Based on the mining of the thick No. 3 coal seam in Majiliang Coal Mine, this study focuses on goaf water inrush, using multiple technologies and theories to explore the formation mechanism of water inrush channels during mining and related monitoring and early warning methods. A 3D FLAC3D model built with the fluid - solid coupling theory simulates the dynamic evolution of stress, displacement, and seepage fields, revealing a 110m - high water - conducting fracture zone. The Variational Mode Decomposition (VMD) technology processes transient electromagnetic signals, and together with distributed optical fiber monitoring, precisely locates water - filled goafs and captures changes in water inrush channels. Laboratory research shows a negative exponential relationship (R2 = 0.98) between the acoustic emission b - value of soaked coal and its water content, with a 50% drop in the b - value warning of coal instability. In the field application, an early warning system with 12 indicators such as stress - strain and water temperature - water pressure was used in the 8210 working face, successfully warning of 3 roof separation risks with an 89% accuracy rate and giving a 48 - hour - early warning, thus offering a theoretical and technical reference for mine water disaster prevention in similar mines.