Journal of Petroleum Gas & Chemical Engineering
ISSN: 3069-2075 | DOI: 10.64030/3069-2075
Jae-Myong Li, Kum-Hyok Choe, Jong-Chol Pak, Yong-Il Ri and Kyong-Mun Choe
Water inrush is a critical geological hazard restricting safe and efficient mining operations in the Namdock Coal Mine, which is situated in a complex hydrogeological setting with developed faults, fractured aquifers, and variable lithological compositions. This study systematically characterizes the hydrogeological features of the mine area and proposes a novel water inflow prediction method integrating the Entropy Weight Method (EWM) and Adaptive Neuro-Fuzzy Inference System (ANFIS). First, field discharge observations and drilling surveys were conducted to clarify the mine’s hydrogeological structure, including water-bearing strata distribution, fault-conductive pathways, and seasonal water inflow variability. Then, the EWM was applied to quantify discharge risk levels for different mining drifts, identifying high-risk zones with entropy weight values exceeding 0.2221. On this basis, an ANFIS model was established using six key influencing factors (mining depth, coal seam thickness, dip angle, hanging wall failure degree, geological structure, and season) as inputs and measured water inflow as output. The model was trained with 25 groups of field data and validated with 5 groups of test data, achieving a low-test error of 1.0158%— significantly outperforming the traditional BP neural network (8.56% test error). Field application in the mine’s 9-Pit area demonstrated that the integrated method could accurately predict water inflow in high-risk drifts and guide the optimization of mining sequences. This research provides a scientific and efficient technical tool for water inrush prevention in anthracite coal mines with complex hydrogeological conditions.