• 대한전기학회
Mobile QR Code QR CODE : The Transactions of the Korean Institute of Electrical Engineers
  • COPE
  • kcse
  • 한국과학기술단체총연합회
  • 한국학술지인용색인
  • Scopus
  • crossref
  • orcid
Title Detection of Weld Defects in Power Components Using the Attention-Enhanced YOLOv8 Model
Authors 에르셀릭 우우르(Ugur Ercelik) ; 김거식(Keo Sik Kim) ; 박형준(Hyoung-Jun Park) ; 김경백(Kyungbaek Kim)
DOI https://doi.org/10.5370/KIEE.2026.75.2.369
Page pp.369-375
ISSN 1975-8359
Keywords Computer Vision; Weld Defect Detection; Industrial AI; Channel Attention; Spatial Attention; YOLOv8
Abstract In transformer power equipment, major welding defects often cause oil leakage due to flaws such as porosity, spatter, undercuts, and overlaps, as well as cracks and leakage resulting from frequent electromagnetic vibrations. These issues typically arise from reduced weld quality, component fatigue, thermal overload, and long-term operational conditions, ultimately leading to shorter equipment lifespan and serious safety risks. To address these challenges, in this paper, we propose an improved YOLOv8 model with a Convolutional Block Attention Module(CBAM) to detect and resolve multiple types of small defects on weld beads. The CBAM module is integrated into the neck part of the YOLOv8 network to enhance target features from both channel and spatial dimensions. The proposed YOLOv8-CBAM model demonstrated the most balanced performance across all metrics, reaching the highest precision (75.2%) and recall (77.6%). Although its mAP@50 (81.0%) was slightly lower than YOLOv5, it still outperformed YOLOv10 and YOLOv11, and its competitive mAP@50-95 (49.0%) confirmed its robustness in detecting defects of varying sizes.