Evaluation And Prediction Of Highway Slope Stability Based On The Fuzzy Set Theory

Based on systematically consulting, summarizing and analysing the literature at home and abroad, and combined with field monitoring results, the influence factors on the stability of highway slope were analyzed, and the main factors influencing the slope stability were selected as the evaluation indexes. Combined the theory of fuzzy aggregation, fuzzy pattern recognition and fuzzy optimization to build the fuzzy similar clustering evaluation model of highway slope stability. And introducing the fuzzy similar clustering model into the RBF neural network to build the model of fuzzy similar clustering RBF neural network of highway slope stability. Through research, the following research achievement have been reached:1. Through the research of literature data and field investigation, together with the analysis of present situation of highway landslide disasters, landform characteristics and engineering geological environment in Hunan province, this thesis analyzes, summarizes and classifies the main factors which influence the stability of highway slope to regard the soil bulk density, cohesion, internal friction angle, slope height, slope angle, pore water pressure ratio as the fuzzy evaluation indexes in analyzing the slope stability.2. This thesis uses binary contrast sorting method to calculate the index weight of the six major influencing factors. This thesis also uses the weighted fuzzy clustering algorithm put forward by Professor Chen Shouyu to establish the fuzzy clustering prediction model of highway slope stability, and carries out corresponding improvement about the fuzzy clustering iteration model to optimize the calculation algorithm of the model.3. Combining the fuzzy clustering theory and fuzzy pattern recognition with fuzzy optimization theory, the fuzzy similar clustering model of highway slope stability has been built, the fuzzy clustering algorithm has been further optimized, and the computational efficiency and accuracy of evaluation has also been improved. According to the comparison research about different fuzzy similar clustering levels β, this thesis concludes that the best similar clustering levelβ is 0.8.4. The fuzzy similar clustering model has been introduced into the RBF neural network to build the fuzzy similar clustering and neural network model. Combined with the analysis of engineering instances, the model can reliabily evaluate and predicte the stability of highway slope, which has further optimized evaluation method of the slope stability.