Prediction of compressive strength of roller compacted concrete using regression analysis and artificial neural networks

The compressive strength is most reliable parameter to evaluate the ability of concrete in resisting compression. The paper presents a study on prediction of the compressive strength of roller compacted concrete using multiple regression analysis (MRA) and artificial neural networks (ANN). The compressive strength of roller compacted concrete was obtained experimentally at 3, 7 and 28 days of curing. The samples were prepared by varying the percentage of cement and superplasticizer. The data were organized in three different groups randomly using R statistical software. The models were executed with cement content, coarse and fine aggregate, superplasticizer content, water content and days of aging as input parameters that were used to predict compressive strength which is the output parameter. The analysis was performed using multiple regression and artificial neural networks methodology. Statistical measures like root-mean-square error (RMSE), mean absolute error (MAE) and coefficient of determination are used to assess the performance of models. The determination coefficient from multiple regression analysis is found to be 0.975 and 0.886 for testing and validating the data correspondingly, whereas the determination coefficient from artificial neural network analysis is found to be 0.9 for both testing and validating the data. The results obtained from ANN are highly accurate because of its own topology.