Characteristics inversion of underground goaf based on InSAR techniques and PIM
Hongdong Fan, Tengteng Li, Yantao Gao, Kazhong Deng, Hongan Wu
Abstract
The nature of coal mining areas, such as private exploitation, illegal exploitation, is not only a waste of a large amount of coal resources, but is also very likely to cause geological problems such as surface collapse and cracks. However, existing supervision methods are mostly in the form of mass report, on-site inspection, and geological exploration, which are not suitable for underground goaf detection in large-scale non-target areas. Therefore, this paper presents a method for locating and inversion of underground goaf based on a combination of interferometric synthetic aperture radar (InSAR) techniques and a probability integral model (PIM). Firstly, the Line-of-Sight deformation of the subsidence basin above goaf generated by combining InSAR and Offset-Tracking algorithm is taken as the real land subsidence. Then, according to the obtained surface deformation, the initial ranges of eight goaf locational parameters are determined, and the correlation function between the surface deformation and the location of underground goaf is constructed. The Oppositional-based Learning Chaotic Bat Algorithm (OLCBA) is used to search the optimal parameters to satisfy the minimum error between the ground deformation calculated by PIM and the Line-of-Sight deformation obtained by SAR. Therefore, the optimal parameters are considered as the characteristics of underground goaf. The simulation results show that the maximum relative error appears in the dip angle of the inversed parameters of the goaf location, which is 3.70%. Taking a known working face of the Daliuta coal mine as the test object and 11 TerraSAR-X images as the data sources, the land subsidence is obtained and compared with GPS measurements, then the underground mining characteristics of the working face is successfully retrieved. Although the maximum relative error of the dip angle is 81.5%, the absolute error is only 1.63° and the relative errors of the other main parameters are less than 15.03%.