Abstract: Carbon dioxide emission is one of the main reasons for the greenhouse effect. Oxy-fuel combustion has an extensive research prospect as an effective carbon emission reduction and storage technology.The ignition temperature of pulverized coal in oxy-fuel combustion in coal-fired power plants is an important indicator of burner design and operational safety,which has complex correlations with the composition of coal,coal particle size,and the atmosphere of combustion. Therefore,the research of the ignition temperature prediction model of pulverized coal oxy-fuel combustion is very meaningful. In this study,the ignition temperatures of five coal samples were measured in dropper furnace under 30%,35%,40%,50%,60%,70%,80%,90% and 100% volume fraction of O_2 in CO_2 atmosphere.The relationship between coal ignition temperature and the oxygen concentration and the composition of pulverized coal was analyzed.The research finds that the coal ignition temperature decreases significantly with the increase of oxygen concentration,and the degree of decrease is higher when the coal sample contains more volatile.A machine learning sample base with 45 sets of coal ignition temperature in the experiment and 69 sets of ignition temperature collected from recent year' s research with the same measurement was established.The ultimate analysis and proximate analysis of the pulverized coal,the coal particle size and the oxygen volume fraction were selected as the input features,and the ignition temperature was the target output,a random forest model optimized by genetic algorithm( GA-RF model)was constructed and the ignition temperature of pulverized coal in oxy-fuel combustion was accurately predicted,with the accuracy of R~2>0.99,RMSE<16,MAE<8.The feature importance of ignition temperature shows that the ignition temperature increases immediately when the H content is over 5%,which is proved by the existing ignition data of pulverized coal.