Abstract: At the present stage, more and more mines in China are entering the deep mining and lower group coal mining period. The coal seam mining is seriously threatened by karst water, the hidden danger of water inrush in the deep mine is unclear, and the lack of scientific and effective water inrush monitoring and early warning means in the mining process is the main reason for the water inrush disaster in the mine.In order to consolidate the foundation of early warning and monitoring of water disasters, a “double drive” micro earthquake early warning framework for coal mine water disasters is proposed. Using data drive and model drive, the risk level and scope of water inrush from the floor of the working face are monitored dynamically in real time, and the trend of water inrush risk is predicted intelligently.Under the framework of data driving, taking the temporal and spatial evolution law of microseismic events as the breakthrough point, through the inversion and attribute analysis of the source mechanism of microseismic events, it provides a basis for judging the triggering cause of microseismic events, the rupture trend and the formation of water diversion channels. In combination with the changes of hydrological dynamic data, it establishes the corresponding criteria for water inrush and evaluates the risk of water inrush.Under the framework of model driven, a deep learning model with multiple algorithms such as classification prediction and cluster analysis is constructed. The typical microseismic event cluster is used as the model input to quantitatively and dynamically predict the spatial range and concentration of future microseismic events, and then determine the water inrush risk level and risk area. Based on the “double drive” early-warning technology of coal mine water disaster based on microseismic data and model, the corresponding regional 3D intelligent early-warning platform of coal mine water disaster is developed, which realizes the dynamic intelligent early-warning prediction of water disaster risk characteristics and 3D visual display of dangerous areas.The practice has proved that the “double drive” microseismic early warning system using deterministic data research and intelligent model prediction has a remarkable effect on predicting the level and scope of water inrush risk, and has realized accurate early warning and prevention and control of high-risk areas of floor water damage.