Abstract: In order to avoid the phenomenon of "information isolated island" in the early warning and management of coal mine disasters in China, and comprehensively improve the digital level of disaster parameter monitoring and the intelligent level of data analysis, taking Cuncaota No.2 Coal Mine of China Energy Shendong Coal Group Co., Ltd.( hereinafter referred to as "Cuncaota No.2 Coal Mine") as the research object, combined with big data analysis and statistical technology, the data collected by monitoring are analyzed and trained in depth, and the mine production and disaster management work are comprehensively considered. A mine disaster intelligent management and early warning system composed of edge data resource layer, cloud data processing layer and application terminal layer is constructed. The intelligent closed-loop management of mine disaster perception - analysis - decision - execution is realized. The field application in No.2 Coal Mine shows that the accuracy of mine disaster early warning is significantly improved, the number of disaster events and the average downtime maintenance time of working face are reduced, and the intelligent level of mine safety management is improved by replacing traditional safety inspection with automatic monitoring.