Research On The Displacement Prediction Of Bazimen Landslide Under The Hysteresis Effect Of Rainfall

Research data shows that geological disasters have occurred frequently in recent years,affecting the operation of public facilities and threatening the safety of life and property.Three Gorges reservoir area is a high-incidence area of landslide disasters,and huge losses are caused by landslide disasters every year.Since the impoundment of the Three Gorges Reservoir in2003,the cyclical fluctuations of the reservoir water level combined with heavy rainfall result in many newly born landslides,aggravating the deformation of the original landslides and resurrect plenty of old landslides.Disaster prevention work has become more severe.This thesis takes the Bazimen landslide as the research object,and studies the deformation characteristics and mechanism of the landslide from internal factors and external factors through a combination of various professional detection data: the water fluctuation period of the Three Gorges Reservoir area,hydrometeorological data in the area and survey reports.According to the regular characteristics of the changes in the accumulative displacement data on the surface in the past few years and automatic surface displacement monitoring data,the hysteresis quality between the deformation of the Bazimen landslide and the rainfall is preliminarily determined by using different data processing methods and analysis angles.Based on the researches above,the hysteresis quality between deformation and rainfall is finally determined by Distributed Lag Model.And then this hysteresis quality was used to predict by Grey Neural Network Model,which can be used for the prevention and treatment of this type of landslide.The specific content and results are summarized as follows:(1)The basic conditions of Three Gorges Reservoir area,including topography,geomorphology,stratum and material composition,geological structure,and hydrogeological environment are explained.The internal and external influence factors that cause the Bazimen landslide deformation are qualitatively analyzed and studied from the macroscopic deformation,deformation characteristics and mechanism over the years.(2)Based on the accumulated surface displacement data of the Bazimen landslide,the total annual and each stage variation of the surface displacement at the monitoring points as well as the cumulative rainfall is calculated,and we analyze and study the influencing factors that lead to the deformation of the Bazimen landslide from the above dimensions.The research results are: heavy rainfall is the main reason that make Bazimen landslide in a state of instability.Its influence on landslide deformation is greater than reservoir water level.(3)According to the data characteristics of the automatic ground displacement monitoring data of the Bazimen landslide,the daily deformation displacement is selected as the explained variable and the daily rainfall as the explanatory variable.Distributed lag model is established,performed and verified by Eviews.It is finally determined that the length of the lag period of the rainfall response is 12 days.The results show that hysteresis quality of rainfall of the Bazimen landslide based on the distributed lag model conforms to the actual law of landslide deformation.(4)According to the established 12-period Distributed Lag Model,the contribution of rainfall to the daily deformation displacement of the landslide is calculated during the lag period.It is found that the contribution of the previous rainfall to the daily deformation displacement of the landslide first increased and then decreased with the extension of time.The second day of the previous rainfall had the largest contribution to the current landslide displacement,which was 17.807%.(5)The accumulated rainfall during the lag period is included as one of the influencing factors into the "small sample" and "non-linear" Gray Neural Network Model,and the data sample of the "step section" of the Bazimen landslide in 2020 is predicted through Matlab data analysis software;According to the prediction results,the average absolute percentage error(MAPE)of the landslide severe deformation stage is 11.983%,the accuracy is 88.017%,and the maximum deviation is less than 2mm,which is within a reasonable range.In the recognition of the daily deformation displacement mutation value and the displacement change trend,the accuracy is high,which can provide a reference for the prevention of the Bazimen landslide.