A hybrid genetic algorithm-neural network model for optimizing thermal and mechanical characteristics of 3D-printed parts

Purpose Fused filament fabrication (FFF) optimization remains a critical challenge due to the complex interplay of process parameters affecting the physicochemical and mechanical properties of printed parts. This study aims to optimize the FFF process by analyzing the effects of liquefier temperature, platform temperature and print speed using artificial neural networks (ANNs) and genetic algorithms (GAs). Design/methodology/approach A hybrid approach combining ANNs and GAs was employed to model and optimize the process parameters. The GA systematically explored the parameter space to identify optimal settings that maximize material properties such as degree of crystallinity, tensile strength, Young’s modulus and elongation at break. The ANN, trained on an extensive experimental dataset, provided predictive insights into these properties, enabling a more efficient and precise optimization. A heat transfer model was utilized to predict the temperature profile of the filaments during deposition, validating the optimized conditions and offering insights into the thermal history of the printed parts. Findings The results demonstrate that the optimized process parameters significantly enhance material properties and improve inter-layer bonding. The study identifies the critical influence of liquefier temperature and print speed on mechanical performance, while also revealing limitations in traditional optimization methods. Furthermore, the thermal model predictions align well with experimental findings, reinforcing the validity of the proposed optimization strategy. Originality/value This study provides a novel hybrid approach to optimizing FFF process parameters by integrating machine learning techniques with traditional experimental and analytical models. The findings offer valuable insights into improving 3D printing efficiency and part quality, with potential applications in various industrial and research domains. Future work may explore the generalizability of the findings to different materials and incorporate real-time process monitoring for enhanced adaptability and precision.