Scholarlink

<p>Roller tank lugs are the guiding devices in vertical shaft hoisting systems, playing a crucial role in ensuring the stable operation of the hoisting equipment. Existing size measurement methods for roller tank lugs suffer from poor real-time performance and complex processing procedures. To address these issues, this paper proposes a real-time, multi-target size detection technology for roller tank lugs based on YOLOv8. First, the YOLOv8 detection algorithm is utilized to identify and locate the roller tank lugs. The resulting bounding boxes are then associated across consecutive frames using the BoT-SORT algorithm, achieving continuous tracking and identification of multiple roller tank lugs within the field of view. Secondly, based on the generated tracking targets, a size measurement scheme for roller tank lugs is proposed, which can effectively eliminate perspective projection errors. Then, the size measurement module is integrated into the YOLOv8 detection network to achieve end-to-end, real-time size detection of the roller tank lugs. Experimental results demonstrate that the proposed method achieves a detection accuracy of 98.08% on the roller tank lug dataset, with a detection frame rate of 45.9 FPS. Furthermore, the size measurement error of the roller tank lugs is controlled within 3&#xa0;mm, fully satisfying practical engineering requirements.</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

Research on precision measurement methods for dynamic multi-objective roller tank lugs dimensions in shaft hoisting systems

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月10日

查看全文
  • Zong Li,
  • Yucan Shen,
  • Feng Gao,
  • Senliang Gai,
  • Meiling Zhang,
  • Fangfei Cao,
  • Xiang Lu

摘要

Roller tank lugs are the guiding devices in vertical shaft hoisting systems, playing a crucial role in ensuring the stable operation of the hoisting equipment. Existing size measurement methods for roller tank lugs suffer from poor real-time performance and complex processing procedures. To address these issues, this paper proposes a real-time, multi-target size detection technology for roller tank lugs based on YOLOv8. First, the YOLOv8 detection algorithm is utilized to identify and locate the roller tank lugs. The resulting bounding boxes are then associated across consecutive frames using the BoT-SORT algorithm, achieving continuous tracking and identification of multiple roller tank lugs within the field of view. Secondly, based on the generated tracking targets, a size measurement scheme for roller tank lugs is proposed, which can effectively eliminate perspective projection errors. Then, the size measurement module is integrated into the YOLOv8 detection network to achieve end-to-end, real-time size detection of the roller tank lugs. Experimental results demonstrate that the proposed method achieves a detection accuracy of 98.08% on the roller tank lug dataset, with a detection frame rate of 45.9 FPS. Furthermore, the size measurement error of the roller tank lugs is controlled within 3 mm, fully satisfying practical engineering requirements.