基于D-InSAR技术和SVR算法的开采沉陷监测与预计  

Combining D-InSAR and SVR for monitoring and prediction of mining subsidence

作　　者：陈炳乾[1] 邓喀中[1] 范洪冬[1] 

机构地区：[1]中国矿业大学环境与测绘学院江苏省资源环境信息工程重点实验室,江苏徐州221116

出　　处：《中国矿业大学学报》2014年第5期880-886,共7页Journal of China University of Mining & Technology

基　　金：国家重点基础研究发展计划(973)项目(2013CB227904);江苏省2013年度普通高校研究生科研创新计划项目(CXZ Z13_0936);江苏省基础研究计划(自然科学基金)青年基金项目(BK20130174);江苏高校优势学科建设工程项目(SZBF2011-6-B35)

摘　　要：针对传统开采沉陷监测和预计方法存在诸多缺陷且两者无法实现集成化的问题,提出了利用合成孔径雷达差分干涉测量技术(D-InSAR)进行开采沉陷监测,并将监测结果与支持向量回归算法(SVR)相结合进行开采沉陷动态预计,最终实现开采沉陷监测与动态预计的一体化.首先利用D-InSAR技术获取开采沉陷的影响范围与发展趋势;然后将监测结果作为SVR算法的训练与学习样本建立预计函数;最后在已建立预计函数的基础上采用滚动预测方法进行开采沉陷动态预计.以陕西省大柳塔矿某工作面为例,采用所提出的方法,使用13景TerraSAR-X雷达影像进行实验研究与分析.结果表明:D-InSAR监测结果能够很好地反映开采沉陷的影响范围与发展趋势,其开采沉陷的最大绝对和相对预计误差分别为19mm,5.4%.实验结果证明了该方法的可行性.The current methods used for monitoring and dynamic prediction of mining subsid- ence carry many limitations and defects; moreover, the tasks of monitoring and prediction of mining subsidence are always performed separately in these methods, rather than combining in an integrated system. Therefore, a model is proposed that integrates differential synthetic ap- erture radar interferometry （D-InSAR） technology and the support vector regression （SVR） al- gorithm to monitor and dynamically predict mining subsidence. The detaile steps can be de- scribed as follows. The D-InSAR technology is first used to monitor the range of influence and the development trend of mining subsidence. Based on the monitoring results obtained by the D-InSAR technology, the SVR algorithm is used to describe the nonlinear function relationship between the monitored data and future subsidence. Finally, a method of rolling prediction based on the established SVR function is adopted to update the training and learning samples of SVR, thus allowing the algorithm to use the latest monitored data to dynamically predict future mining subsidence. To verify the applicability of the proposed methodology, it was applied to Daliuta mining area in Shanxi province, China, where thirteen TerraSAR-X images were used. The experimental results show that the monitoring results very satisfactorily reflect the range of influence and the trend in the development of mining subsidence. The subsidence is predicted with a maximum absolute error accuracy of 19 mm and a maximum relative error of 5. 4%. These values demonstrate the accuracy and feasibility of the proposed model.

关 键 词：开采沉陷 沉陷监测 动态预计 D—InSAR 支持向量回归 

分 类 号：P228.6[天文地球—大地测量学与测量工程]