Monitoring And Prediction Of Regional Settlement Along High-speed Railway Based On InSAR Technology And BP Neural Network

Land subsidence is widely concerned by relevant researchers at home and abroad.It is a geological disaster caused by factors such as urban development and excessive exploitation of underground resources.It has a certain impact on the safe operation of buildings and the South-to-North Water Diversion Project.At the same time,uneven se ttlement occurs in the region,and the settlement difference value is large,which will have a greater impact on the smoothness of the high-speed railway track.It is necessary to study the regional settlement characteristics and the development trend of s ettlement along the high-speed railway.Safe operation is of great significance,and its acquisition of time series settlement values and change trends can provide certain reference information for relevant departments.The main work of this paper is as follows:(1)Using the SAR data of the elevating rail during the period from May2018 to August 2020 in the covered study area,based on the SBAS-In SAR technology,the average annual subsidence rate and the settlement sequence value of the elevating rail w ere obtained,and the regional deformation along the high-speed railway was analyzed.The analysis shows that the average annual subsidence rate and the change trend of the subsidence sequence value under the lifting rail mode have a high consistency.A to tal of 7 more obvious subsidence areas have been detected,with a total subsidence area of 1,620 km~2.The total length of the high-speed railway within the settlement is 121km,and the average annual settlement rate is-26mm/yr-18mm/yr.(2)Taking the average annual settlement rate as the research object,the factors affecting regional settlement are further analyzed,and the possible influence of regional settlement on high-speed railways is also analyzed.The study found that human activities such as oi l extraction and salt fields will affect land subsidence.The different degrees of uneven settlement along the high-speed railway in the study area will affect the smoothness and stability of the rack Certain influence.(3)The BP neural network and its o ptimized three prediction models are used to train and predict the settlement values of seven settlement points in the settlement area along the high-speed railway.The results found that:the two-layer wavelet decomposition of the settlement sequence va lue to reduce noise,the noise reduction effect is the best,and the settlement value sequence value after noise reduction is smoother.The genetic algorithm and particle swarm algorithm have a certain optimization effect,and the optimization effect of ge netic algorithm is the best,The root mean square error of the predicted value is the smallest,followed by the particle swarm algorithm.