Wyniki wyszukiwania - BazTech

Ograniczanie wyników

1 Acta Geophysica

1 2021

Znaleziono wyników: 1

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: intact rocks

Sortuj według:

Ogranicz wyniki do:

Prediction of physico-mechanical properties of intact rocks using artifcial neural network

Hassan Amjed, Sanuade Oluseun A., Olaseeni Olayiwola G.

Acta Geophysica

2021

Vol. 69, no. 5

1769--1788

The determination of the physico-mechanical characteristics of rocks is very essential for the planning and implementation of engineering structures as well as for the classification of the rock mass. These physico-mechanical properties are often obtained directly in the laboratory by using standard tests on specific core samples and/or cut samples. However, these experiments are difficult to perform, destructive, time-consuming, costly, and are impossible to execute in some cases due to the complex nature of some rocks. Hence, there is a need to develop an indirect approach to estimate these physico-mechanical properties of rocks. The artificial neural network (ANN) technique has been proven to be well suited for developing predictive models for the estimation of the physico-mechanical characteristics of rocks. Therefore, this study presents new ANN models to predict uniaxial compressive strength (UCS), dry unit weight (DUW), Brazilian tensile strength (TS), point load index (Is(50)), porosity (ɸ), and the Schmidt hardness (RN) based on the seismic P-wave velocity (Vp), and to compare the ANN models with conventional empirical models. Three error indexes including determination coefficient (R2 ), average absolute percentage error, and root mean square error were determined to assess the reliability of the newly developed ANN models. The results show that the developed models were able to predict the UCS, DUW, TS, Is(50), ɸ and the RN from the Vp of intact rocks with high accuracy, determination coefficient (R2 ) of more than 0.89 was achieved. The ANN models also showed better performance compared to the conventional empirical models.