With the advancement of digital construction, the number of information devices in each area of the power grid room is growing, posing management challenges related to security and efficiency. When combined with the Digital Twin (DT) technology, intelligent scheduling and monitoring of equipment operating status based on the DT room is possible. In this paper, the network performance metrics data of the physical layer of the information room is mapped to the DT layer. The data from DT layer contains rich informatio related to the physical network layer. Furthermore, a spatio-temporal model based on graph convolutional neural network and long short-term memory is proposed to analyze the network performance metrics data, such as network traffic mapped to DT layer. Then, by means of the Blockchain for its characteristics of security, traceability, and non-tampering, the predicted results can be securely shared among consensus nodes that join in the private chain network inside the computer room, providing data support for computer room managers to analyse the status of networks and optimize them.