| Disasters caused by linear ground features and urban surface deformation will bring great threats to the society.Compared with conventional GNSS technology,interferometric Syometry Aperture Radar(InSAR)is an active remote sensing technique for obtaining surface deformation information at all times of the day.With the advantages of being free from the special weather restrictions such as cloud,rain and fog,it has become one of the geodetic techniques for high resolution large-area deformation monitoring.In this paper,the Dashengguan Yangtze River Bridge was firstly taken as an application case,the SBAS-InSAR technology was used to monitor its deformation characteristics,the BP neural network time series prediction model was introduced,and the existing deformation time series was used to establish an optimal model to predict the short-term shape variables of the bridge,providing model support for the subsequent research.Then,taking Yinzhou District of Ningbo City as the research object,aiming at the problem that the offshore of Yinzhou District is seriously affected by water vapor,this paper respectively uses GACOS and surface meteorological information atmospheric correction model to reduce the influence of atmospheric delay,improve the monitoring accuracy of InSAR,obtain more accurate surface deformation results,and provide accurate deformation data for the study of the subway deformation of Yinzhou District.The characteristics and causes of settlement of metro and its surrounding areas are analyzed,and the BP neural network time series prediction model is used to predict the short-term future shape variables of metro area,and its accuracy is evaluated.The main contributions and conclusions of this paper are as follows:(1)The SBAS-InSAR technology was used to obtain the deformation characteristics of Dashengguan Yangtze River Bridge in 2018-2019,and the BP neural network model was established by using the deformation rate and time deformation series data respectively to establish the optimal deformation prediction model and realize the short-term prediction of the bridge deformation.The mean absolute error(MAE),mean square error(MSE)and root mean square error(RMSE)are 1.28 mm,1.34 mm and 1.53 mm respectively,indicating that the time series deformation model can make good use of InSAR monitoring results to accurately predict the short-term future shape variables of the bridge.(2)The SBAS-InSAR technology was used to obtain the 2018-2020 deformation characteristics of Yinzhou District of Ningbo City,and the GACOS model and the atmospheric correction model of surface meteorological data were used to reduce the impact of atmospheric delay on SBAS-InSAR technology.The atmospheric correction model of surface meteorological data had higher accuracy,and the phase standard deviation of 106 interference pairs was improved.The corrected monitoring value is closer to the precision level monitoring value of the same period.According to the deformation monitoring data and the comprehensive analysis of field exploration,the land subsidence of Yinzhou District is mainly due to the engineering subsidence caused by frequent infrastructure construction.(3)Focusing on the analysis of the deformation characteristics of Yinzhou Metro settlement area,the main reason is the settlement caused by construction.The BP neural network time series prediction model was established on six subway lines according to the existing time deformation series,and the mean square error of the six subway lines was1.392 mm,2.099 mm,1.432 mm,1.132 mm,1.524 mm and 3.150 mm,respectively,by comparing the predicted value with the expected value.The results show that BP neural network can make good use of InSAR time series deformation monitoring data of subway lines to predict the future deformation,and further demonstrate the feasibility of combining BP neural network time series prediction model and temporal InSAR technology for deformation monitoring. |
