Real-time perception of distributed energy resource access status based on an event-driven architecture

Abstract

High-penetration distributed energy resources (DERs) require distribution operators to identify access-state transitions quickly enough to support automatic voltage regulation, curtailment, reconnection holding, and isolation, yet periodic telemetry creates redundant traffic and masks stale states. This study aims to provide low-latency, control-ready perception for asynchronous DER access data. An event-driven architecture is developed in which field gateways use volatility-adaptive triggering and critical-rule overrides to publish informative measurements. A central latest-value cache records event age and source quality, and an event-age-aware graph attention-gated recurrent model estimates network states, classifies five access modes, and screens outputs using a power-flow residual. Confidence-gated decisions are then translated into reactive-power support, soft curtailment, reconnection hold, or isolation commands. In an OpenDSS-Python co-simulation on a modified IEEE 33-bus feeder with twelve heterogeneous DER terminals, the method reduced traffic by 72.8%, achieved 0.0046 p.u. voltage RMSE, 0.946 macro-F1, and 48 ms P95 detection latency. Under 5% packet loss and 120 ms delay, macro-F1 remained 0.938 and voltage RMSE 0.0053 p.u.; closed-loop recovery was shortened to 0.68 s with 0.53 kWh curtailed energy. The proposed framework provides scalable perception-to-action support for automated DER access control.