Irrigation Loess Landslide Deformation Monitoring And Forecast Model

Loess is widely distributed in China.Due to the unique geological structure of loess,geological disasters in the loess-covered areas are frequent.And it seriously affects local people's property and security.Therefore,it has a great significance to carry out regular monitoring,analysis,and prediction of loess landslides.Landslide Monitoring and Prediction research has always been one of the hot topics in the field of engineering geology.However,most of the studies on the monitoring and prediction of the irrigated loess landslides currently use a single deformation monitoring technique,and the corresponding prediction model has low accuracy and poor applicability.In order to solve these problems Taking the Miaodian Landslide as the research area,this paper uses a variety of deformation monitoring technologies such as the total station and meteorological instruments to monitor the stability of the Miaodian Landslide,and analyzes the formation reasons of the irrigated loess landslide through combined with geological investigation.In addition,this paper also uses several single prediction models and combined prediction models to predict landslide deformation,and analyzes the accuracy and feasibility of these models.The main research contents of this paper are as follows:(1)A comprehensive monitoring network for the deformation of Miaodian landslide is established according to the geological investigation and taking into account factors such as irrigation,rainfall,and topographically complex irrigated loess landslides.This paper uses various monitoring methods to get continuous deformation monitoring data and meteorological and hydrological monitoring data of the Miaodian landslide.(2)Aiming at the problems of the single method for monitoring the deformation of loess landslides at present.This paper compares the monitoring results of the total station,GNSS and crack meter,and analyzes the stability of Miaodian landslide from three aspects: time,space and influencing factors.It shows that there is a strong non-linear correlation between them.The correlation could be described by a power function model.(3)Aiming at the problems of the current landslide single prediction model's accuracy is low,and the data needed for modeling is huge.This paper proposes a prediction of landslide deformation,which uses unbiased grey Verhulst model.It shows that the unbiased grey Verhulst model can completely eliminate the inherent error of the grey Verhulst model,the average absolute error and relative error of the model are 7.3mm and 4.8% respective.Compared with the common landslide prediction model such as exponential smoothing model,kalman filter model and GM(1,1)model,the model has higher prediction accuracy.And it is also suitable for deformation prediction with S-type trend.(4)Aiming at the problems of the poor applicability of the current landslide combination forecasting model This paper uses several prediction models,like linear combination,modified linear combination and combined forecasting theory based on error correction principle,to predict the deformation of Landslide.And the prediction results of these models are compared with that of combination model with optimum weight.It shows that the prediction ability of the modified classic weight method is the most stable,and its extreme minimum prediction error is the smallest.While the combination model with optimum weight has the smallest average absolute error.