Research And Application Of Spatio-temporal Correlation Model Of Surface Movement And Deformation Response In Deep Mining

With the large-scale mining of coal resources,coal resources with better shallow and better geological conditions have been basically exhausted,and most coal mines in China have switched to deep-well mining.The amount of coal under buildings,railways and water bodies in deep mines is large,so the research on the subsidence law of deep mine is the key to find the way of mining technology.This article aims at common problems of surface monitoring,law analysis,and prediction models in the deep-well mining area.Taking the Huainan mining area as an example,it comprehensively used theoretical research,simulation experiment,field measurement,observation and analogy and other methods to comprehensively analyze the conditions of deep mining.On the basis of the mechanism and law of surface movement and deformation,the research on the prediction model of surface spatio-temporal correlation and the inversion method of geometric characteristic parameters during the deep-well mining was carried out.The main results are as follows:(1)In this paper,the multiple regression model of surface subsidence,movement and deformation parameters was established in Huainan mining area.An improved algorithm integrating extreme learning machine(ELM)was proposed,which effectively improved the accuracy of surface movement and deformation parameters.Based on recent years' data collected from surface observation stations of Guqiao,Gubei,Zhangji,Zhuji,Dingji,Paner,Pansan and Panbei working faces in Huainan mining area,and through regression analysis,the research constructed the quantitative and qualitative relationships among dynamic parameters of surface movement basin(starting distance,subsidence velocity of the maximum subsidence point,duration of surface movement,and etc.),angle parameters(advance effect angle,lag angle of the maximum subsidence velocity,angle of draw,angle of critical deformation),probability integral method prediction parameters(subsidence factor,horizontal movement constant,tangent of main influence angle,propagation angle of extraction)and geological mining conditions(mining height,mining depth,coal seam dip angle,mining dimension of the working face,and etc.).A prediction model of surface movement basin related parameters was constructed by integrating the multiple regression model and ELM.The accuracy and reliability of the model have been verified by taking the subsidence coefficient of a typical mining area as an example.This could provide a new method for the calculation of surface movement basin parameters in mining area lacking of measured data.(2)An improved Boltzmann(IB)mining subsidence prediction model was established,which was suitable for surface prediction of the working face in deep mining.It could significantly improve the prediction accuracy of the whole subsidence basin and the boundary part.Aiming at the non-convergent boundary of probability integral method(PIM)prediction model in deep well,based on the analysis of surface movement and deformation mechanisms,and the idea of combination model and Boltzmann function,a new unit subsidence basin was constructed by combining two different parameters of the unit subsidence basin according to a certain proportion.The prediction formulas of strike main section,dip main section and any point were derived.In order to solve the complex problem of model parameters,the multi population genetic algorithm(MPGA)was introduced into the model solving process.The simulation experiment and engineering case study results showed that the IB prediction model could lead to stable parameter solving results and had good resistance to random errors and gross errors of observation points.In order to analyze the variation law of IB prediction parameters,IB prediction model parameters and PIM parameters of typical deep mining face in Huainan mining area were obtained.Regression equations among IB prediction model parameters,PIM parameters and geological and mining conditions were established by regression.(3)A surface dynamic prediction model was proposed.By combining the proposed model with InSAR data,a three-dimensional dynamic prediction method of surface deformation based on InSAR data at the edge of surface moving basin was developed.By analyzing the change law of the surface movement deformation parameters(IB model parameters)during the mining of 1414(1)working face,a dynamic prediction model was constructed(Dynamic Improved Boltzmann,DIB).The dynamic prediction model parameters based on surface and absolute deformation values were presented.On the basis of this research,and surface deformation obtained by the D-InSAR,a InSAR-DIB dynamic prediction model based on the edge correction model was constructed,effectively extending the application scope of D-InSAR technology in the mining area.(4)A method of obtaining geometric feature parameters of mining was developed based on surface deformation and prior information of surface movement and deformation parameters,which significantly improved the efficiency of obtaining geometric feature parameters in unconsolidated layers in deep mines.First,the parameter system describing mining geometric parameters was established during the mining process and under the stable conditions.Then,with unknown geometric parameters,and known working surface variables and relevant parameters of the predicted model,the mining geometric parameter identification equations were established.Finally,two different inversion methods were built in the mining process of the working surface and under stable conditions.The feasibility of the inversion model was verified by simulation experiments and engineering applications.Figure [91] Table [41] Reference [146]...