Inversion Method Based on IPOA-RELM for Borehole Resistivity Monitoring of Mining-Induced Fractures in Overlying Rock

Summary As coal mining progresses, the overlying roof rock is damaged, forming mining-induced fractures. To perform full-space resistivity inversion of borehole resistivity data for these fractures, a new method is proposed that combines Tent mapping, nonlinear convergence factors, the improved Pelican Optimization Algorithm (IPOA), and Regularized Extreme Learning Machine (RELM). Compared to other neural network methods, this algorithm offers more efficient and accurate inversion in complex geological environments. Applied to field data from a coal mine, the inversion identified the first major fracture zone development at 41 meters and the maximum development of the water-conducting fracture zone at 51 meters. The results align well with field data from double-ended blocking segmented water injection tests, providing technical support for accurately monitoring mining-induced fractures.