The Analysis Of Bank Slope Multi-fields Monitoring Data Based On BP Neural Network

Landslide is one of the geological disasters which occur the most frequently and cause the most serious losses. No matter from the consideration of China’s disaster prevention and mitigation, the national requirements of the Belt and Road construction, or the landslide disaster prevention and control theory and technology innovation, setting up the slope multi-fields monitoring and forecasting system and strengthening the research of slope stability, will both have important theoretical and realistic significances.This paper first introduces the present situation of the slope monitoring and the analysis method of slope stability, and analyzes the feasibility of artificial neural network theory which is applied to the analysis of slope multi-fields, and then around this central subject to obtain and analyze the multi-fields information data of the bank slope, combined with slope monitoring of Majiagou in the Three Gorges, It sets up a optical fiber multi-fields monitoring system of slope, and obtains the multi-fields information data of the bank slope; On this basis, the BP neural network model of the multi-fields information of the bank slope is established. This paper analyzes the correlation between the multi-fields by using the method of association rules, Finally, through the training of the BP neural network model, it makes an sensitive analysis of the effect factors of slope stability and builds the nonlinear mapping relationship between the multi-fields of slope. The paper’s main achievements are as follows:(1) This paper successfully introduces artificial neural network theory into the multi-fields analysis of slope stability, demonstrates the feasibility and advantages of this method which is applied to the analysis of the slope stability, and shows the application prospect of this method.(2) This paper summarizes basic principle of DFOS, analyzes the technical superiority of DFOS, on this basis, it establishes the bank slope multi-fields monitoring system, and puts forward the monitoring plans of the deformation field, temperature field, seepage field and stress field.(3) This paper establishes the BP neural network analysis model of slope fields data based on software Clementine, introduces the general steps of the model which is used for slope stability analysis, and provides a new method for the slope stability analysis.(4) This paper Introduces the method of association rules in data mining to analyze and deal with the slope fields information data, dig out the rule of high and low water level which affects the rate of slope deformation, and establishes the close links between the seepage fields temperature field、stress field and deformation field.(5) Combined with monitoring projects of the Majiagou in the three gorges, It sets up the analysis model of BP neural network, analyzes the sensitivity of the slope stability, and quantitatively calculates the water level of influence coefficient which is 0.558, explaining that this deformation type of Majiagou slope belongs to the water leading type; Through the model evaluation, the predicting deformation is in line with the target deformation, accordingly, the nonlinear mapping relationship between the temperature field, seepage field, stress field and deformation field are established.