ML Model for Predicting Tensile Strength of 3D-Printed Components Using Bayesian Optimization

The increasing popularity of 3D printing as a manufacturing process has highlighted the need for predictive models, which can estimate the mechanical properties of printed components before production. This study focuses on developing a machine learning (ML) model for predicting the tensile strength of 3D-printed components, manufactured using Fused Deposition Modeling (FDM). The model uses key printing parameters (layer height, infill pattern, infill density, print speed, wall thickness) and provides respective predictions of resulting tensile strength thereby reducing the number of test prints to identify suitable printing parameters. Gaussian Process Regression (GPR) and Bayesian optimization were employed to train the model, utilizing tensile test data from only 57 3D-printed samples. The model performance, evaluated using the coefficient of determination (R²), showed promising results (0.948). This research underscores the potential of ML models to streamline the 3D printing process by optimizing parameter selection and minimizing material waste, thus contributing to the broader adoption of additive manufacturing.