Hybrid Data Fusion in Guided Wave Tomography: Bridging Defect Detection and Life Assessment
e-Journal of Nondestructive Testing · 2026-07 · 31 (7)
DOI: 10.58286/33329 ↗At a glance
Journal article in e-Journal of Nondestructive Testing, 2026-07. Written by Igor Sakhabiev and 1 co-authors.
Abstract (original language)
This study proposes a hybrid Ultrasonic Guided Wave Tomography (UGWT) framework for inspecting Containment Liner Plates (CLPs) in nuclear power plants. The integrity of carbon steel liners is essential for containment leak-tightness, while conventional reconstruction techniques often suffer from coherent noise and signal interference. To address this, a hybrid data fusion approach is introduced, combining unsupervised Principal Component Analysis (PCA) for statistical anomaly detection with a supervised Support Vector Machine (SVM) for probabilistic classification. To overcome defect data scarcity, geometric data augmentation is employed to generate a synthetic training dataset from limited experimental mock-ups. A fused Damage Index is constructed by integrating the statistical variance extracted by PCA with the probabilistic output of the SVM. Experimental results demonstrate that the proposed fusion strategy improves the DI matrix correlation with the ground truth by up to 78% compared to conventional approaches. The improved input damage map leads to enhanced tomographic reconstructions with accurate defect localization and reduced artifacts.
Cite
APA
Sakhabiev, I., & Cho, Y. (2026). Hybrid Data Fusion in Guided Wave Tomography: Bridging Defect Detection and Life Assessment. e-Journal of Nondestructive Testing, 31(7). https://doi.org/10.58286/33329
Vancouver
Sakhabiev I, Cho Y. Hybrid Data Fusion in Guided Wave Tomography: Bridging Defect Detection and Life Assessment. e-Journal of Nondestructive Testing. 2026;31(7). doi:10.58286/33329
BibTeX
@article{Sakhabiev2026,
author = {Sakhabiev, Igor and Cho, Younho},
title = {Hybrid Data Fusion in Guided Wave Tomography: Bridging Defect Detection and Life Assessment},
journal = {e-Journal of Nondestructive Testing},
year = {2026},
volume = {31},
number = {7},
doi = {10.58286/33329}
}
EUVIMED