Machine learning driven analysis of dimensional accuracy in FFF printed PLA-cenosphere-graphene composites

This study investigates the influence of cenosphere and graphene nanoplatelet (GNP) reinforcements on the physical, mechanical, and dimensional accuracy characteristics of PLA-based composites fabricated using Fused Filament Fabrication (FFF). The incorporation of cenospheres effectively reduced the composite density from 1.25 g/cm³ for neat PLA to 1.19 g/cm³ and 1.14 g/cm³ for 5 wt% and 10 wt% cenosphere additions, respectively, demonstrating the lightweighting capability of hollow ceramic fillers. The void content increased progressively with reinforcement loading due to increased melt viscosity, particle agglomeration, and localized air entrapment during processing, with values ranging from 2.14% to 8.09% for the reinforced composites. Despite the increase in porosity, graphene nanoplatelets improved the overall stiffness and stress transfer efficiency of the composites through enhanced interfacial interaction and restriction of polymer chain mobility. Mechanical characterization revealed substantial enhancement in hardness, reaching a maximum value of 91.5 Shore D for the hybrid PCG-5 composite, attributed to the synergistic reinforcement effect of rigid cenospheres and high-stiffness graphene nanoplatelets. Dimensional accuracy analysis identified layer thickness as the most influential processing parameter, contributing nearly 79% of the total variation, with lower layer heights consistently improving geometric fidelity. Statistical analyses using Taguchi design, ANOVA, contour mapping, and regression modeling confirmed this trend, while Random Forest regression further validated the experimental findings with a high prediction accuracy (R² = 0.94) and minimal prediction error. Feature importance analysis also established layer thickness as the dominant governing factor, followed by printing speed and raster angle. Overall, the integration of experimental optimization and machine learning provides an effective framework for improving dimensional precision and mechanical performance in additively manufactured PLA hybrid composites intended for lightweight engineering applications.