Comparative Studies Of Artificial Neural Networks And Adaptive Neuro-Fuzzy Inference System Based Approach For The Circular Sliding Slopes Stability Analysis

This thesis deals with the following aspects concerning the adaptive neuro-fuzzy inference system based approach for the circular sliding slopes stability analysis.(1) A comparision study was conducted to the fitting capability and popularization forecast of the adaptive neuro-fuzzy inference system and artificial neural networks by using a multi-peaks function. The results show that the adaptive neuro-fuzzy inference system is faster in convergence and more competent for prediction and more suitable for prediction of the circular sliding slopes stability analysis. Meanwhile ANFIS has better stability.(2) An artificial neural network system and an adaptive neuro-fuzzy inference system based approach were established by using the artificial neural network system and the adaptive neuro-fuzzy inference system. The approaches have been used for analyzing the problem of circular sliding slope stablity. The research results show that the later approach has relatively higher speed of convergence and better stability and higher accuracy of inversion and can satisfy the requirements of engineering application.(3) The approach of adaptive neuro-fuzzy inference system has been used to calculate safety factor and stability state of Pan-Gu Temple slope beside Henyang-Zaomuppu highway in Hunan. Province and predict the stability state. The inversion results were used to forcast the stability of the slopes. It has been proved that the fact on building site are in good agreement with the simulation results. At meanwhile we also estimate and analysis the reinforcement project of the father design.