Akademik Veri Yönetim Sistemi
Nondestructive Testing and Evaluation, 2025 (SCI-Expanded, Scopus)
With advances in technology, non-destructive testing (NDT) combined with artificial intelligence (AI) offers a highly effective approach to determining concrete strength. Accurately estimating the strength of reinforced concrete without causing damage is essential for assessing both existing and under-construction structures. However, many studies in the literature rely solely on P-wave velocity or single concrete designs, often overlooking the effects of reinforcement and different curing conditions. This study presents an innovative AI-based approach to estimating the strength of both reinforced and unreinforced concrete under varying curing conditions. A total of 334 concrete samples were subjected to various curing treatments, and their strength was assessed using uniaxial compressive strength tests and ultrasonic methods. Data obtained from NDT techniques were analysed using machine learning (ML) algorithms, including Random Forest (RF), XGBoost (XGB), and Artificial Neural Networks (ANN). Key parameters such as p- and S-wave velocities, curing time, curing type, water/cement ratio, and reinforcement diameter were incorporated into the models. Among the tested AI methods, XGB demonstrated the highest accuracy in predicting concrete strength. XGB reduced RMSE by 12% compared to RF. These findings highlight the potential of machine learning in improving the efficiency and reliability of concrete strength estimation.