MECHANICAL BEHAVIOR PREDICTION OF CARBON FIBER-REINFORCED ONYX IN FDM USING INTEGRATED STATISTICAL AND MACHINE LEARNING APPROACHES

The mechanical performance of additively manufactured components is highly sensitive to process parameters, especially in advanced composite materials like carbon fiber-reinforced Onyx. This study presents a comparative optimization framework combining Response Surface Methodology (RSM) and machine learning (ML) to model and enhance the tensile and flexural strengths of Fused Deposition Modeling (FDM) printed Onyx composites. Key parameters including infill pattern, infill density, and nozzle temperature—were systematically varied using a Taguchi L9 design, and mechanical testing was performed according to ASTM standards. Statistical analysis revealed infill pattern as the most significant factor affecting strength properties. RSM provided reliable predictions with R² values of 97.61% (tensile) and 95.93% (flexural), while ML models, particularly XGBoost coupled with Bayesian optimization, achieved superior prediction accuracy with zero average error. Both methods converged on the same optimal parameters hexagonal infill, 60% infill density, and 265 °C nozzle temperature highlighting the consistency and robustness of the integrated approach. The results demonstrate that combining traditional statistical methods with advanced machine learning offers a powerful pathway for precise process control and mechanical optimization in polymer composite additive manufacturing.