IEEE Transactions on Power Delivery · 2022 · 15 citations · 47 references
EngineeringSystems EngineeringEnergy SystemsReal-time SystemsRenewable Energy SystemsDistributed Energy ResourcesPower Grid AnalysisPower System StabilityPower SystemsSmart Grid SystemsSmart GridsDistribution AutomationDistribution Network ModelingDistribution SystemsElectricity Service RestorationGrid Restoration PlanningOperational Decision SupportPower System ModellingTransmission System Restoration
Situational awareness (SA) is critical to properly operating active power distribution systems during normal and outage conditions. Appropriate SA tools should provide an accurate estimate of system voltage and current variables, the operational network topology, and the power injections from distributed energy sources (DERs), including behind-the-meter (BTM) photovoltaics (PVs). Obtaining an accurate SA, especially by estimating network topology and gross load demand, is increasingly challenging with the proliferation of BTM DERs with intermittent power generation. Moreover, the SA required for distribution system restoration is even more challenging after a medium to a prolonged outage due to inaccurate estimates of cold load pickup (CLPU) and switch statuses. This paper proposes an integrated real-time model update (RTMU) module for SA enhancement to help distribution system operators (DSOs) understand the power system conditions in dynamic and DER-rich environments. The proposed RTMU consists of several modules to obtain the required level of SA for operational decision-making. It includes estimators for a) BTM PV power, b) network topology, and c) CLPU. The proposed approaches leverage multiple data resources, deep learning approaches, and domain knowledge of the power system to provide the required level of SA to DSOs. The dependencies among these modules are actively leveraged to enhance SA under normal conditions and during power outages. We demonstrate and analyze the performance of the proposed RTMU on a modified IEEE 123-node feeder and a utility distribution system from the Western United States.
47
Sepp Hochreiter, Jürgen Schmidhuber · Neural Computation · 1997 · 98.1K citations
Dropout: a simple way to prevent neural networks from overfitting
Nitish Srivastava, Geoffrey E. Hinton, Alex Krizhevsky et al. · 2014 · 34.3K citations