A neural network model for slope stability computations

Abstract

This paper pertains to the prediction of stability of slopes by establishing empirical relations for stability coefficients adopting machine-learning process analysing available data (computed by well-known and established limit equilibrium methods) in the form of tables or charts. Stability coefficients obtained by using the developed equations are subsequently used for evaluating the factor of safety of homogeneous slopes. As in the developed expressions, all information reported in the tables/or charts are implicitly built in, their use removes the tediousness and mistakes that generally occur in present practice of using either tables or charts. Unlike the other available statistical or artificial neural network models that find the factor of safety values directly, the present model estimates the same using the predicted stability coefficients including the depth factor as well and is more versatile in its applicability. © 2018 ICE Publishing: all rights reserved.