Scholarlink

<p>Thermal spray practitioners have long struggled to define robust plume characteristics process windows—combinations of particle temperature, velocity, and plume geometry that reliably yield acceptable coatings—especially as powder characteristics, equipment wear, and ambient conditions evolve. The <i>AccurasprayHUB</i> addresses this challenge by unifying data capture, structuring, and analysis within a single platform. During an initial lab deployment, the Hub continuously collects summary statistics (mean temperature, velocity, and geometry) for each spray run and automatically enriches them with detailed metadata—stand-off distance, gas and powder flow rates, powder batch IDs, gun maintenance records, and laboratory data when available. These per-run feature vectors are then subjected to density-based clustering to identify the natural “normal” operating regimes without requiring explicit “bad coating” examples. From the resulting high-confidence clusters, the Hub derives multidimensional min–max envelopes that define process windows tailored to specific powders and gun health states. By delivering reproducible data pipelines, transparent feature engineering, and modular analytics, the <i>AccurasprayHUB</i> not only accelerates the discovery of stable operating conditions—but also bridges the longstanding divide between thermal spray experts and the machine learning community, paving the way for scalable, data-driven process control.</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

Unsupervised Process Window Identification in Thermal Spray Operations with the AccurasprayHUB

  • 期刊论文

  • 发表日期：

    2026年3月11日

查看全文
  • T. Garcin,
  • A. Sabsabi,
  • D. Georgaris,
  • J. F. Henri,
  • D. Lessard,
  • W. Jibran,
  • J. N. Robert,
  • A. Nadeau

摘要

Thermal spray practitioners have long struggled to define robust plume characteristics process windows—combinations of particle temperature, velocity, and plume geometry that reliably yield acceptable coatings—especially as powder characteristics, equipment wear, and ambient conditions evolve. The AccurasprayHUB addresses this challenge by unifying data capture, structuring, and analysis within a single platform. During an initial lab deployment, the Hub continuously collects summary statistics (mean temperature, velocity, and geometry) for each spray run and automatically enriches them with detailed metadata—stand-off distance, gas and powder flow rates, powder batch IDs, gun maintenance records, and laboratory data when available. These per-run feature vectors are then subjected to density-based clustering to identify the natural “normal” operating regimes without requiring explicit “bad coating” examples. From the resulting high-confidence clusters, the Hub derives multidimensional min–max envelopes that define process windows tailored to specific powders and gun health states. By delivering reproducible data pipelines, transparent feature engineering, and modular analytics, the AccurasprayHUB not only accelerates the discovery of stable operating conditions—but also bridges the longstanding divide between thermal spray experts and the machine learning community, paving the way for scalable, data-driven process control.