Answers are grounded in this paper and cite their sources.
Renewable-dominated power systems need operational digital twins that do more than mirror assets: they must convert streaming evidence into state awareness and secure control actions. This paper presents DT-Grid, a digital twin (DT) framework for real-time state awareness and operational optimization. DT-Grid treats the twin as an evidential control layer with four coupled functions: a topology and data twin, robust temporal state assimilation, confidence-envelope construction, and rolling optimal power flow (OPF). The state-awareness module solves a Huber-weighted, temporally regularized estimation problem and exposes residual information to the optimization layer. The dispatch layer then uses this evidence as a security margin while scheduling conventional generation, renewable acceptance, and corrective actions. We evaluated the framework on the PGLib IEEE 118-bus benchmark driven by Open Power System Data Germany load, wind, and solar profiles over 365 operating points and three measurement seeds. DT-Grid reduced injection root mean square error (RMSE) from 464.6 MW under static weighted least squares (WLS) to 217.0 MW, improved bad-data F1 from 0.140 to 0.149, and reduced the realized overload proxy by 97.9% compared with persistence-driven OPF. The results indicate that state evidence is most valuable when it is carried into dispatch constraints, while pure temporal smoothing can still produce lower phase-angle error in some operating points.
No takes yet. Share an insight, caveat, or question.
Chen et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: