Research On Slope Monitoring And Early Warning Model Based On LSTM And FEM

Monitoring and early warning is the main means of geological disaster prevention,monitoring is the basis of early warning,and early warning is the purpose of monitoring.In recent years,scholars at home and abroad have conducted in-depth research on the method of slope monitoring and early warning,and have achieved a lot of research results.However,in general,the existing early warning model has three major problems: single consideration factor,short early warning time,it is difficult to provide sufficient time for personnel evacuation,and the slope deformation process is not fully considered,so it is difficult to obtain a good early warning effect.Therefore,the existing research results are difficult to effectively complete the tasks of "what time may occur" and "strive to achieve24-hour early warning" of geological disasters.Based on this,this paper systematically summarizes the author’s practical achievements in monitoring and early warning in recent years.Taking a slope in Quanzhou City,Fujian Province as an example,the long short-term memory network(LSTM)model and the finite element(FEM)model are used to construct a monitoring and early warning model.Model.The future 24-hour displacement prediction model constructed by the longterm and short-term memory network realizes the monitoring and early warning of the slope 24 hours in advance;the monitoring data is combined with the finite element model for real-time stability analysis,which realizes the real-time evaluation of the slope state.The main results obtained in this paper are as follows:(1)Ground disaster surveys and monitoring points were arranged for the disaster sites:monitoring equipment was laid out for the displacement,pore water pressure,groundwater level,soil moisture content and rainfall of the disaster sites according to the hydrogeological conditions and soil material properties of the disaster sites;(2)Construct a displacement prediction model for slope in the next 24 hours: The LSTM model is optimized by sparrow search algorithm(SSA),and an SSA-LSTM displacement prediction model is constructed.Achieve accurate prediction of displacement changes on slopes over the next 24 hours;(3)Stability analysis of slopes: A slope stability analysis model based on strength reduction was established by using the finite element software ABAQUS.According to the stability analysis results,the danger area of the landslide was judged,and the current state of the slope was further understood;(4)A comprehensive monitoring and early warning model of slope is established: the displacement prediction model is updated through the online transfer learning method,which realizes the accurate prediction of slope displacement change in the monitoring process,and quantifies the displacement result into an early warning level in combination with the improved tangent angle model.When the warning level is at the warning level,the real-time stability analysis model of the driving slope evaluates the current state of the slope according to the monitoring data.Through the joint work of the neural network part of the early warning model and the finite element part,the comprehensive early warning of the slope is realized.