Answers are grounded in this paper and cite their sources.
Corrosion of steel reinforcement, particularly in marine and aggressive environments, significantly impairs the seismic performance of reinforced concrete beam–column joints by accelerating deterioration and altering the failure mechanisms. This study explores the application of machine-learning techniques for predicting the load-bearing capacity of corroded, seismically detailed reinforced concrete (RC) beam–column joints under cyclic loading. An extensive experimental dataset comprising both uncorroded and corroded values was compiled, incorporating key structural parameters such as reinforcement indices, sectional loss ratios, and axial load ratios. Ten predictive models, including neural networks, instance-based models, and ensemble models, were trained and evaluated for predicting the load-bearing capacity using the correlation coefficient ( r ), coefficient of determination ( R 2 ), and the root mean squared error. The results exhibit that the stacking regressor model achieved the highest predictive accuracy ( R 2 = 0.96 ), outperforming all individual algorithms in modeling the complex nonlinear effects of corrosion on seismic capacity. The findings highlight the reliability of ensemble-based approaches, particularly stacking, in capturing complex nonlinear interactions. The proposed ML framework provides a reliable, rapid, and noninvasive tool to support assessment of seismic vulnerability and retrofit prioritization in aging RC infrastructure. This approach obviates the requirement for resource-intensive simulations and proposes a significant potential for practical implementation in construction decision-making and health monitoring.
No takes yet. Share an insight, caveat, or question.
Dangwal et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: