Abstract: For the purpose of the dynamic and real-time early warning of abnormal gas emission, the probability distribution of the time series of gas concentration at the working face was analyzed. The joint normal test of Shapiro-Wilk and Lilliefors was used to deeply excavate the distribution characteristics of time series of gas concentration at the working face. Taking a driving face in Pansan coal mine as an example, a real-time normal test of the time series of gas concentration during fault crossing was carried out. The research results show that when the factors influencing gas emission are similar in effect and none of them plays a decisive role, the time series of gas concentration is normally distributed, that when disobeying the normal distribution, the fault has a significant influence on the gas emission at the working face, which may lead to disasters, that through the real-time normal test of the time series of gas concentration in the working face, the gas emission state can be identified, and that the distribution characteristics of the time series of gas concentration can be taken as the basis of the early warning, which can play a helpful role in the prediction and early warning of gas disasters.

Key words: gas concentration, time series, normal distribution, hypothesis test, identification of abnormality, disaster warning