Scholarlink

This paper investigates the damage properties of polylactic acid (PLA) polymers elaborated through fused deposition modeling (FDM) technology. Experimental tensile tests are conducted to investigate the influence of various FDM parameters on the mechanical performance of 3D-printed samples. Specifically, we focus on failure strain (εr) and the ultimate tensile strength (UTS) as key metrics. Our study involves the systematic variation of FDM parameters to assess their impact on quasi-static damage characterization. Additionally, leveraging insights from existing literature, we employ machine learning (ML) regression models to predict mechanical properties, particularly failure strain and the ultimate tensile strength. The findings offer valuable insights for optimizing FDM parameters to enhance the mechanical performance of PLA polymers in additive manufacturing applications. The random forest regression (RFR) and the XGBoost regression (XGB) models yield a lower value of root mean squared Error (RMSE), expressed as a percentage, compared to the linear regression (LR) model. That indicates that the RFR and XGB models are better at accurately predicting the ultimate tensile strength and failure strain of FDM-printed PLA parts.

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

PLA Damage Properties in FDM Printing: A Hybrid Experimental and Machine Learning Study

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Manel Dhouioui,
  • Boutheina Ben Fraj,
  • Hamdi Hentati,
  • Mounir Ben Amar,
  • Mohamed Haddar

摘要

This paper investigates the damage properties of polylactic acid (PLA) polymers elaborated through fused deposition modeling (FDM) technology. Experimental tensile tests are conducted to investigate the influence of various FDM parameters on the mechanical performance of 3D-printed samples. Specifically, we focus on failure strain (εr) and the ultimate tensile strength (UTS) as key metrics. Our study involves the systematic variation of FDM parameters to assess their impact on quasi-static damage characterization. Additionally, leveraging insights from existing literature, we employ machine learning (ML) regression models to predict mechanical properties, particularly failure strain and the ultimate tensile strength. The findings offer valuable insights for optimizing FDM parameters to enhance the mechanical performance of PLA polymers in additive manufacturing applications. The random forest regression (RFR) and the XGBoost regression (XGB) models yield a lower value of root mean squared Error (RMSE), expressed as a percentage, compared to the linear regression (LR) model. That indicates that the RFR and XGB models are better at accurately predicting the ultimate tensile strength and failure strain of FDM-printed PLA parts.