Currently, the development of accurate and reliable models for predicting the behavior of rock mass joints is one of the most common interests among researchers, engineers and geologists. An alternative to address this type of problem more efficiently can be neuro-fuzzy systems, which combine the advantages of Fuzzy Controllers and Artificial Neural Networks (ANN). Therefore, the objective of this paper is to use Adaptive Neuro-Fuzzy Inference Systems (ANFIS) to predict the shear strength and corresponding dilation of unfilled discontinuities of rock masses, incorporating the uncertainties of the variables that govern their shear behavior. It was found that the proposed ANFIS models can be considered a useful tool to predict the shear behavior of clean discontinuities, as they require only some information about the joints characteristics, the intact rock that constitutes their walls, and the boundary conditions imposed on them, without the need for costly and complex laboratory tests.
The development of accurate and reliable models for predicting the behavior of rock mass discontinuities is one of the most common interests among researchers, engineers and geologists. However, Grima (2000) points out that exact solutions to problems in engineering geology rarely exist, since the relationships between the different variables that describe a given geological phenomenon are not known precisely, besides the intrinsic presence of uncertainty and imprecision in their values.
An alternative to address problems of this type more efficiently may be the neuro-fuzzy systems, which combine the advantages of Fuzzy Controllers and Artificial Neural Networks (ANN). While the first one attempts to reproduce the psychology of the human brain by means of the Fuzzy Logic (Zadeh 1965), the second simulates its physiology. According to Singh et al. (2012), this combination provides an interesting tool to describe complex, nonlinear and multivariate problems, such as those observed in geotechnical works designed and built-in rock masses. Regarding rock joints, Matos et al. (2018), Matos et al. (2019) and Dantas Neto et al. (2019) reported that the use of neuro-fuzzy systems can be a pertinent resource to model their shear behavior.
