2016, Tese de Doutorado
The stability of slopes is a topic of great interest to the geotechnical engineer, given the significant economic losses, and even human, resulting from the slopes collapse. It’s estimated that the landslides outbreak has already caused thousands of deaths and tens of billions of dollars in annual losses worldwide. The phenomena of instability of slopes are conditioned by many factors, such as climate, the lithology and structures of rock, the morphology, the anthropic and others. The analysis of geological and geotechnical conditions of landslides provides an appraisal of each of the factors involved in the processes of instability of slopes, allowing the achievement of results of interest with regard to the mode of action of factors. The current work aims at the use a Hybrid system that uses the Neural Network and Fuzzic Logic (Neuro-Fuzzy) to create a model that, in qualitative form, provides a prediction of the potential of slope rupture. To fulfill this objective, we studied the factors involved in the processes of instability of slopes, and how these factors are interrelated. Parametric analyzes were carried out to provide data for the Neuro-Fuzzy model. It is presented in this work, one history case well documented for the validation of the Neuro-Fuzzy Model and Among the main findings includes the capability of Hybrid Neuro-Fuzzy Modeling in predicting the potential of slope rupture, appearing as a tool capable of assisting in the detection of slopes with potential for rupture.