Experimental validation of an IoT-integrated maturity and machine learning method for concrete strength prediction

Monitoring the strength of concrete during the early curing phase is essential for ensuring safe removal of formwork and effective scheduling of construction activities. Traditional methods of measuring compressive strength involve destructive testing of cubes, which yields only limited data points and can postpone decision-making in construction. This research provides experimental validation for an Internet of Things (IoT)–based maturity monitoring system that forecasts the compressive strength of M25 concrete in the context of Indian weather conditions. Waterproof DS18B20 temperature sensors placed within concrete cubes consistently recorded the internal temperature at one-minute intervals over a span of 28 days. The recorded temperature data were utilized to calculate the maturity index using the Nurse-Saul equation as outlined in ASTM C1074. The impact of various datum temperatures (− 10 °C, − 5 °C, 0 °C, 2.5 °C, 5 °C, and 10 °C) on strength prediction accuracy was assessed using a sensitivity analysis under laboratory condition within study data. Compressive strengths evaluated experimentally at 3, 7, 14, and 28 days were used to validate the projected strengths derived from the maturity technique. The findings show that, within the allowable bounds suggested by ASTM C1074, a datum temperature of 0 °C offers the lowest prediction error and exhibits good agreement with experimental data. Using temperature history, curing time, and datum temperature as input variables, machine learning models such as Artificial Neural Networks, Random Forest, and Support Vector Machine were created to further examine nonlinear relationships between curing parameters and compressive strength. Several data split ratios (70-15-15, 80-10-10, and 90-5-5) and statistical measures, including R2, RMSE, MAE, and MAPE were used to assess the model’s performance.