Prediction of the Shear Behaviour of Clean Joints in Soft Rocks using Perceptron
Silvrano Adonias Dantas Neto1; Buddhima Indraratna2; David Americo Oliveira2
1 Federal University of Ceara; 2 University of Wollongong
Resumo
It is known that the economic and safe design of structures in rock masses depends on a reasonable prediction of their shear behaviour which is usually governed by the conditions of the rock discontinuities. A number of analytical models predicting the shear behavior of rock joints have been developed on the basis of experimental data obtained from direct shear tests performed under both CNL (Constant Normal Load) and CNS (Constant Normal Stiffness) conditions (Patton, 1966; Barton, 1973, Indraratna et al., 1998, Indraratna and Haque, 2000; Indraratna et al., 2005; etc). However, the use of these analytical models sometimes become difficult due the availability of laboratory tests or even the difficulty of obtaining some of their parameters. The objective of this paper is to present a non-conventional model that can be used to adequately predict the shear behaviour of clean joints based on an artificial neural network (ANN) technique known as perceptron. The laboratory test results presented by Indraratna and Haque (2000) on idealized saw-tooth synthetic rock joints under different boundary conditions (CNL and CNS) were considered in the development of the proposed ANN model. The following parameters were considered as input for the ANN model training: normal stiffness (kN), asperity height (a), initial asperity angle (i0), initial normal stress (n0) and the horizontal displacement (h). The output of the ANN model is the shear stress () at a particular horizontal displacement (h) requested, allowing for the development of the complete stress-displacement curve. The ANN model that presented the best performance in the current study is composed of two hidden layers and an A5-15-5-1 architecture, where each number represents the number of neurons per layer. The coefficients of correlation between the actual test results and the model output values used to evaluate the behaviour of the neural model during training and validation were 0.999 and 0.998, respectively. The shear strength obtained with t
Palavras-chave: rock; clean joints; shear behaviour; perceptron