Integration of Artificial Neural Networks with Rebound Hammer Testing for Enhanced Concrete Retrofitting Assessment

The structural assessment and retrofitting of aging concrete infrastructure are critical for public safety and sustainability. Non-destructive testing (NDT) methods, such as the rebound hammer test, provide surface hardness estimates correlated to concrete compressive strength, aiding in preliminary evaluation stages. However, rebound hammer results are influenced substantially by factors including surface condition, moisture, carbonation, and aggregate properties, complicating direct strength estimation. This paper explores the integration of Artificial Neural Networks (ANNs), a powerful machine learning tool, to accurately predict concrete compressive strength from rebound hammer data while accounting for influencing environmental and material variables. Extensive experimentation, including data acquisition from various aged structures and destructive core testing for validation, is coupled with ANN development and training to improve predictive accuracy beyond traditional empirical correlations. The results demonstrate ANN's superior ability to manage non-linear relationships and data variability inherent in rebound hammer outputs, facilitating enhanced and reliable retrofitting decisions. Moreover, this study underscores how this approach promotes cost-efficient, sustainable structural rehabilitation through precise strength mapping, prioritization of repairs, and minimization of material waste.