Scholarlink

<p>Interlayer bonding remains one of the main weaknesses of parts produced by Fused Deposition Modeling (FDM), especially in the build (Z) direction where cracks often initiate. This is caused by limited chain diffusion and thermal consolidation during deposition. Improving and predicting interlayer bond strength is therefore essential for structurally reliable applications. In this work, the influence of key FDM parameters (printing temperature, printing speed, layer height, and extrusion multiplier) on the tensile strength of PETG specimens printed along the Z-axis was investigated. Specimens were printed in triplicate to assess both mechanical performance and variability. Machine-learning models, including Linear Regression (LR), Random Forest Regression (RFR), Gradient Boosting Regression (GBR), eXtreme Gradient Boosting Regression (XGBR), AdaBoost Regression (ABR), and an Artificial Neural Network (ANN), were implemented using the scikit-learn Python library to predict average interlayer bond strength from process parameters. The results show that nonlinear ensemble models outperform linear regression, confirming their ability to capture the complex interactions governing interlayer bonding. Among the tested methods, AdaBoost Regression provided the best performance, with R<sup>2</sup> ≈ 0.70, RMSE ≈ 4.42&#xa0;MPa, and MAE ≈ 3.55&#xa0;MPa on the test set. Feature importance analysis identified printing temperature as the dominant parameter, with additional coupled effects involving speed, extrusion multiplier, and layer height. Two-parameter response surfaces and complementary microstructural observations further supported the mechanical trends and provided a physically grounded interpretation of the process–property relationships. Overall, this study offers practical guidance for optimizing FDM conditions to improve the strength of PETG printed parts.</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

Predicting interlayer bond strength in FDM-printed PETG parts: an experimental and machine-learning study

  • 期刊论文

  • 发表日期：

    2026年8月3日

查看全文
  • André Chateau Akué Asséko,
  • Adélaïde Leroy,
  • Benoît Cosson

摘要

Interlayer bonding remains one of the main weaknesses of parts produced by Fused Deposition Modeling (FDM), especially in the build (Z) direction where cracks often initiate. This is caused by limited chain diffusion and thermal consolidation during deposition. Improving and predicting interlayer bond strength is therefore essential for structurally reliable applications. In this work, the influence of key FDM parameters (printing temperature, printing speed, layer height, and extrusion multiplier) on the tensile strength of PETG specimens printed along the Z-axis was investigated. Specimens were printed in triplicate to assess both mechanical performance and variability. Machine-learning models, including Linear Regression (LR), Random Forest Regression (RFR), Gradient Boosting Regression (GBR), eXtreme Gradient Boosting Regression (XGBR), AdaBoost Regression (ABR), and an Artificial Neural Network (ANN), were implemented using the scikit-learn Python library to predict average interlayer bond strength from process parameters. The results show that nonlinear ensemble models outperform linear regression, confirming their ability to capture the complex interactions governing interlayer bonding. Among the tested methods, AdaBoost Regression provided the best performance, with R2 ≈ 0.70, RMSE ≈ 4.42 MPa, and MAE ≈ 3.55 MPa on the test set. Feature importance analysis identified printing temperature as the dominant parameter, with additional coupled effects involving speed, extrusion multiplier, and layer height. Two-parameter response surfaces and complementary microstructural observations further supported the mechanical trends and provided a physically grounded interpretation of the process–property relationships. Overall, this study offers practical guidance for optimizing FDM conditions to improve the strength of PETG printed parts.