Evaluation of Tensile Properties of 3D-Printed PA12 Composites with Short Carbon Fiber Reinforcement: Experimental and Ma-Chine Learning-Based Predictive Modelling

The present study investigates the tensile properties of 3D-printed PA12 composites reinforced with short carbon fibers, focusing on the impact of printing parameters on material performance. We employed both experimental testing and machine learning-based predictive modeling to evaluate the influence of layer thickness, extrusion width, and raster angles on failure stress, failure strain, and stress–strain curves. Four machine learning models, including Gaussian process regression (GPR), gradient boosting regression (GBR), random forest (RF), and artificial neural network (ANN), were developed and trained on the experimental data. The results indicated that ANN and GPR models outperformed RF and GBR in predicting mechanical properties, with ANN demonstrating the highest accuracy across all tasks. A SHAP analysis was conducted to interpret the models, revealing that raster angles significantly influence failure stress predictions, while extrusion width predominantly affects failure strain predictions. The ability of the models to predict entire stress–strain curves provides a comprehensive understanding of the material’s mechanical behavior, which is crucial for applications requiring detailed material response data. This study highlights the potential of machine learning models, particularly ANN, in predicting the tensile properties of 3D-printed composites. The findings offer valuable insights for optimizing the 3D printing process to achieve desired material characteristics and pave the way for further research in integrating these predictive tools into additive manufacturing workflows for real-time optimization and quality control.