Machine learning and RSM-based optimization of mechanical properties in MXene-reinforced PLA composites for additive manufacturing applications

The mechanical properties of fused deposition modeling (FDM)-printed polymer composites depend on both material composition and printing parameters, necessitating systematic optimization and reliable predictive modeling. This study investigates the effects of material composition and process parameters on the mechanical properties of FDM-processed, MXene-reinforced polylactic acid (PLA) composites using response surface methodology (RSM) and machine learning (ML). Thirty-one experimental runs were conducted by varying MXene content (0–0.05 wt.%), outer shells (2–4), orientation angle (0°–90°), nozzle temperature (190 °C–210 °C), and printing speed (20–60 mm/s). Tensile strength, Young’s modulus, and Vickers hardness ranged from 1.11–27.00 MPa, 326–2,969 MPa, and 44.6–84.3 HV, respectively. Maximum tensile strength was obtained at 0.05 wt.% MXene and 0° orientation, whereas Young’s modulus was highest at 0.025 wt.% MXene with four outer shells. The highest hardness of 84.3 HV was obtained at 0.025 wt.% MXene and 45° orientations, with three outer shells, a nozzle temperature of 200 °C, and a printing speed of 60 mm/s. ANOVA indicated that the developed regression models reported R2 values above 0.90 and insignificant lack-of-fit tests (p > 0.05). MXene content affected tensile strength and hardness, while outer shells and nozzle temperature significantly influenced stiffness. Among the evaluated ML models, LightGBM showed the highest model-fit R2 values of 0.964, 0.965, and 0.935 for tensile strength, Young’s modulus, and hardness, respectively. Evaluation using five additional experimentally tested conditions showed reasonable agreement between predicted and measured responses, with R2 values of 0.846, 0.881, and 0.856 for tensile strength, Young’s modulus, and hardness, respectively. SHAP and Pearson analyses further identified the dominant process parameters. These results demonstrate the potential of combining RSM and ML for data-driven analysis and optimization of FDM-processed PLA/MXene composites for potential applications in automotive components, sports equipment, and customized functional parts.