Prediction of UCs Sterength of Limestone Using Neural Network and ANFIS (Case Study)

Geotechnical and Strength parameters such as uniaxial compressive strength (UCS), modulus of deformation, cohesion and internal friction angle are the most important parameters for practical design as well as numerical modeling in rock mass. UCS as an index parameter is very important. This parameter can be estimated using direct and laboratory methods, which in many cases are expensive and time consuming. So far, various experimental models have been proposed to estimate rock compressive strength using different parameters, mainly for rocks in a particular area and there are many uncertainties for its use in other areas. Artificial neural network is a powerful tool used to create predictive models and studies have shown the superiority of this technique over classical statistical methods. The purpose of this paper is to estimate UCS of limestone rocks using elastic modulus, effective porosity, specific gravity, Poisson's ratio. In this paper, two models of multilayer neural network (MLP) and fuzzy logic adaptive system (ANFIS) are used to model UCS. Based on the results, both constructed networks have acceptable performance for estimating rock compressive strength and selecting the most accurate network based on the quantity and quality of the statistical population used. In this paper, the designed neural network performs somewhat better than the fuzzy logic network.