Experimental Validation of an IoT-Integrated Maturity and Machine Learning Method for Concrete Strength Prediction

The Internet of Things (IoT) has the power to transform concrete testing in real time by adopting non-destructive testing of concrete cubes by measuring simple phenomena like the internal temperature of concrete. This study validates an IoT-integrated maturity method for M25 concrete under Indian conditions using embedded DS18B20 temperature sensors and the Nurse–Saul formulation in ASTM C1074 as the reference framework. Twenty-four cubes of 150mm size are cast, of which 8 were equipped with DS18B20 temperature sensors. Readings are collected at an interval of 1 minute, and later converted in hours. The temperature of the cube is observed continuously for 28 days. The Nurse-Saul equation is adopted for verification. The sensitivity of Datum temperature is verified by using various temperatures like − 10°C, − 5°C, 0°C, 2.5°C, 5°C, and 10°C. The results of maturity-based and actual compressive strength are compared for predefined datum temperatures, in which 0°C comes with the lowest error. To strengthen these results, three machine learning regression models are adopted: Random Forest (RF), artificial neural network (ANN), and Support Vector Machine, using time, temperature, and datum temperature as input parameters to predict the compressive strength of concrete. From which RF model reserves first place by achieving R² = 0.972, RMSE = 2.86 MPa, MAPE = 6.12%, MAE = 1.97 MPa followed by ANN and SVM models. The results combines IoT based maturity monitoring with datum temperature and calibrated ML models for accurate early predictions suitable for safer removal of formwork for improved construction practice.