Abstract: In view of the requirements of intelligent safety monitoring system for multi-system underground integration and “man-machine-environment” emergency linkage in Guide to Intelligent Mine Construction （version 2021） issued by the National Energy Administration and National Mine Safety Supervision Bureau in 2021, the common characteristics of equipment at the sensing layer of different systems and the problems of irregular false alarm and large numbers of real-time collected data are analyzed. At the same time, from the perspective of edge computing equipment, combined with the requirements of the development trend of intelligent mine for the real-time and nondrop-frame loss of underground multi-system fusion data acquisition and processing, this paper deeply studies three key technologies of multi-system sensor layer equipment fusion, such as unified acquisition of sensor layer equipment of safety monitoring system, nondrop-frame of collected data, and abnormal data detection of sensor layer equipment. The data acquisition middleware method is used to establish a unified acquisition architecture, the double buffer queue is used to realize multi-threaded data acquisition without losing frames, and the sensor layer data anomaly real-time detection algorithm with time series as the calculation factor is proposed, and the detailed implementation method is given.