Real-time coordination of distributed energy resources (DERs) is crucial for regulating the voltage profile in distribution grids. By capitalizing on a scalable neural network (NN) architecture, machine learning tools can attain decentralized DER decisions by minimizing the average loss of prediction. This paper aims to improve these learning-enabled approaches by accounting for the potential risks associated with reactive power prediction and voltage deviation. Specifically, we advocate to measure such risks using the conditional value-at-risk (CVaR) loss based on the worst-case samples only. Additionally, we propose a mini-batch selection algorithm based on the CVaR value to accelerate the training process. Numerical tests using real-world data on the IEEE 123-bus test case have demonstrated the computation and safety improvements of the proposed risk-aware learning algorithm for decentralized DER decision making in distribution systems.
PecanStreet Dataport: https://www.pecanstreet.org/dataport/
PSCC2022 https://arxiv.org/pdf/2110.01490.pdf