Fuzzy logic approach to conventional rubble mound structures design

Abstract

Stability of rock slopes are usually evaluated by empirical formulae, none of which deal with uncertainties. In fact, stability number and damage level are closely related random variables and their relationship can best be modeled by methods that explicitly take uncertainty into account as vagueness, ambiguities, and imprecision. Fuzzy logic (FL) approach is one of these methods that can deal with nonlinear, complex and uncertain systems. In this paper, the use of non-traditional FL technique is employed as a means to develop efficient predictive model for designing conventional rubble mound structures. A total of 579 experimental small and large scale test data from Van der Meer are used for calibration and verification. FL model results are compared with empirical Van der Meer model in addition to the artificial neural network (ANN) model of Mase et al. It is shown that one can forecast the stability number of conventional rubble mound structures with more significant accuracy by the FL approach than previous models.