It is important for the early prediction and prevention of coal fire to study the variation law of indicator gas in the process of coal spontaneous combustion.Based on this, the temperature programmed rise experiment of coal samples is carried out to analyze the changing trend of oxidation products, alkane gas, olefin gas and the ratio of each indicator gas with temperature.Combined with the mathematical method of grey correlation analysis, the specific values of the correlation degree between the 11 index gas concentrations or the ratio of the index gas concentrations and the coal temperature are obtained, and the matching degree of each index and the temperature change trend is judged according to the numerical value order, and then the top six indicators suitable for predicting the coal spontaneous combustion at five different temperature stages are selected, and the top three prediction indicators are constructed coal spontaneous combustion prediction system supplemented by the last three prediction indicators.Among them, C₂H₆, C₂H₆/C₃H₈and C₃H₈are the top three prediction indicators with temperature of > 70-110 ℃, and the grey correlation degree is 0.810 4, 0.807 5 and 0.763 8 respectively.It can be seen that the prediction index system is mainly based on alkane gas, alkane ratio and alkene ratio, which has a stronger correlation with the trend of temperature change.The research results provide certain support for enriching and improving the prediction index system of coal spontaneous combustion and more reasonable prediction of coal spontaneous combustion process.