Machine learning based assessment of corrosion related damages in reinforced concrete low-rise buildings

Abstract

A machine learning-based approach for evaluating corrosion-related damages in low- rise RC buildings is presented in this study. Traditional assessment methods often rely on visual inspections, which can be subjective and inefficient. To address this limitation, the research integrates empirical data from field investigations and advanced computational techniques to establish a data-driven framework for evaluating structural degradation due to corrosion. Quantitative assessments of key parameters named concrete quality, carbonation depth, compressive strength, concrete cover thickness, and corrosion risk were conducted, using non-destructive tests named ultrasonic pulse velocity test, carbonation test, rebound hammer test, cover meter test, and half-cell potential test during field investigations in RC buildings. A systematic categorization of corrosion-related damages and their causative factors was developed, providing a structured understanding of deterioration mechanisms. Based on empirical data, including the measured key parameters, a Damage Index (DI) equation and a corresponding Damage Assessment Framework were formulated to quantify degradation levels of structural components in RC buildings affected by corrosion. The DI equation and Damage Assessment Framework were validated using machine learning-based regression techniques to enhance predictive accuracy. The findings confirm that the machine learning model developed in this research can effectively identify patterns and predict structural degradation levels, offering valuable insights into corrosion progression. This research underscores the importance of proactive maintenance and data-driven decision-making in structural health monitoring. By integrating machine learning into corrosion assessment, the study advances predictive maintenance strategies, ensuring the long-term durability and resilience of RC structures.