Scholarlink

<p>Accurate detection of corrosion severity in stud connections is essential for maintaining the safety and reliability of skid-mounted equipment. Traditional inspection methods are inefficient and labor intensive, limiting their practical application. To address these challenges, this study proposes a corrosion grade detection method based on an improved YOLOv8 deep learning model. First, the corrosion grade of flange stud connections is classified by studying and analyzing the model to match the actual corrosion situation. Second, to enhance model efficiency, a lightweight DWGhostConv module is introduced to reduce parameters and computational cost while maintaining strong feature extraction capability. The network is further optimized with the DWGhostC2f module, improving structural efficiency. Experimental results on a custom corrosion dataset demonstrate that the proposed DWGhost-YOLOv8 achieves comparable accuracy to models such as YOLOv8n, YOLOv5n, and YOLOv6, with differences in precision, recall, and mean average precision (mAP) within 2%. Meanwhile, model parameters, computational complexity, and model size are reduced by 20.4, 15.6, and 17.7%, respectively. The experimental results highlight the potential of this model as a lightweight, accurate, and practical solution for intelligent corrosion detection in industrial environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Corrosion grade detection of flange stud connections based on improved YOLOv8

  • 期刊论文

  • 发表日期：

    2025年10月3日

查看全文
  • Meng Wang,
  • Biao Huang

摘要

Accurate detection of corrosion severity in stud connections is essential for maintaining the safety and reliability of skid-mounted equipment. Traditional inspection methods are inefficient and labor intensive, limiting their practical application. To address these challenges, this study proposes a corrosion grade detection method based on an improved YOLOv8 deep learning model. First, the corrosion grade of flange stud connections is classified by studying and analyzing the model to match the actual corrosion situation. Second, to enhance model efficiency, a lightweight DWGhostConv module is introduced to reduce parameters and computational cost while maintaining strong feature extraction capability. The network is further optimized with the DWGhostC2f module, improving structural efficiency. Experimental results on a custom corrosion dataset demonstrate that the proposed DWGhost-YOLOv8 achieves comparable accuracy to models such as YOLOv8n, YOLOv5n, and YOLOv6, with differences in precision, recall, and mean average precision (mAP) within 2%. Meanwhile, model parameters, computational complexity, and model size are reduced by 20.4, 15.6, and 17.7%, respectively. The experimental results highlight the potential of this model as a lightweight, accurate, and practical solution for intelligent corrosion detection in industrial environments.