Stability Evaluation and Parametric Optimization of Coal‐Concrete Composite Bearing Systems Under Mine‐Water‐Induced Deterioration: Experiments and FEINN Analysis

ABSTRACT In recent years, the valorization of underground spaces in abandoned coal mines has attracted increasing research attention. Nevertheless, irrespective of the intended application, engineering implementation generally relies on sealed load‐bearing units jointly formed by remnant coal pillars and concrete structures to meet the dual requirements of structural stability and long‐term sealing performance. Therefore, elucidating the load‐bearing and instability mechanisms of concrete‐protective coal‐pillar composite systems under mine‐water exposure, as well as their sensitivity to key parameters, is essential for the safe utilization of underground spaces and the optimization of dam design. This study focuses on the composite dam of the Daliuta Underground Reservoir. A mine‐water‐simulating saline solution was prepared, and uniaxial and triaxial compression tests were performed on coal rock and concrete under water‐induced deterioration to obtain key mechanical parameters and characterize degradation behavior. Building on these results, a coupled FEINN framework integrating physics‐informed neural networks and the finite element method was introduced to systematically evaluate the stability responses of different dam configurations and design parameters under extreme hydraulic pressure. Moreover, a unified normalized multi‐indicator evaluation scheme was established to enable quantitative ranking and selection among multi‐parameter design alternatives. The results provide quantifiable evidence to support the planning, design, operation, and risk mitigation of coal‐mine underground reservoirs, and offer theoretical foundations for the green reutilization and long‐term safe operation of underground spaces in closed mines.