To realize real-time monitoring and short-term forecasting and forewarning of coalmine ventilation systems(CVS), in this paper, we first established a joint surface and underground CVS safety management system consisting of main ventilation fan, safety-partition linked passageways, and air-required locations. We then applied chaos theory to identify the air quantity and gas concentration of underground partition boundaries, and adopted a fixed data quantity, multi-step progressive, weighted first-order local-domain method to setup a chaos prediction model and a CVS safety forecasting and forewarning system formed by the normal change level, orange forewarning level, and red alarm level. We next conduct the on-field application of the system in a coalmine in Jining, Shandong, China. The results showed that (1) in the statistical scale of 5 min, the changes in both air quantity and gas concentration along CVS partition airflow boundaries were characteristic of chaos and could be used for short-term chaos prediction, and the latter was more chaotic than the former;(2) the setup chaos prediction model had a higher prediction precision and the established safety prediction system could not only predict the variation in CVS stability but also reflect the rationality of underground mining intensity. Thus, this CVS safety forecasting and forewarning system is of better application value. To realize real-time monitoring and short-term forecasting and forewarning of coalmine ventilation systems (CVS), in this paper, we first established a joint surface and underground CVS safety management system consisting of main ventilation fan, safety-partition linked passageways, and air-required locations. We then applied chaos theory to identify the air quantity and gas concentration of underground partition boundaries, and adopted a fixed data quantity, multi-step progressive, weighted first-order local-domain method to setup a chaos prediction model and a CVS safety forecasting and forewarning system formed by the normal change level, orange forewarning level, and red alarm level. We next conduct the on-field application of the system in a coalmine in Jining, Shandong, China. The results showed that (1 ) in the statistical scale of 5 min, the changes in both air quantity and gas concentration along CVS partition airflow boundaries were characteristic of chaos and could be used for short-term chaos prediction, and the latter chaotic than the former; (2) the setup chaos prediction model had a higher prediction precision and the established safety prediction system could not only predict the variation in CVS stability but also reflect the rationality of Thus, this CVS safety forecasting and forewarning system is a better application value.