Artificial Intelligence and Machine Learning Techniques to Predict the Compressive Strength of Concrete at High Temperature

Abstract

The nature of the components used to make concrete is significantly impacted by high temperatures, which in turn lessens the concrete's strength qualities. Increasing the concrete's compressive strength to the optimum level takes effort and time. However, the use of supervised machine learning (ML) techniques enables the first, very accurate prediction of the desired result. This study uses 207 data points to anticipate the compressive strength of concrete at high temperatures using a decision tree (DT), an artificial neural network (ANN), bagging, and gradient boosting (GB). The chosen models were run using Anaconda navigator programme and Python code. Both information about the input variables and the output parameter are needed by the programme. One output parameter (compressive strength) was chosen out of a total of nine input parameters (water, cement, coarse aggregate, fine aggregate, fly ash, superplasticizers, silica fume, nano silica, and temperature). Statistics such as the coefficient correlation (R2), mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE) were used to assess the effectiveness of the deployed ML algorithms. R² for individual models using DT and ANN was 0.83 and 0.82, respectively, but R² for models using the ensemble approach and gradient boosting was 0.90 and 0.88. This shows a significant link between the actual and expected results. Ensemble methods performed worse than the k-fold cross-validation, coefficient correlation (R²), and fewer errors (MAE, MSE, and RMSE). Sensitivity studies were also performed to see how each input variable contributed to the results. It has been established that employing the ensemble machine learning method would raise the model's level of performance.