Abstract: In order to enhance the intelligence and refinement level of operation and maintenance management in thermal power plants, this paper proposes a device status monitoring and fault diagnosis method based on a digital twin platform. This method uses high-precision 3D models as physical carriers, deeply integrates internal mathematical models with multi-source heterogeneous data, and constructs a virtual reality mapping system that runs through the entire life cycle of the power plant. Through the research and implementation of key technologies such as visual query, dynamic presentation, model driven fault diagnosis, and 3D linkage positioning, a complete application loop from data integration, status monitoring to fault diagnosis is formed.