Evaluation of the relationship between the surface hardness of magmatic building blocks and uniaxial compressive strength values with regression analysis and artificial neural networks

Uniaxial compressive strength (UCS) values of rocks are the most important input parameter in rock mechanics and engineering applications. This parameter can be determined by laboratory tests and indirect methods. This study aimed to predict the UCS value with two different non-destructive testing techniques. To this end, the uniaxial compressive strength (UCS) and the values of Leeb hardness (HL) with low application energy and Schmidt hammer rebound hardness (SHR) with high application energy, which are among non-destructive testing techniques, of 95 different magmatic rocks (plutonic, volcanic, and pyroclastic) were determined. Simple regression (SR), multiple regression (MR), and artificial neural network (ANN) methods were employed to predict the UCS value. The models obtained using these methods were compared with each other. It was revealed that the model developed by ANN had the highest correlation number.