Abstract: To grasp the overall safety state of coal mine in real time, this paper introduces the concept of safety situation awareness into the field of coal mine safety, and constructs a coal mine safety situation awareness system which integrates intelligent perception, safety situation assessment and risk intelligent early warning. The system collects and processes data comprehensively and effectively through the Internet of Things monitoring, analyzes the national coal mine accident records and extracts the impact factors of coal mine risk, extracts the chain of accident causes through Bayesian network analysis, and establishes the index system of safety situation multilevel prediction, constructs the multilevel prediction model of coal mine risk change trend by the combined application of Bayesian Network, Rough Set Theory and Support Vector Machine; finally, the system presents the results of overall safety situation assessment of coal mine to users through visual charts. The system provides intuitive and effective prediction results for the overall safety situation of coal mines, and provides ideas for the perception study of coal mine safety situation.