Abstract: Aiming at the problems of frequent faults and lagged maintenance of coal mine electromechanical equipment, an intelligent monitoring and diagnostic system based on digital twin is designed and constructed. The system adopts a four-layer architecture of "entity layer-data layer-model layer-application layer", integrates multi-source sensors, virtual modeling, and CNN algorithm to achieve dynamic perception, trend prediction, and active early warning of equipment operation status. The system conducts a pilot deployment on equipment such as roadheaders, belt conveyors, main ventilation fans and others in Binhu Coal Mine of Zaozhuang Mining Industry. After 3 months of trial operation, the fault response time is shortened from 3.5 hours to 0.8 hours, the recognition accuracy rate is improved from 86.4% to 96.2%, and the number of unplanned shutdowns is decreased by 66.7%. The innovation of this research lies in proposing a digital twin system architecture for coal mine electromechanical equipment, integrating and optimizing a CNN intelligent diagnostic model that adapts to complex working conditions, and the engineering application of the system is achieved under the mine environment of multi-equipment, the feasibility and promotional value of the technology are verified.