Predicting the Experimental Results of the Tensile Strength of Parts Produced with a 3-D Printer Using ABS Filament with Artificial Intelligence

ABSTRACT This study used artificial intelligence (AI) to predict the tensile strength of materials produced through fused deposition modeling (FDM) type three-dimensional (3-D) printing. Samples were produced using a 3-D printer with printing speeds of 40, 70, and 100 mm/s, nozzle temperatures of 190°C and 220°C, raster angles of 0°, 45°, and 90°, layer thicknesses of 0.1 and 0.2 mm, and a constant infill ratio of 100 %. Tensile strength data were obtained through tensile testing. In the conducted experiments, the highest tensile strength was recorded at parameters of 40 mm/s printing speed, 220°C nozzle temperature, 90° writing angle, and 0.1 mm layer thickness, yielding 43.27 MPa, whereas the lowest tensile strength was observed at parameters of 100 mm/s printing speed, 190°C nozzle temperature, 0° writing angle, and 0.2 mm layer thickness, resulting in 7.33 MPa. Tensile strength values obtained from standardized mechanical tests were used to train and evaluate several machine-learning algorithms, including random forest, decision tree, ridge, lasso, and elastic net regressions. Among these, the random forest algorithm yielded the highest prediction accuracy with an R2 value of 0.948. The results demonstrate that AI can effectively model the complex relationships between 3-D printing parameters and mechanical performance, offering a powerful tool for optimization in additive manufacturing processes.