Stone is one of the materials that has always been considered because of its strength, resistance and beauty. According to the use of stone as a decoration of facades, it is necessary to double its beauty by performing operations and processes on it. One of the important stages of this process is the sub-stone stage, which takes into account the characteristics of the stone and its application, and separate methods and materials are used for each stone. In this study, with the aim of investigating the effect of physical-mechanical and lithological properties on the final quality of the stone surface after subbing, 15 granite samples from different quarries in the country were selected. First, the physical-mechanical and lithological properties of the samples were determined, and then the samples were subjected to Saab treatment. Numerical values of gloss and roughness were measured with a gloss meter and roughness meter, before and after the sub operation. The average of the measured values was calculated and three indices of gloss improvement (GII), roughness improvement (RII) and quality improvement (QII) were defined for the samples. Using SPSS statistical software and the collected data, the factors affecting the quality of the samples were investigated after sub. Among of the physical-mechanical properties, the Schmitt face hardness with a coefficient of determination of 0.803 and among the lithological and texture properties, the texture coefficient with a coefficient of determination of 0.376 were among the most important factors affecting the quality improvement index of the stone surface than All the properties of the stone were identified as the main factor on the quality of the stone surface after subbing. Based on this, linear and non-linear equations were generalized to measure gloss, roughness and surface quality of the samples, and the coefficient of determining linear and non-linear relationships for the three indicators of gloss improvement, roughness improvement and stone surface quality improvement were respectively 0.727, 0.826, 0.572, 0.618, 0.723, 0.804 were obtained. In order to accurately eva‎luate and eva‎luate the performance of the developed model, the statistical indicators of normalized root mean square error (NRMSE) and error variance (VAF), coefficient of determination (R2) and PI index were calculated for the developed model. Due to the higher numerical values of non-linear relationships compared to linear relationships, it was found that non-linear equations can provide more accurate predictions.

ابراهيم قاسمي

مهدي نصيري سروي , راحب باقرپور