Advancing 3D-printed fiber-reinforced concrete for sustainable construction: A comparative optimization based study of hybrid machine intelligence models for predicting mechanical strength and CO₂ emissions

With the increasing demand for sustainable and resilient marine infrastructure in marine environments, 3D printing technologies offer a promising solution for fabricating customized, durable components under challenging conditions. 3D-printed fiber-reinforced concrete (3DPFRC) presents significant potential for rapid, waste-reducing construction of complex geometries, making it suitable for marine structures such as sea walls, breakwaters, underwater pipelines, and floating platforms. However, optimizing 3DPFRC for mechanical performance and environmental sustainability remains a complex challenge. This study proposes advanced machine learning (ML) models to simultaneously predict compressive strength and CO₂ emissions of 3DPFRC, enabling both mechanical and environmental performance evaluation. A user-friendly graphical user interface (GUI) was also developed to facilitate practical deployment by engineers without programming expertise. Four hybrid ML models were evaluated: CNN-LSTM, RA-PSO, XGBoost-PSO, and SVM-PSO. RA-PSO outperformed others with an R² of 0.9819 (training) and 0.9674 (testing) for compressive strength and 0.97 (training) and 0.94 (testing) for CO₂ emissions, alongside the lowest MSE (48.24 MPa²) and highest F1-score (0.9519). This superior performance is primarily due to RA-PSO’s adaptive parameter tuning and randomized search, which maintain population diversity and prevent premature convergence, enabling the model to capture complex nonlinear interactions in 3DPFRC mix parameters. Sensitivity analysis revealed that water content (34 %), silica fume (30 %), and the water-to-binder ratio (23 %) were the most influential parameters on compressive strength. These findings confirm RA-PSO as a highly reliable tool for optimizing 3DPFRC mix designs while minimizing environmental impact, particularly in Sustainable Marine and Civil Infrastructure Applications.