Hybrid RSM-ANFIS modeling for concurrent optimization of fused filament fabrication and graphene content in high-performance PLA nanocomposites

Abstract

Fused Filament Fabrication (FFF) offers unparalleled geometric complexity but faces significant challenges in achieving optimal mechanical performance and part accuracy in advanced nanocomposites. This study introduces a comprehensive approach to concurrently optimize both the material composition (Graphene concentration) and the process parameters (layer height, infill pattern, and extrusion temperature) to fabricate high-performance Polylactic Acid/Graphene (PLA/GR) nanocomposites. PLA/GR filaments were prepared by mixing in a high-energy planetary ball mill, followed by melt extrusion. A Central Composite Design (CCD) within Response Surface Methodology (RSM) established the experimental matrix, systematically linking input variables to key outputs: tensile strength, thermal characteristics, and dimensional accuracy. Initial experimental results demonstrated a maximum tensile strength of 53 MPa at 4% GR, 0.25 mm layer height, 220°C extrusion temperature, and Gyroid infill. Thermal analysis confirmed the material’s thermal stability with only a minor shift in the melting temperature (T m ∼ 157°C), while dimensional analysis showed that all fabricated parts met strict tolerance requirements (C pk > 1.33). To achieve global optimal parameters, a hybrid predictive model combining RSM with a Genetic Algorithm-Adaptive Neuro-Fuzzy Inference System (GA-ANFIS) was developed. The model accurately predicted a maximum tensile strength of 60.56 MPa at the optimal set (4.5% GR, 0.22 mm layer height, 225°C, and Gyroid infill). Experimental validation confirmed this prediction with a maximum tensile strength of 59 MPa, yielding a minimal prediction error of 2.576%. This work establishes an integrated modeling framework for concurrent material and process optimization, enabling rapid, reliable production of high-strength FFF nanocomponents.