Compressive Strength Prediction of Coal Ash-Reinforced Concrete Using Machine Learning

Coal ash, a waste product from thermal power plants has gained a lot of interest as a key ingredient in concrete to enhance its characteristics while also contributing to eco-friendly construction practice. Random forest (RF), artificial neural networks (ANN), and random tree (RT) are the soft computing techniques used to predict the concrete compressive strength (CS) with coal ash as a reinforcement material. The main aim of the present work was to find the feasibility of using these methodologies to predict the CS of modified concrete containing coal ash as supplementary material. A dataset from the past published literature was assembled to create the model and the same dataset was split into 70/30 training and testing subsets. Three performance evaluation indices namely root mean square error (RMSE), correlation coefficient (CC), and mean absolute error (MAE) were used to find the optimum model. The analysis results revealed that the ANN model was the best among the applied soft computing techniques with a CC value of 0.9694 and 0.9614, RMSE values of 2.6561 and 3.1867, and MAE values of 1.8723 and 2.4249 for both training and testing stages. Furthermore, a sensitivity study was performed by deleting one input parameter at a time and the result showed that cement content has a considerable impact on the CS prediction of coal ash concrete. All things considered, the present research shows how effectively these computing techniques can be used to predict the modified concrete CS when incorporating coal ash as a supplementary ingredient.