Scholarlink

<p>The straw granulator serves as a critical device for processing straw biomass resources, with channel blockages being a frequent challenge during straw particle production. Channel blockages significantly compromise production efficiency and damage equipment safety, potentially leading to severe mechanical damage. This study presents the SGB-YOLOv5 straw granulator blockage monitoring system, engineered to accurately identify the stacked states of strip-like materials amidst dynamic multi-object interference. By counting straw particles, the system monitors the blockage state of the granulator and triggers an emergency stop upon reaching the blockage alarm threshold. Leveraging a robust datasets, the SGB-YOLOv5 monitoring algorithm extracts visual feature maps via a feature extraction network, which are subsequently analyzed. Post-training evaluations revealed that SGB-YOLOv5 attained a MAP value of 97.5%, with a per-instance decision time of 0.011&#xa0;s, outperforming other advanced network models and satisfying accuracy standards for practical deployment. Finally, a comprehensive blockage monitoring system for straw granulators was tested to determine the confidence thresholds, ensuring accurate detection within a range of 0.47–0.84. Furthermore, a production test was conducted based on this, validating its effectiveness and accuracy, to some extent, it estimated the production output of the straw granulator. The findings confirm that the blockage monitoring system effectively detects and issues alerts for the pellet blockage states in straw granulators.</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

SGB-YOLOv5: straw granulator blockage monitoring system

  • 期刊论文

  • 发表日期：

    2025年3月7日

查看全文
  • Haoyang Tong,
  • Dongyang Gao,
  • Zhixu Wang,
  • Longlong Feng,
  • Yue Li,
  • Xuewei Bai

摘要

The straw granulator serves as a critical device for processing straw biomass resources, with channel blockages being a frequent challenge during straw particle production. Channel blockages significantly compromise production efficiency and damage equipment safety, potentially leading to severe mechanical damage. This study presents the SGB-YOLOv5 straw granulator blockage monitoring system, engineered to accurately identify the stacked states of strip-like materials amidst dynamic multi-object interference. By counting straw particles, the system monitors the blockage state of the granulator and triggers an emergency stop upon reaching the blockage alarm threshold. Leveraging a robust datasets, the SGB-YOLOv5 monitoring algorithm extracts visual feature maps via a feature extraction network, which are subsequently analyzed. Post-training evaluations revealed that SGB-YOLOv5 attained a MAP value of 97.5%, with a per-instance decision time of 0.011 s, outperforming other advanced network models and satisfying accuracy standards for practical deployment. Finally, a comprehensive blockage monitoring system for straw granulators was tested to determine the confidence thresholds, ensuring accurate detection within a range of 0.47–0.84. Furthermore, a production test was conducted based on this, validating its effectiveness and accuracy, to some extent, it estimated the production output of the straw granulator. The findings confirm that the blockage monitoring system effectively detects and issues alerts for the pellet blockage states in straw granulators.