Scholarlink

<p>The formation and propagation of fracture fields under explosive loading significantly influence rock rupture morphology and blasting effectiveness. However, the transient nature of the explosive event and the complex coupling of blast-induced stresses pose considerable challenges in accurately monitoring crack propagation during experimental investigations. This study introduces, for the first time, an innovative approach for precise detection of explosion-induced cracks using Distributed Fiber Optic Sensing (DFOS) integrated with deep learning techniques. Initially, the feasibility of DFOS for monitoring crack propagation under explosive loading is validated. Subsequently, millimeter-level precision in tracking crack extension paths is achieved through polar coordinate transformation and color mapping techniques. A novel strain profile image generation approach is also developed to facilitate the visual representation of DFOS strain data. The spatial distribution of cracks is successfully identified using the YOLOv5 deep learning model, achieving precision, recall, and F1 scores of 0.927, 0.933, and 0.930, respectively, demonstrating the high efficacy of the proposed detection method. This approach eliminates the need for traditional visual imaging and presents a pioneering solution for detecting internal fracture fields induced by explosions.</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

A New Fine Detection Approach for Explosion Cracks Based on Distributed Fiber Optic Sensing and Deep Learning

  • 期刊论文

  • 发表日期：

    2025年4月27日

查看全文
  • Jin Li,
  • Renshu Yang,
  • Liyun Yang,
  • Jinjing Zuo,
  • Xiang Zhang,
  • Yuanyuan You,
  • Yiqiang Kang

摘要

The formation and propagation of fracture fields under explosive loading significantly influence rock rupture morphology and blasting effectiveness. However, the transient nature of the explosive event and the complex coupling of blast-induced stresses pose considerable challenges in accurately monitoring crack propagation during experimental investigations. This study introduces, for the first time, an innovative approach for precise detection of explosion-induced cracks using Distributed Fiber Optic Sensing (DFOS) integrated with deep learning techniques. Initially, the feasibility of DFOS for monitoring crack propagation under explosive loading is validated. Subsequently, millimeter-level precision in tracking crack extension paths is achieved through polar coordinate transformation and color mapping techniques. A novel strain profile image generation approach is also developed to facilitate the visual representation of DFOS strain data. The spatial distribution of cracks is successfully identified using the YOLOv5 deep learning model, achieving precision, recall, and F1 scores of 0.927, 0.933, and 0.930, respectively, demonstrating the high efficacy of the proposed detection method. This approach eliminates the need for traditional visual imaging and presents a pioneering solution for detecting internal fracture fields induced by explosions.