Abstract: The accurate identification and early warning of coal seam spontaneous combustion has always been a worldwide problem that urgently needs to be solved in the field of coal spontaneous combustion prevention and control. It is necessary to build a fine division of coal spontaneous combustion stage and accurate early warning methods. Through the establishment of a tiered early warning system for coal spontaneous combustion, the early and accurate prediction and active prevention and control of coal spontaneous combustion can be realized, and the occurrence of spontaneous combustion or secondary disasters can be effectively prevented. Fresh coal samples were collected from the working face on site, crushed and mixed into test coal samples, and a large-scale spontaneous combustion test bed was used to simulate the actual oxidation and heating environment of the coal seam in the working face, and the coal spontaneous combustion test was carried out. Through further analysis of the collected data, the change rule of the index gas product volume fraction and generation rate of each reaction stage in the process of natural oxidation and heating of coal seam with coal temperature was obtained, and the characteristic temperature points of each stage of coal spontaneous combustion reaction were determined. By comparing and analyzing the experimental data, the index gases of coal-oxygen reaction process (CO,O2, C2H4, C2H6, φ(CO2)/φ(CO), Graham, chain alkyl ratio)of the fine division of the coal bed oxidation heating process was determined, and the relationship between thresholds and coal characteristic temperature was obtained. The whole process of coal spontaneous combustion is finely divided into six early warning levels, and a relatively complete warning system for coal spontaneous combustion has been established, and the judgment standards at each stage have been quantitatively determined. Combined with the benefit performance of the hierarchical early warning system in the application of the field working face, it verifies the feasibility of the constructed early warning method for predicting and forecasting the spontaneous combustion process of coal seams in coal mines and adopting corresponding prevention and control measures to suppress the development of coal seam fires. The research results have important theoretical research value and practical guiding significance for the prediction and active prevention and control of coal spontaneous combustion in coal mines, which can further improve the safety of the mines, and guarantee the safety of the personnel working underground and realize the green mining of coal resources.