Research On Multi-index Monitoring And Forecasting Algorithm Of Rainfall Type Rock Landslide

Our country is one of the worst countries in the world for landslides.Landslides are gravity-driven movement of rocks,soil and debris down a slope and can cause serious economic losses and human casualties.With the development of science and technology and people’s in-depth research on landslides,a variety of monitoring means are constantly applied in landslide monitoring.How to choose the appropriate monitoring methods for the types of landslides,and comprehensively use these monitoring data to build an accurate and reliable landslide displacement prediction model and effectively improve the accuracy of landslide prediction are the difficulties and hot spots of current research.In order to solve these problems,this thesis researches landslide displacement prediction and constructs a landslide displacement prediction model on the basis of in-depth study of deformation monitoring theory and data processing methods for rainfall-type rocky landslides,taking the landslide in Huangshi City,Hubei Province as an example.The main research contents of this thesis are as follows.1.The landslide monitoring scheme is formulated according to various factors affecting the deformation of rainfall-type rocky landslides,including surface displacement monitoring,infrasound monitoring,rainfall monitoring and water content monitoring,and the data transmission methods commonly used in landslide monitoring are introduced.2.To address the problems of discrepancy,incompleteness and irregularity of the actual sensor data,a variety of landslide monitoring data pre-processing methods are studied,mainly the abnormal data detection method,the missing data filling method and the data normalization method.And a brief analysis of the landslide prediction dataset is carried out.3.For the surface displacement prediction problem of rainfall-type rocky landslides,a BP neural network landslide displacement prediction model based on the optimization of GA algorithm and PSO algorithm is proposed.The experimental results show that the BP neural network model is suitable for landslide displacement prediction,the GA algorithm and PSO algorithm can optimize the BP neural network well,and the RMSE of the PSO-GA-BP neural network fusion model reaches 0.901mm,MAE reaches 0.611mm,and the goodness-of-fit R~2reaches 0.979.The fusion model can obtain satisfactory fitting results and maintain a high prediction accuracy.