Scholarlink

<p>This paper presents a novel in-situ optimization approach for Fused Deposition Modeling (FDM) Additive Manufacturing which utilizes high-resolution data from a 3D scanner and Artificial Intelligence for minimizing layer-specific surface roughness. The procedure starts with the print head retracting post-layer deposition, which is followed by a 3D scanner creating a detailed point cloud of the surface. These data are then pre-processed to calculate the surface roughness, and a metaheuristic optimization algorithm, supported by a neural network, updates the print settings for the subsequent layer. The analysis of the different algorithms revealed that Particle Swarm Optimization was the most efficient, with a better fitness evaluation and faster convergence than the Genetic Algorithm. The study focuses on the effect of sequential layer deposition on surface roughness and shows that the quality of the previous layer has a significant impact on the quality of the next. In addition, a Digital Twin architecture is implemented that enables seamless integration between the FDM printer and the 3D scanner and provides a user-friendly interface for real-time parameter monitoring and surface visualization. This approach not only improves manufacturing precision, but also sets new standards for accuracy and reliability in Additive Manufacturing. It demonstrates the transformative potential of adaptive manufacturing techniques that integrate real-time data acquisition, advanced AI and robust optimization algorithms.</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

Digital twin-driven real-time optimization of layer-specific surface roughness in FDM 3D printing

  • 期刊论文

  • 发表日期：

    2025年7月7日

查看全文
  • Abdelhamid Ziadia,
  • Mohamed Habibi,
  • Sousso Kelouwani

摘要

This paper presents a novel in-situ optimization approach for Fused Deposition Modeling (FDM) Additive Manufacturing which utilizes high-resolution data from a 3D scanner and Artificial Intelligence for minimizing layer-specific surface roughness. The procedure starts with the print head retracting post-layer deposition, which is followed by a 3D scanner creating a detailed point cloud of the surface. These data are then pre-processed to calculate the surface roughness, and a metaheuristic optimization algorithm, supported by a neural network, updates the print settings for the subsequent layer. The analysis of the different algorithms revealed that Particle Swarm Optimization was the most efficient, with a better fitness evaluation and faster convergence than the Genetic Algorithm. The study focuses on the effect of sequential layer deposition on surface roughness and shows that the quality of the previous layer has a significant impact on the quality of the next. In addition, a Digital Twin architecture is implemented that enables seamless integration between the FDM printer and the 3D scanner and provides a user-friendly interface for real-time parameter monitoring and surface visualization. This approach not only improves manufacturing precision, but also sets new standards for accuracy and reliability in Additive Manufacturing. It demonstrates the transformative potential of adaptive manufacturing techniques that integrate real-time data acquisition, advanced AI and robust optimization algorithms.