Phase Identification with Single-Phase Meter and Concentrator Based on NMF Dimension Reduction and Label Propagation

The accurate phase information of single-phase meter in low-voltage distribution network is important for refined management. For a long time, China's low-voltage distribution network has been expanding rapidly, but the management is insufficient, and the maintenance of topology files including meter phase information is not timely, which leads to meter phase information errors. At present, it mainly relies on manual inspection and installation of carrier meter for phase identification. The former is time-consuming and inefficient, while the latter is expensive. This paper presents a high accuracy phase identification method of single-phase meter in low-voltage distribution networks based on time sequence voltage data of single-phase meter and concentrator. Firstly, the single-phase meter voltage data and concentrator three-phase voltage data matrix are formed, and the KNN method is used to fill the missing data. The non-negative matrix factorization (NMF) dimension reduction method is used to transform the high-dimensional voltage time series data into low dimensional data, which aim to solve the problem of data redundancy and high computational cost caused by high dimension. Then the label propagation algorithm based on graph theory is used to process the reduced data iteratively, which lead to identify the phase of each single-phase meter. Finally, the identification accuracy of the method proposed in this paper reaches 100
^%
on the data of a distribution network in Nanjing. It is also compared with other similar work to verify the accuracy and superiority of the method.