Goaf Geometric Parameters Inversion And Dynamic 3-D Deformation Prediction In Mining Area Based On InSAR Technology

Detecting underground goafs and predicting the dynamic mining-induced three-dimensional(3-D)deformation is critical for the research and interpretation of mining subsidence mechanism,prevention and control of potential geological disasters,ecological environment management and mining process optimization.The traditional goaf detection methods are time-consuming,labor-intensive,and expensive.Compared with traditional goaf detection methods,establishing the link between goaf parameters and sufficient ground measurements can realize large-scale goaf detection.The construction of a mining-induced 3-D deformation prediction model also requires a large amount of reliable surface measured deformation.Common geodetic surveying techniques(e.g.,leveling and GPS)can only provide sparse "point" or "line" 3-D deformation observations,and it is difficult to reflect the "surface" mininginduced deformation characteristics.These geodetic techniques hinder the development of large-scale underground goaf detection,and reduce the prediction accuracy of 3-D surface deformation in mining areas.Interferometric synthetic aperture radar(InSAR)technique greatly makes up for the shortcomings of traditional geodetic techniques.Therefore,InSAR technique provides a new development opportunity for the largescale unknown goaf detection and the mining-induced 3-D deformation prediction.The studies on retrieving mining-related goaf parameters are rare.Most of them need to obtain the precise model parameters of the mining subsidence model in advance,which is not suitable for unknown goaf.Therefore,this thesis proposes an inversion method of underground goaf geometric parameters based on cross-iteration.Moreover,it is necessary to know the goaf parameters,the overburden properties,the shape of working face and the model parameters to predict 3-D deformation by using mining subsidence model,which hinders the practical application and development of this method.However,the existing 3-D deformation prediction methods based on single-track InSAR do not consider the spatial continuity between ground deformation points,and cannot obtain the predicted deformation with complete spatio-temporal continuity characteristics.It is difficult to accord with the current situation of increasing deep mining.Therefore,this thesis considers the acquisition of constraint parameters and spatio-temporal prediction,and carries out related research on dynamic 3D deformation prediction based on singletrack InSAR.The major contributions are as follows:(1)Based on the simplified Probability Integration Method(PIM),the functional relationship between the subsampling InSAR observations and goaf geometric parameters is constructed,which provides a theoretical model for the next goaf parameter inversion.Due to the high nonlinearity and complexity of the PIM model and the excessive InSAR-derived LOS(Line of Sight)deformation,constructing the function model will undoubtedly restrict the inversion efficiency of goaf parameters.Therefore,this method firstly simplifies the PIM model to reduce the difficulty of model parameter acquisition.Then,the adaptive quadtree subsampling method is used to effectively reduce the redundancy of the InSAR observations.Finally,a function model of subsampling LOS deformation and goaf parameters is established to improve the inversion efficiency.(2)This thesis presents a cross-iteration-based method for retrieving underground goaf parameters.This method avoids the influence of inaccurate model parameters on the results,and expands the practical application of InSAR technology in mining areas.Aiming at the problem that the PIM model parameters calculated by sufficient measured data are easily affected by missing data and measurement errors.An iterative inversion algorithm with cross-update between grouped parameters is proposed to overcome the influence of model parameter errors.Subsequently,simulated experiment and real experiments were selected to verify the performance and reliability of this method.The results indicate that the inverted goaf geometric parameters are basically consistent with the measured values.Therefore,this algorithm provides a new solution for high-precision unknown goaf detection.(3)This thesis develops a method for predicting dynamic mininginduced 3-D displacements from a single-track InSAR dataset based on the prior constraints and the Space-time Kalman Filter(STKF)model.This method enriches the means of mine disaster risk assessment and prevention.Aiming at the difficulty of obtaining parameters in the mining subsidence model,this method firstly constructs a 3-D decomposition model of LOS deformation according to the proportional relationship between horizontal movement and vertical deformation gradient,and transforms the 3-D deformation prediction into1-D LOS deformation prediction.Then,the STKF model is introduced to solve the problem that the single-point deformation prediction method ignores the spatial relationship between ground deformation points by performing spatio-temporal filtering,interpolation and prediction on the InSAR time series deformation.Afterwards,the constraint parameters obtained by the proposed goaf detection method are used to decompose the dynamic LOS prediction deformation.Finally,the Datong mining area was chosen to verify the reliability and performance of this method.The results show that the accurately predicted 3-D surface deformation has complete spatio-temporal characteristics.