Optimizing compressive strength prediction in eco-friendly recycled concrete via artificial intelligence models

The use of recycled coarse aggregates (RCA) in concrete is currently recognized for its positive impact on reducing environmental hazards and promoting sustainable development, a trend observed worldwide. However, despite these benefits, RCA’s disadvantages on concrete properties have led to limited attention and usage. Developing an optimal RCA mixture design in the laboratory is time-consuming and can cause construction delays. In this study, various artificial intelligence models were employed, including the group method of data handling neural network (GMDH-NN), combinatorial method of group method of data handling (GMDH-Combi), and multiple linear regression method (MLR), to address the challenge of estimating the compressive strength of RCA concrete. Performance evaluation of these models utilized metrics such as the correlation coefficient (R), mean square error (MSE), mean absolute error (MAE), and square root mean square error (RMSE). The GMDH-Combi model demonstrated superior performance compared to other methods. In the testing subsets, it achieved R, MSE, RMSE, and MAPE values of 0.89, 53.56, 7.32, and 13.61, respectively. In the training subsets, the model showed R, MSE, RMSE, and MAPE values of 0.87, 53.77, 7.33, and 12.77, respectively.