Abstract: The paper analyzed the nature of hazard and case reasoning of coal mine gas accident,introduced the early warning methods and steps of the ELM model which was an extreme learning machine.Taking the coal mine gas accident as an example,the selection of characteristic variables and characteristic parameters was discussed,and the rationality of the application of ELM model in the case reasoning of accident forewarning was analyzed.Simulation results which use SVM and BP neural network model as reference showed that the accuracy of forewarning based on ELM model is higher than that of BP neural network model and SVM model,the accuracy of ELM model in forewarning is verified.It is feasible to apply ELM model to forewarning of coal mine safety in production.