Scholarlink

<p>This research explores the application of supervised machine learning techniques, integrated with simulation model, to predict the clad layer geometry in laser direct energy deposition (L-DED), which is one of the most well-known processes in laser-based additive manufacturing. The study focuses on the deposition of Inconel 718 powder on stainless steel 304 substrates and Goldak's heat source model has been used to simulate melt pool temperature and clad characteristics under varying laser power, scan speed, and powder feed rate. Gaussian process regression (GPR) and support vector regression (SVR) models were trained on simulation data to predict clad layer dimensions and dilution ratios. The GPR model, particularly with a squared exponential kernel, demonstrated superior predictive accuracy over SVR. It presents an optimization contribution for L-DED process, giving a better framework for improvement of material fabrication technologies.</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

Simulation-Based Machine Learning for Predicting Clad Layer Geometry in Laser Direct Energy Deposition

  • 期刊论文

  • 发表日期：

    2025年2月26日

查看全文
  • Ebrahim Gholami,
  • Saeed Batebi

摘要

This research explores the application of supervised machine learning techniques, integrated with simulation model, to predict the clad layer geometry in laser direct energy deposition (L-DED), which is one of the most well-known processes in laser-based additive manufacturing. The study focuses on the deposition of Inconel 718 powder on stainless steel 304 substrates and Goldak's heat source model has been used to simulate melt pool temperature and clad characteristics under varying laser power, scan speed, and powder feed rate. Gaussian process regression (GPR) and support vector regression (SVR) models were trained on simulation data to predict clad layer dimensions and dilution ratios. The GPR model, particularly with a squared exponential kernel, demonstrated superior predictive accuracy over SVR. It presents an optimization contribution for L-DED process, giving a better framework for improvement of material fabrication technologies.