine more reliable thinning values. In this study, a data processing method based on the support vector machine regression algorithm was proposed, which was adjusted and modified considering the general characteristics of pipe thinning phenomena. Using datasets of thickness measurements constructed by assumed wall thickness shapes and measurement uncertainty, it was confirmed that the proposed method reduces the uncertainty and bias of evaluated thinning values. |
| Document Type: |
article |
| File Description: |
electronic resource |
| Language: |
English |
| ISSN: |
1738-5733 |
| Relation: |
http://www.sciencedirect.com/science/article/pii/S1738573325000853; https://doaj.org/toc/1738-5733 |
| DOI: |
10.1016/j.net.2025.103517 |
| Access URL: |
https://doaj.org/article/37cbdf44d95f49d2921206337ca2d2b7 |
| Accession Number: |
edsdoj.37cbdf44d95f49d2921206337ca2d2b7 |
| Database: |
Directory of Open Access Journals |
