Prediction of HFRC compressive strength using HS-based SIRMs connected fuzzy inference system

A single input rule modules (SIRMs)-connected Fuzzy Inference System optimized by Harmony Search model is used to predict compressive strength for steel-polypropylene hybrid fiber reinforced concrete (HFRC). The proposed model is evaluated using a dataset with 114 real experimental steel-polypropylene data gathered from literature. 82 data (72% of the total data) were used in training the model. Predictions from the model for both training and testing datasets achieve coefficient of determination more than 0.96 with root-mean-square error (RMSE) less than 2.40 MPa. These results show that the proposed model gave accurate and reliable predictions of steel-polypropylene HFRC compressive strengths. The predictions from neural networks model gave accuracy with coefficient of determination more than 0.89 with RMSE 3.89 MPa, while predictions from the multiple regression analysis model achieve accuracy with coefficient of determination 0.95 with RMSE 2.59 MPa. The proposed model gave better accuracy for testing dataset, when compared to neural networks model. Comparison with multiple regression analysis model showed that the proposed model gave better predictions for training and testing datasets. These results show that the proposed model can be used for HFRC mix design in deciding a suitable mix proportion for HFRC.