Topology Identification of Radial Distribution Networks Using Smart Meter Data

This article proposes a novel approach for identifying the topology of radial distribution networks. It uses smart meter measurements from loads (e.g., houses) of the network to estimate the voltage sensitivity coefficients with respect to changes in the loads currents. These coefficients are then used to reconstruct the network topology. An enhanced graph learning algorithm with backtracking is proposed to generate a cluster of graphs, from which the best fitting candidate is selected using a set of optimization criteria. The performance of the proposed method is first assessed using the European Low-Voltage IEEE test case, followed by an extensive performance evaluation in noisy conditions on randomly generated topologies. Finally, measurements from a real low-voltage network in Australia are used to validate the method.