Landslide Displacement Prediction And Stability Early Warning Method Under Reservoir Water Level Rise And Fall

Landslides are a common geological disaster,and the number of landslides in China accounts for 70%of the total number of geological disasters in the country every year.In order to reduce the losses caused by landslides,many scholars have conducted a lot of research on landslide prevention and mitigation,and the early warning and forecasting of landslides is the core of disaster prevention and mitigation.In this paper,based on the analysis and summary of domestic and foreign research on landslide deformation and stability analysis as well as early warning and prediction,the research background of Liang Shui Jing landslide in Three Gorges reservoir area is used to study the seepage characteristics,deformation law,displacement prediction and stability analysis of landslide under the condition of reservoir water level rise and fall,and the research work of the paper and the results achieved are as follows:（1）Based on the hyperbolic model of shear displacement and shear stress of soil body,the calculation method of local stability coefficient of landslide under the effect of reservoir water level change and rainfall is deduced.The sensitivity analysis of external factors（reservoir level change rate and rainfall intensity）and soil parameters（cohesion,internal friction angle and shear expansion angle）affecting landslide stability shows that the influence of reservoir level change on landslide local stability coefficient is slightly greater than that of rainfall intensity;internal friction angle>cohesion>shear expansion angle in soil parameters.（2）The seepage characteristics of the landslide in Liangshuijing were analyzed by numerical simulation with different rates of reservoir water level rise and fall（0.2m/d-1.2m/d）and rainfall intensity（10-year,20-year and 50-year event）,which revealed the seepage pattern of the landslide during the reservoir water level change and rainfall.The infiltration line pattern in the process of reservoir water level decrease is roughly in the form of gentle decrease first,then rapid decrease,then slow decrease;the infiltration line pattern in the process of rise is linear decrease from the slip surface,then gentle decrease,then rapid rise to the reservoir water level,forming an inverse curve pattern.The groundwater level in the landslide body lags behind the change of reservoir water level during the change of reservoir water level,and the degree of lag is positively correlated with the rate of water level change.The pore water pressure at the back edge of the landslide body is mainly influenced by the rainfall intensity,and is positively correlated with the rainfall intensity.The pore water pressure in the reservoir level change area and nearby areas is mainly influenced by the rise and fall of the reservoir water level,the pore water pressure at each monitoring point is positively correlated with the fall rate;the pore water pressure at monitoring point B near the reservoir level change area is positively correlated with the rise rate,the pore water pressure at monitoring point C is negatively correlated with the rise rate,and the pore water pressure at monitoring points D,E and F is stable because they are below the water level.water pressure is more stable.（3）Through numerical simulation,different rates of reservoir water level rise and fall（0.2m/d-1.2m/d）and rainfall intensity（10 years,20 years,50 years）are designed to analyze the landslide deformation in Liangshuijing,revealing the law of landslide deformation in the process of reservoir water level change and rainfall,the horizontal displacement of landslide increases and then decreases as the rate of fall increases in the declining stage of reservoir water level,and the horizontal displacement increases and then decreases at 1.0 The horizontal displacement is the largest at 1.0 m/d,and the rainfall when the water level decreases makes the landslide produce obvious displacement.The stability coefficient of the landslide body is negatively correlated with the decline rate of the reservoir water level,but the decrease rate of the stability coefficient gradually slows down.The horizontal displacement of the landslide is positively correlated with the rising rate in the rising stage,and the total displacement is smaller compared with the falling stage.The stability coefficient of the landslide body is positively correlated with the rising rate of reservoir water level,but the rising trend will gradually become slower.（4）The displacement data are decomposed by the CEEMDAN algorithm,and then the hyperparameters of the BiLSTM model of the components are calculated by using Bayesian optimization,and the Bo-BiLSTM model is constructed to predict the displacement components,and a landslide displacement prediction method based on the CEEMDAN-Bo-BiLSTM model is established.The results show that the prediction set error obtained by this method is significantly smaller than that of other displacement prediction models,and the values of RMSE,MRE and MAE are the minimum values,and the deterministic coefficient R² is 0.994,which has the highest prediction accuracy.The predicted displacement data were used to improve the calculation of displacement tangent angle,and the stability judgment of the landslide was made jointly with the results of the percentage of the landslide local stability coefficient F_Di>1(F_Di/n)based on the displacement obtained.