Through decades of development, the analysis theories about slope stability have generally formed a relatively perfect system. The typical methods are the limit equilibrium method, the limit analysis method and FEM. But the traditional methods may meet difficulties because the factors which affect the slope stability are random and fuzzy variables. This paper does some study on a new method based on the principle of adaptive neural fuzzy inference system. It has certain significance to the theory research and engineering.In this paper, the main theory of adaptive neural fuzzy inference system is expressed system, and it feasibility to evaluate the slope stability is showed. The adaptive neural fuzzy inference model is established for the estimation of slope stability, and the cases of stable and failure slopes are used to train the network. The testing data is used to test the fitting and general-prediction capability of the trained model. The result indicates that the trained model is practical for the estimation of slope stability.This paper also does some deeply study on how to choose the training data, and how to gain the best network structure when using the adaptive neural fuzzy inference system to estimate the stability of the slope. Some principle advices are given on the choice of training data and membership functions. The result may have some effect on promote this method to practical use in engineering.
|