Abstract: In order to ensure the safe and efficient production of coal mines, and forecast the faults and the disa sters timely and accurately, this paper analyzes the current situation and existing problems of coal mines big data in Yankuang Group, and explores the construction and application of big data platform for coal industry monitoring. Massive data is collected, processed and stored by using the internet, cloud computing and big data technology. Through data mining technology and video analysis technology to explore the production rules of the site, the ability of coal mines can be improved in dealing with potential disasters and identifying potential safety hazards. Taking the big data analysis of the coal mine belt conveyor monitoring system and video monitoring system as examples, the data is processed by using the manifold space imbalance processing technology and watershed algorithm. The experimental results show that the manifold space imbalance processing technology can increase the AUC index by about 20%, and the coal flow detection based on machine vision can also achieve better performance. Meanwhile, the combination of big data and coal industry monitoring system is discussed, and the specific solution of alarm data fusion application is proposed.