Abstract: The operation state of underground water accumulation and mined-out areas in coal mines is the basic condition for underground efficient production work, while the water-blocking coal pillar, as the object directly exposed to water, is a potential hazard factor. In order to understand the internal force and seepage pressure changes in the water-blocking coal pillar in time, we take the water-blocking coal pillar between lane 32 and lane 33 of No.42108 auxiliary transport roadway in Inner Mongolia as the research object, design and set up a monitoring system to collect real-time data; we introduce relevant theories and inequality methods from probability theory into sensor classification and anomaly data determination, establish a single sensor graded alarm method, which can accurately discover anomalies and improve the accuracy of data anomaly screening; we adopt matrix QR decomposition method and least square method, and simultaneously conduct comprehensive analysis on borehole stress and seepage pressure monitoring data, establish interval ranges, and provide a comprehensive safety classification method based on borehole stress and seepage pressure monitoring. The research results show that the single-point graded early warning and comprehensive analysis early warning technology methods can accurately evaluate and early warn the fixed point area and overall safety of the water-blocking coal pillar; at the same time, the single-point graded results are generally slightly higher than the comprehensive analysis early warning level of the water-blocking coal pillar, indicating that risk in a single point does not determine the overall safety status of the coal pillar, and the water-blocking coal pillar still has a high reliable operating characteristics.