Abstract: Aiming at the commonly existing pain points of data silo effect, insufficient disaster early warning capability and others in coal mine safety monitoring system, a systematic transformation scheme for multi-source perception data access and fusion is implemented at Hecaogou Coal Industry Co., Ltd. in Yan'an City. The core innovation of this scheme includes deploying edge protocol gateways to solve the heterogeneous network communication barriers and difficult protocol conversion problems between professional monitoring systems such as gas, rock burst and others and production systems, and constructing a unified data transmission architecture based on MQTT protocol; Establish a standardized multi-source data preprocessing process, develop a composite disaster risk evaluation model based on fuzzy comprehensive evaluation method, and achieve accurate identification and cross-system linkage control of disaster risks. Significant results are achieved after transformation: the data integrity rate of the system is improved to 99.96%, and the Mean Time Between Failures (MTBF) is increased by 872.7%; The daily average effective transportation volume of the main transportation system is improved by 22.5%, the monthly average number of overload shutdowns is decreased from 4.3 to 0.6, and a composite type of disaster risk event is successfully predicted. The practice proves that this technical path effectively breaks down data barriers, significantly improves the intelligent level and risk prediction foresight of mine safety management, and provides a replicable technical paradigm for the upgrading and transformation of coal mine safety monitoring system.