Fuente: Journal of applied polymer Lugar: RESEARCH ARTICLE Methodology followed in this study.


ABSTRACT
Polylactic acid (PLA) is widely used in fused filament fabrication (FFF) due to its biocompatibility and low cost, but its inherent brittleness and limited mechanical strength restrict its application in load-bearing and structural components. Reinforcing PLA with nanofillers such as carbon fiber (CF), graphene (Gr), and multi-walled carbon nanotubes (MWCNTs) offers a promising route to overcome these limitations, yet the combined influence of build parameters and reinforcement type remains underexplored. In this study, PLA composites were fabricated via FFF with systematic variation of raster orientation (RO), layer height (LH), and print speed (PS), and their tensile, compressive, and flexural properties were evaluated. Among the reinforcements, PLA-MWCNT composites demonstrated the highest compressive strength (121.25 MPa at 0° RO, 0.1 mm LH, 30 mm/s PS), while PLA-CF composites exhibited superior tensile performance, and PLA-Gr composites offered balanced mechanical behavior. To predict and optimize these outcomes, machine learning (ML) models—Linear Regression, Random Forest Regression, and Extreme Gradient Boosting Regression—were employed, with XGBR achieving the highest accuracy (R
2 up to 0.99). This integrated experimental–computational framework identifies reinforcement–parameter combinations most suitable for real-world applications such as lightweight automotive components while also highlighting the need for larger datasets to further strengthen ML-driven predictions.