• Apply digital twins to the scenario of power grid data centers. • Propose an integrated modeling and maintenance framework based on digital twins. • Propose four types of models in virtual entities and introduce the Function Digital Twin (FDT). • Apply the maintenance method with anomaly detection and Root Cause Analysis (RCA) in the real scenario. The increasing complexity of the power grid data center (PGDC) operation and maintenance system complicates system modeling and maintenance, hindering system application and business operations. This paper presents an integrated modeling and maintenance framework based on digital twin. The modeling approach redefines four types of models in virtual entities and introduces the function digital twin (FDT) to describe the constituent elements of physical entities. The maintenance method extracts FDT behaviors through data analysis, formulating rules based on the modeling approach to achieve anomaly detection and analyzing anomaly associations for root cause analysis (RCA). The proposed framework has been applied to a real PGDC scenario, demonstrating its authenticity and practicality.
Xiao et al. (Wed,) studied this question.