Page 120 - MSAM-4-3

Materials Science in Additive Manufacturing Sustainable manufacturing composite material optimization

Figure 7. Root mean squared error (left) and mean absolute error (right) values over training epochs for both
training and validation sets

Table 4. Comparative analysis of AI‑MCDM and of surface roughness, thermal distortion compensation,
conventional parameter selection and real-time defect detection, to further automate FDM-
based impeller manufacturing. Other hybrid AI methods,
Parameter selection Tensile Energy Wear rate
method strength consumption (mm /N·m) including RL and digital twin models, hold promise
3
(MPa) (Wh) for predicting machine-specific variations in FDM
Conventional approach 38.7 140 0.62 performance. Such advancements would contribute to
AI-MCDM-optimized 41.5 145 0.55 the development of next-generation smart manufacturing
platforms, where AI systems continually learn and evolve
Abbreviation: AI-MCDM: Artificial Intelligence–multi-criteria FDM processes in real-time to optimize efficiency, reduce
decision-making.
waste, and improve product quality.
enabling real-time adaptive control in the manufacturing 4. Conclusion
process.
The findings of this study validated the performance of
Figure 7 demonstrates stable convergence behavior AI-based optimization and MCDM methods in optimizing
of the ANN model, with decreasing error trends and the mechanical characteristics and sustainability of
no indication of overfitting, thereby confirming the FDM-based impeller fabrication. The combination of Fuzzy
reliability and generalization capability of the predictive AHP and TOPSIS with AI predictive modeling enabled the
model. These results demonstrate the potential of identification of optimal trade-offs among tensile, flexural,
AI-MCDM-based optimization in FDM impeller and compressive strengths, wear resistance, and energy
manufacturing. The AI-MCDM-based approach enables efficiency. Higher infill densities and shell thicknesses led
data-driven enhancement of material efficiency, mechanical to notable improvements in mechanical properties and
performance, and energy consumption. The integration wear resistance; however, these gains were accompanied
of real-time sensors in the AI system enables dynamic by increased energy consumption, highlighting the
adjustment of print parameters, ensuring high process importance of AI-driven, eco-friendly parameter
reliability and sustainability.
optimization. The AI-optimized process yielded 7% higher
Collectively, the results of this study suggest that tensile strength and 11% lower wear rate, while maintaining
future research should focus on extending AI-MCDM energy consumption within 3.5% of baseline FDM
frameworks to additional objectives, such as optimization parameters. Overall, the results highlighted the superiority

Volume 4 Issue 3 (2025) 12 doi: 10.36922/MSAM025200033