Reduced Voltage-Dependency by Categorical Location Information and Distance Along Street Metric for Meter-Transformer Mapping in Distribution Systems

Deep penetration of distributed energy resources (DERs) and electric vehicles (EVs) introduce benefits but may cause the overloading of service transformers in distribution networks. Such situations require real-time transformer loading information, where an accurate mapping between smart meters and distribution transformers is a prerequisite, e.g., summing the downstream smart meter consumptions. Due to arbitrary curvature in streets, we propose to employ a density-based clustering based on voltage magnitudes (continuous) and street name (categorical) information. However, a density-based approach may only be able to localize a meter to a street segment. Hence, we use a second-stage spectral clustering with distance along the street (DAS), a novel feature, to obtain meter clusters, each with a common parent transformer. For mapping transformers to meter clusters, we use the nearest cluster center approach based on the location since voltage measurements may not be available at transformers. Moreover, we provide a theoretical guarantee for such an approach. Finally, we illustrate the usefulness of the proposed algorithm on long streets, which is a challenging scenario due to many possible incorrect combinations of meter-transformer mapping. The proposed algorithm has been tested on modified IEEE 8-,
$6{9}$
-, 123-bus test systems and real distribution feeders from a utility in the Southwestern United States, demonstrating outstanding performance.