Graph-Based Meter Placement For State Estimation In Unbalanced Lv Distribution Grids

Advanced control and protection schemes in lowvoltage (LV) distribution networks rely on real-time monitoring applications, such as state estimation algorithms. These algorithms estimate nodal voltages and branch currents using a limited number of measurements, making an optimal placement of smart meters critical for accuracy. This study explores the use of graph centrality metrics to optimize meter placement and enhance unbalanced state estimation. The proposed approach is applied to a real LV network in Gradiˇsˇce, Slovenia, operated by Elektro Ljubljana, and modeled as a graph from its IEC Common Information Model (CIM) representation. Different centrality metrics including PageRank, Current Flow Betweenness Centrality, and Katz Centrality are evaluated. Results show that when fewer than 25-30% of potential measurement points are installed, the error trend has not yet stabilized, and no single metric consistently gives the best performance. However, beyond 30% coverage, PageRank-based placement provides lower and more stable estimation errors.