Data-Driven Topology Estimation with Limited Sensors in Radial Distribution Feeders

Abstract

Topology estimation is a central part of the wider state estimation problem in electrical networks. We describe a method for data-driven topology estimation in radial distribution feeders with limited sensors. Our algorithm, based on voltage event correlation, estimates a fixed, unknown topology using voltage magnitude measurements collected over several hours and stored in the high performance time-series Berkeley Tree Database. In addition to a topology estimate, our correlation-based algorithm returns a short, human-interpretable snapshot of measurement data that validates the topology estimate. We test our correlation based algorithm on microsynchrophasor data collected on an operational distribution feeder.