Novel hybrid models to predict the capillary water absorption values of building stones

Abstract

The water that moves upward inside the stone is affected by capillary water absorption (CWA). This upward trend in historic and old buildings causes rapid destruction. In the simple regression model, the relationship between CWA with resistance (compressive strength CS) and its physical characteristics, such as (water absorption by weight ( W_a), surface porosity (n), P-wave velocity (V_p), dry unit weight (ρ_d)) can be extracted. Several samples were collected from several rocks. To aim the prediction design, SVR and ANFIS models were proposed, in which two main ANFIS variables and the optimal quantity of SVR’s decisive variables were indicated via linking with a multi-verse optimizer (MVO) algorithm. So as to develop models, the gathered data records for estimating CWA from published articles were separated into two parts after being sorted randomly. Numerous stones containing igneous, metamorphic, and travertine were gathered from Turkey and Anatolia to use in experimental affairs. The outputs of hybrid models for predicting the CWA of building stones depict that both models have the justifiable capability in the prediction process, indicating the acceptable correlation between recorded and predicted values of CWA. All in all, it is proved that the MVO-SVR has the best performance among the comprising approaches.