Smart meters phase identification for topology verification: practical challenges and insights from a case study

For historical reasons, many distribution system operators (DSOs) do not have complete and accurate information about their networks. In particular, the topology of the low-voltage distribution power grid, which is required to carry out numerous analyses necessary to the energy transition, is generally imprecise. This research work presents a novel approach to verify the distribution network topology on the basis of imperfect data. More precisely, the dynamic data retrieved from a partial coverage of smart meters are utilised for that purpose. Relying on the clustering of electrical phases on the basis of the correlation between voltage time series, the algorithmic solution proposed has the advantage of yielding interpretable results, in contrast to black-box models. Promising results are reported for a real test case: an area managed by RESA, a Belgian DSO, with a coverage of smart meters inferior to 20%. Numerous ambiguous situations in the dataset of the DSO have been correctly identified by the algorithm, whose output was perfectly logical from a human’s perspective. To finish with, the interpretability of the results helped to identify ideas to be investigated as future work in order to further improve the algorithmic solution presented.