Machine learning-driven optimization of energy absorption in PLA composites reinforced with carbon fibers through FFF additive manufacturing

Purpose This study aims to optimize the energy absorption capacity (EAC) of polylactic acid (PLA) composite materials reinforced with short micro carbon fibers, fabricated via fused filament fabrication (FFF) 3D printing. The objective is to understand how key process parameters influence mechanical performance and to explore the potential of machine learning in enhancing material property prediction and optimization. Design/methodology/approach A series of controlled experiments were conducted by varying nozzle diameter, nozzle temperature, and fan speed. The EAC of the 3D-printed samples was determined through impact testing and density measurements. Three regression-based machine learning models-first-order, second-order, and second-order with interaction terms-were developed and evaluated to predict EAC based on process parameters. Findings The second-order model with interaction terms showed the highest predictive accuracy (R2 = 0.9984, adjusted R2 = 0.9955, predicted R2 = 0.9758). Optimization analysis identified the ideal conditions (0.4 mm nozzle diameter, 200 °C nozzle temperature, 100% fan speed), achieving a maximum EAC of 72.64 kJ.m/g. The study revealed the significant effect of printing parameters on EAC, confirming that fine-tuning these variables enhances composite performance. Originality/value This research highlights the effective integration of machine learning with additive manufacturing to predict and optimize mechanical properties of PLA composites. It provides a valuable framework for improving energy absorption through data-driven tuning of 3D printing parameters, contributing to the advancement of high-performance, customized polymer composites.