| The seasonal frost heave deformation of the subgrade in seasonally frozen soil areas is a key factor affecting the construction and safe operation of high-speed railways.The stability and millimeter-level deformation control of the subgrade require greater accuracy in obtaining deformation along the line.However,traditional frost heave deformation analysis methods rely on discrete frost heave monitoring points or laboratory tests to calculate geotechnical parameters for direct model calculations,neglecting the spatiotemporal variability of geotechnical parameters and the existence of various uncertainties.This results in discrepancies between predicted and measured deformation values,making it difficult to reveal the overall spatiotemporal distribution of frost heave deformation.Data assimilation methods can fuse multi-source observation data to invert the probability distribution of forecast parameters in models,providing a new method for reducing parameter uncertainty in deformation forecast models.This paper,based on data assimilation theory,takes the Menyuan section of the Lanzhou-Xinjiang high-speed railway as the research object and conducts multi-scale experiments,multi-physics field analysis,and multi-source information fusion.This establishes a theoretical basis and methodological support for the spatiotemporal prediction of subgrade frost heave deformation and provides an important decision-making basis for dynamic assessment and control of subgrade frost heave risk.The main work and achievements of this paper are as follows:(1)The research status of domestic and foreign subgrade frost heave deformation monitoring,analysis,and prediction is sorted out,the existing problems in current subgrade frost heave prediction research are discussed,and the research content and technical route of this paper are proposed based on this.(2)A frost heave deformation analysis method based on data assimilation and optimized coupling between SAR post-processing deformation and induced factors is proposed.The SAR observation data is calculated based on PS-InSAR and post-processed to obtain the frost heave deformation.A frost heave model function is constructed based on the Taylor series expansion for the relationship between time,temperature,and deformation.An Ensemble Kalman Filter(En KF)coupling post-processing displacement and frost heave model framework is established,and the quantitative response relationship between subgrade frost heave displacement and inducing factors is explored.The results show that periodic displacement changes have a phenomenon of winter uplift and summer subsidence,and this periodic fluctuation is strongly negatively correlated with temperature changes.After data assimilation,the induced parameter results are continuously optimized,the uncertainty of model parameters is gradually reduced,and the predicted values approach the observed values,with good consistency in time and space.(3)A prediction method based on the En KF algorithm to dynamically correct the parameters of the water-heat-force multiphysics coupling model is proposed.A water-heatforce multiphysics field is constructed based on COMSOL,the sensitivity of parameters to frost heave results is analyzed using the Sobol method,and the influence of different ensemble sizes and observation errors on frost heave assimilation results is analyzed.The results show that as the assimilation progresses,more observation data are integrated into the stress-strain model,causing the parameter estimates to be increasingly affected by the measured data,and the overall predicted results are relatively close to the observed data.In addition,the sensitivity of the subgrade filler layer to frost heave is higher than that of the graded layer,and the ensemble sample size and observation error affect the assimilation effect.(4)Based on the Iterative Ensemble Kalman Filter(i En KF)assimilation algorithm,the spatial variability of the model parameters for frost heave mechanisms is investigated.By constructing a COMSOL and data assimilation framework,the parameters of the subgrade are inversely estimated.The results show that this method can obtain a good estimation of the spatial variability of geotechnical mechanical parameters.Furthermore,as the i En KF assimilation algorithm gradually merges the displacements of various observation points and undergoes different iterations,the mean value of the assimilated Young’s modulus tends to approach the reference field. |
