Optimization of the manufacturing process of materials produced from ABS and PLA filaments using artificial intelligence

This study aims to optimize the mechanical performance and surface quality of parts produced using ABS and PLA filaments via the Fused Filament Fabrication (FFF) method, by employing artificial intelligence-based techniques. Within the scope of the thesis, key manufacturing parameters such as nozzle temperature, bed temperature, layer height, print speed, and infill density were systematically varied using a multi-level experimental design. A total of 324 specimens were fabricated for both materials. Experimental data were evaluated in terms of performance metrics such as tensile strength and surface roughness, and artificial neural network (ANN)-based predictive models were developed. Additionally, multi-objective optimization was conducted using NSGA-II and Particle Swarm Optimization (PSO) algorithms to enhance both mechanical and surface properties. The findings indicated that PLA exhibited superior performance, with an average tensile strength of 48.2 MPa and surface roughness of 6.1 µm, compared to ABS. The ANN model demonstrated high predictive accuracy, enabling the digital estimation of process outcomes. Optimization results showed reduced production time and improved material efficiency for both materials. Overall, the results highlight that AI-assisted approaches offer a robust framework for improving quality and supporting sustainability in FFF-based additive manufacturing processes.