Scholarlink

<p>In light of recent interest in understanding the mechanisms behind liquid metal embrittlement (LME) cracking behaviours, methods have been developed to observe in-situ LME cracking using the half-sectioned resistance spot welding. To facilitate analysis of the collected video footage, an object detection algorithm methodology was developed to detect and measure LME cracking directly from the video frames. The method employed in this experiment uses an unmodified YOLOv9-C model with SORT integrated in the workflow to detect, track, and measure cracks. Results show that YOLOv9 is feasible to be used as a tool to facilitate this laborious task, indicated by a parity plot with R<sup>2</sup> = 0.982, slope of 0.9775 +/- CI and a relative error of 4% to human measurements.</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

Advanced machine vision and machine learning for adaptive control of LME cracking in third-generation AHSS welding

  • 期刊论文

  • 发表日期：

    2026年7月15日

查看全文
  • Hao Gao,
  • JiUng Kim,
  • Reza Bakhtiari,
  • Ali Afzal,
  • Hasan Habib,
  • Hassan Ghassemi-Armaki,
  • Ahmad Aminzadeh,
  • Elliot Biro

摘要

In light of recent interest in understanding the mechanisms behind liquid metal embrittlement (LME) cracking behaviours, methods have been developed to observe in-situ LME cracking using the half-sectioned resistance spot welding. To facilitate analysis of the collected video footage, an object detection algorithm methodology was developed to detect and measure LME cracking directly from the video frames. The method employed in this experiment uses an unmodified YOLOv9-C model with SORT integrated in the workflow to detect, track, and measure cracks. Results show that YOLOv9 is feasible to be used as a tool to facilitate this laborious task, indicated by a parity plot with R2 = 0.982, slope of 0.9775 +/- CI and a relative error of 4% to human measurements.