Predicting Tensile Strength in FDM-Printed PLA Parts Using Machine Learning

Additive Manufacturing (AM) is transforming the manufacturing industry, with Fused Deposition Modeling (FDM) being one of the most widely used techniques due to its affordability and ease of use. However, the mechanical performance of FDM-printed parts is highly dependent on process parameters, making it essential to understand and optimize their influence. This study investigates the impact of six critical FDM parameters including extruder temperature, printing speed, infill density, infill pattern, layer height, and the number of wall layers, on the tensile strength of 3D-printed Polylactic Acid (PLA) parts. Using the dataset from Qian et al. (2024), which includes tensile strength values of 92 PLA samples fabricated under varied parameter settings, we extend their work by applying several machine learning (ML) models to predict and classify mechanical outcomes. Regression algorithms used include Polynomial Regression (PR), Gradient Boosting (GB), Support Vector Machine (SVM), and Extreme Gradient Boosting (XGB), while classification tasks employed Support Vector Classifier (SVC) and XGBoost (XGB). The XGBoost regressor outperformed other models with a test R² of 0.76 (RMSE:5.19, MAE:3.09), followed by Polynomial Regression (R²:0.71, RMSE:5.60, MAE:3.17), Gradient Boosting (R²:0.70, RMSE:5.78, MAE:2.83), and SVM (R²:0.65, RMSE:6.18, MAE:3.19). In classification, the XGB classifier achieved perfect results on the test set with 100% accuracy, precision, recall, and F1-score, whereas the SVC model attained an accuracy of 84.21% and an F1-score of 0.8491. Pearson correlation coefficient analysis revealed infill density, printing speed and wall layers as the most influential parameters (p