Scholarlink

<p>Additive manufacturing (AM) fabricates physical objects by layering materials from a 3D digital model. In metallic AM, the melt pool represents a region of superheated molten material and plays a critical role in determining part quality. Monitoring this region has proven valuable for downstream analytics such as defect detection and microstructure prediction. Melt pool signatures, represented through various modalities (e.g., visual, thermal, acoustic) and metrics (e.g., temperatures, gradients, rates), capture essential process patterns. However, the diversity of materials, process parameters, and sensing configurations across AM systems has limited standardization. To address this, we introduce Melt-Pool-Kinetics, a curated dataset compiled from 32 datasets across 23 sources, totaling 1.9 TB of raw data and released as a 48.6 GB HDF5 collection. Images were processed using cropping, centering, resizing, grayscaling, denoising, and debayering techniques. The dataset can support machine learning applications for <i>in-situ</i> monitoring, process optimization, and control. The dataset is structured into different levels, such as raw, processed, and diverse subsets. The dataset enables future expansion as new melt pool data becomes available.</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 multi-source melt pool compilation for vision-based analytics applications in additive manufacturing

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月19日

查看全文
  • William Jabbour,
  • Mutahar Safdar,
  • Jiarui Xie,
  • Yaoyao Fiona Zhao

摘要

Additive manufacturing (AM) fabricates physical objects by layering materials from a 3D digital model. In metallic AM, the melt pool represents a region of superheated molten material and plays a critical role in determining part quality. Monitoring this region has proven valuable for downstream analytics such as defect detection and microstructure prediction. Melt pool signatures, represented through various modalities (e.g., visual, thermal, acoustic) and metrics (e.g., temperatures, gradients, rates), capture essential process patterns. However, the diversity of materials, process parameters, and sensing configurations across AM systems has limited standardization. To address this, we introduce Melt-Pool-Kinetics, a curated dataset compiled from 32 datasets across 23 sources, totaling 1.9 TB of raw data and released as a 48.6 GB HDF5 collection. Images were processed using cropping, centering, resizing, grayscaling, denoising, and debayering techniques. The dataset can support machine learning applications for in-situ monitoring, process optimization, and control. The dataset is structured into different levels, such as raw, processed, and diverse subsets. The dataset enables future expansion as new melt pool data becomes available.