A Machine Learning Approach to Compressive Strength Prediction of 3D-Printed Biocha-Reinforced PLA Biocomposite

The study pioneers the employment of biochar-reinforced polylactic acid biocomposites in 3D printing to enhance sustainability in additive manufacturing. This research examines the compressive strength of 3D-printed biochar-reinforced polylactic acid biocomposite with fused deposition modeling (FDM). Also, various machine learning algorithms were employed for building prediction models. The selected process parameters under investigation were the %wt. content of biochar (%BC), layer thickness (LT), infill density (ID), and printing patterns (PP). The mechanical compression test was done in accordance with ASTM D695-15, and Taguchi L16 was used for designing the experiments. Various machine learning algorithms, namely linear regression (LR), support vector machines (SVM), extreme gradient boosting (XGB), random forest (RF), and artificial neural networks (ANN), were employed for building prediction models. The result of the mechanical test shows that for 1% biochar content at 100% infill density, the maximum compressive strength and compressive strength to weight ratio were 79 MPa and 16494.21 MPa/kg, respectively. The most accurate prediction of compressive strength for the proposed 3D-printed biocomposite was done by the XGB model with a 98% R-squared value, 1.91 MAE, and 2.71 RMSE for the test dataset.