The abnormality of the mine ventilation system can reflect the risks and hidden dangers existing in the mine production system. By combining the basic information of the demonstration mine with the ventilation monitoring data and production status, the k-nearest neighbor is used to study the abnormal change characteristics of monitoring data of mine ventilation system in different ventilation periods, and a ventilation hazard warning model was constructed and the models were compared and validated. In addition, the dominant factors of the warning level were obtained by combining the gray correlation method. The results show that under different ventilation periods, the accuracy of the warning model is over 95%, which has good application and promotion value. The speed is the main indicator that affects the warning level. The research of this paper can provide theoretical support for realizing the intelligent management of mine risk in advance and short-term early warning.

coal mine ventilation; disaster warning; k-nearest neighbor; grey relational analysis; intelligent management