Monitoring And Prediction Analysis Of Urban Surface Subsidence Along Rail Transit Lines Based On Time Series InSAR Technolog

Ground subsidence is one of the important hidden dangers threatening the structure safety of large infrastructure buildings.Taking the increasingly built urban mass transit as an example,it is easy to induce the nearby ground subsidence during construction and operation,and it is affected by uneven subsidence.However,the causes of the surface are complex,not only because of its huge building loads and long-term strong shocks and impacts during Metro operation,but also because of the nature of the stratum and hydrological activities.The subsidence along the rail transit is related to the safe operation of the metro and the stability of other construction facilities along the surface,so it is necessary to monitor the subsidence along the rail transit in a short period.At present,the subsidence monitoring along metro lines is mainly based on leveling and GNSS,but the ground subsidence caused by general non-intense geological activities is slower and may be difficult to find in a short time(generally,the period of artificial fixed-point subsidence observation is longer).Therefore,although the traditional subsidence monitoring methods have the advantages of high accuracy,they can not achieve large-scale,high-frequency,all-weather monitoring,and the spatial resolution of data sampling is not high,so they can no longer meet the rapid development of rail transit monitoring needs.Currently,the mature time series In SAR technology can monitor large-scale regional surface with high frequency,all-weather monitoring and low application costs.In this paper,Ningbo rail transit is taken as an example,time series subsidence monitoring analysis is carried out on the surface along the Metro Line Based on time series In SAR technology,and the area producing uneven subsidence is mainly analyzed.The BP network prediction model of genetic optimization algorithm is constructed to predict the subsidence trend of uneven subsidence area,and obtain the preceding information of high-risk subsidence area.This monitoring and prediction system is important for the prevention and control of subsidence.It can quickly provide scientific data reference and decision support for the operation and maintenance of Metro and surrounding facilities.The main work and conclusions of this paper are as follows:(1)Selecting the entinel-1A data covering 50 Metro lines,using PS-In SAR and SBAS-In SAR technology to monitor the urban area of Ningbo city,in order to obtain the change of mean surface deformation rate and the cumulative settlement in each time series,the results of the two methods are compared and verified,and the regression function y=0.9115x+0.4722 of the same name point is obtained,and the goodness of fit R~2=0.901of average subsidence rate is obtained.It is proved that the settlement monitoring results based on PS-In SAR technology are reliable.(2)Analyzing the deformation characteristics of the surface 500 meters along Ningbo urban and rail transit lines,the monitoring data show that the subsidence along and within the metro lines is relatively stable in the areas close to the urban center,most areas are in a stable slight upward trend,the subsidence rate is concentrated between-5mm/a and 8mm/a,and the average subsidence rate of the areas along the metro lines and where obvious subsidence occurs is concentrated between-8mm/a and-20mm/a.The large area of subsidence is mainly concentrated in the northwest and southeast of Ningbo.The maximum subsidence rate is-37mm/a and the accumulated subsidence is over-120mm.Among them,the maximum subsidence rate is-31mm/a and the accumulated subsidence is-112mm in the area along the line,which is located near Dongqianhu Station on,Line 4 of Metro.Among them,the analysis of the settlement profile along Line 1 and Line 2 shows that uneven settlement occurs at several stations and sections,and the causes of the settlement along the metro line are discussed.(3)The prediction model of surface subsidence along metro lines is built by using BP network of genetic optimization algorithm.The prediction of time series of subsidence in uneven subsidence areas is made,the first 45 subsidence data are used to learn and train,and the prediction accuracy error is tested with the last 5 subsidence data.The prediction accuracy and performance of GA-BP model after training are compared with that of traditional BP algorithm.The results show that the GA-BP network model can better predict the surface subsidence along the metro lines in the next five months,while maintaining good performance,and its prediction accuracy is significantly better than the traditional BP algorithm.