Construction Of Mining Goaf Subsidence Model By Integrating Time Series InSAR And Improved Probability Integral Metho

Mineral resources are essential to the foundation of national economic and social development,serving as the pillar of industrial modernization and social progress.However,the environmental problems resulting from mining operations remain a challenge,with the issue of settlement being particularly prominent.This can result in surface structural damage that impacts mine safety and puts people’s lives and property at risk.Timely monitoring of surface deformation in goafs is therefore crucial.As surface monitoring technology in goaf areas continues to develop,InSAR technology has emerged as a promising method for mine monitoring,offering new perspectives on the issue.InSAR technology also has certain limitations in mine monitoring,such as noise,atmospheric delay error,monitoring periodicity and other factors,and the surface settlement speed caused by mine mining is fast,the shape variable is large,and spatial discontinuities are formed,resulting in the problem of low accuracy or local missing of the final monitoring results.In order to improve the monitoring accuracy of InSAR in the settlement of the mine surface,obtain accurate settlement area location information,and expand the scope of InSAR technology in practical application,the probability integral method in the mine mining subsidence prediction method is introduced,which has high monitoring accuracy in the settlement center of the goaf area,and is usually applied to the monitoring of surface deformation in the goaf of coal mine,but the convergence speed is too fast when calculating the settlement value at the edge of the goaf,thus affecting its overall monitoring accuracy.In order to solve this problem,this paper uses different algorithms to invert the parameters of the probabilistic integral method model,and constructs the optimal goaf deformation inversion model,so as to realize the organic combination of InSAR and probability integration method.The main research contents and conclusions are as follows:(1)The Small Baseline Subset InSAR(SBAS-InSAR)technique was utilized to monitor surface deformation in different underground mining areas.Sentinel-1A data from 2021 to 2022 were used as monitoring and selection periods for both the Dahongshan iron ore mining area and a coal mining area in Ordos City.Based on this,surface deformation areas caused by mining activities in these two areas were extracted,and it was discovered that some areas exhibited incoherence or low coherence.A total of 4 areas of hidden danger were identified in the iron ore mining area of Dahongshan,and 2 large areas of surface deformation hidden danger were identified in the coal mining area in Ordos City.(2)Combined with the actual site conditions,the C area of Dahongshan iron ore mining area poses significant hidden dangers,as the middle area has lost coherence.The segmented caving method without a bottom column is used for mining in this area,which easily damages the stability of the underground mining surface roof,the main factor causing deformation in the C area.In a coal mining area in Ordos City,Area A is located under buildings and is a key deformation monitoring area.The use of the all collapse method to manage the roof can cause damage to the overlying rock and eventually lead to surface settlement.Due to coherence issues in the monitoring results of the two mining areas,the feasibility of using PS-InSAR technology to monitor mining areas with higher coherence was explored.The accuracy of surface deformation monitoring results in the goaf using PS-InSAR technology was significantly higher than that of SBAS-InSAR technology,with small differences from the actual monitoring results.However,the number and density of permanent scatterers in the mining area limited the continuous surface deformation results that could be obtained using PS-InSAR technology.(3)The deformation feature points were extracted by selecting the monitoring points with a coherence greater than 0.5 in the SBAS-InSAR deformation results.The probability integration method was then employed to monitor the settlement of the mining area.In the fusion process,several algorithms including the improved particle swarm algorithm,sequential quadratic programming algorithm,XGboost algorithm,and traditional empirical parameter method were utilized to invert the parameters of the probability integration method.From the final settlement zone model,it can be observed that the combined XGboost algorithm demonstrated better accuracy in the inversion of the two mining areas compared to other methods.The average absolute error,maximum settlement point absolute error,and root mean square error obtained by this method and the actual monitoring results were 21 mm,11mm,and 26 mm,respectively.In the goaf area of a coal mine in Ordos City,the average absolute error,the absolute error of the maximum settlement point,and the root mean square error were 12 mm,7mm,and 17 mm,respectively.The experimental results indicate that the settlement zone model established by the fusion method and XGboost algorithm can effectively compensate for the limitations of SBAS-InSAR monitoring of mine settlement area center under the condition of insufficient on-site monitoring data.It enhances the surface settlement monitoring ability and expands the application scope of SBAS-InSAR monitoring goaf in mine goaf monitoring.The applicability of the probability integral method in non-coal mines has also been verified.