Hierarchical Clustering Method for Phase Identification in Low Voltage Distribution Networks Based on Voltage Monitoring Data

Acquiring the phase connection information of low-voltage distribution networks (LVDNs) is of paramount importance for power utility to achieve fine-grained operation of LVDNs. This paper presents a hierarchical clustering-based approach for phase identification in LVDNs, leveraging the voltage monitoring data from the advanced measurement infrastructures (AMIs). Firstly, principal component analysis (PCA) is employed to extract salient features from monitoring data. Secondly, a hierarchical clustering algorithm is applied to group the customers into various phase connectivity clusters based on salient features. Finally, the algorithm is applied on a real LVDN in Guangdong province, China, demonstrating an accuracy exceeding 90% for the phase identification task.