Coal resources are an important strategic resource in China. The safety situation of coal production is quite serious, and the foundation of coal mine risk prediction is weak. In this paper, the identification and concentration detection of CO and CH4 released during coal oxidation or oxidative spontaneous combustion are detected by electronic nose (e-nose) gas sensing technology. The influence of temperature and humidity on gas sensors is studied, and the correlation between the degree of coal oxidation and the output frequency of gas sensors is analyzed. The experimental results show that the sensitivity of the five kinds of e-nose gas sensors to CH4 is almost the same as that of CO, and the sensitivity of 112AJ sensor is the best. The e-nose gas sensors can capture the tiny changes of odor release in the initial stage of coal oxidation, and the artificial neural network analysis method is applied to the gas sensor array, which greatly improves the ability of coal mine risk prediction.