Long-Term Reinforced Concrete Damage Detection via IoT and Machine Learning Models

Abstract

This study presents the development and implementation of an integrated system that combines Internet of Things (IoT) sensors and the XGBoost algorithm to predict damage in reinforced concrete structures under tropical climate conditions. The system was deployed on prototype structures equipped with temperature, humidity, strain, and vibration sensors, and data were collected over 60 days. Using supervised machine learning, the XGBoost model achieved high accuracy (93.2%) in classifying structural conditions into the categories of safe, vulnerable, and damaged. Real-time monitoring enabled early anomaly detection, with an average system response time of 3.4 seconds, significantly outperforming manual inspections. The findings demonstrate that this approach provides a viable predictive maintenance solution for infrastructure in high-humidity, variable-temperature environments. While the system performs effectively in a prototype setting, further testing on full-scale structures is needed to validate its robustness and expand its detection scope. This research advances intelligent infrastructure systems through data-driven decision-making and AI-based damage forecasting.