ABSTRACT

The mining of underground coal resources causes long-period and large-gradient deformation of the overlying surface of the working face. Due to the characteristic of long period of deformation, D-InSAR (Differential Interferometry Synthetic Aperture Radar) technology can be easily affected by temporal incoherence. When the surface deformation gradient exceeds the monitoring gradient of D-InSAR technology, the conventional D-InSAR technology will easily cause unwrapping failure, and thus is difficult to realize 3D (Three Dimensional) monitoring of mining-induced surface deformation. In order to solve the problems, a prediction method of 3D surface deformation based on D-InSAR and constraints of DPIM (Dynamic Probability Integral Method) model was studied in this paper. Firstly, SAR image data of the working face after the end of mining were processed by differential interference to obtain the LOS surface deformation during a period of time after the mining. Secondly, according to the relationship between D-InSAR LOS (Line Of Sight) deformation and 3D surface deformation, the D-InSAR observation equation for mining subsidence based on DPIM constraints was constructed. Then, the solution method of the condition equation of DPIM-InSAR (Interferometry Synthetic Aperture Radar) based on GA (Genetic Algorithm) was constructed. Finally, based on obtained LOS deformation, the 3D surface deformation of surface subsidence was obtained using the 3D prediction method based on DPIM-InSAR. The results of simulation experiments effectively verified the feasibility of the method. Differential interference processing was performed for Sentinel-1A data after the mining. The LOS deformation of the 1613 working face of Guqiao South Mine between 20 January 2018 and 25 February 2018 was obtained. 3D surface deformation from the beginning of the mining to completion was obtained using the proposed prediction method of 3D surface deformation and subsidence based on DPIM-InSAR (Dynamic Probability Integral Method). The predicted subsidence values were compared with the measured values. The results showed that the fitting errors of the LOS deformation were mostly within 3 mm. The fitting mean square error was ± 2.37 mm (accounting for 6.7% of the maximum LOS deformation value), and the fitting accuracy was high. The fitting mean square error of subsidence was ± 120 mm (accounting for 7.3% of the maximum subsidence). The results in engineering application show that the proposed 3D predicting method of surface deformation and subsidence based on DPIM-InSAR has certain practical values.

摘要

地下煤炭资源的开采导致工作面上覆面长期大梯度变形。由于变形周期长的特点，D-InSAR（Differential Interferometry Synthetic Aperture Radar）技术容易受到时间不相干的影响。当地表变形梯度超过D-InSAR技术的监测梯度时，传统的D-InSAR技术容易造成解包失败，难以实现对采矿诱发地表变形的3D（三维）监测。针对上述问题，本文研究了一种基于D-InSAR和DPIM（Dynamic Probability Integral Method）模型约束的3D表面变形预测方法。首先，采结束后工作面的SAR影像数据通过微分干涉处理得到采后一段时间的LOS地表变形。其次，根据D-InSAR LOS（Line Of Sight）变形与3D地表变形的关系，构建了基于DPIM约束的开采沉陷D-InSAR观测方程。然后,构建了基于遗传算法的DPIM-InSAR(干涉合成孔径雷达)条件方程的求解方法。最后，在获得的LOS变形的基础上，采用基于DPIM-InSAR的3D预测方法得到地表下沉的3D地表变形。仿真实验结果有效地验证了该方法的可行性。对Sentinel-1A数据挖掘后进行差分干扰处理。获得了古桥南矿1613工作面2018年1月20日至2018年2月25日期间的LOS变形量。使用基于 DPIM-InSAR（动态概率积分法）的 3D 地表变形和沉降预测方法，获得了从开采开始到完成的 3D 地表变形。将预测的沉降值与测量值进行比较。结果表明，LOS变形的拟合误差大多在3mm以内。拟合均方误差为±2.37 mm（占最大LOS变形值的6.7%），拟合精度高。下沉拟合均方误差为±120 mm（占最大下沉的7.3%）。