Phase Identification of LV Distribution Network with Smart Meter Data

Distribution networks require more operational flexibility these days than ever before due to growing penetration of low-carbon technologies such as solar energy and CHPs. The LV network topology information is essential for efficient network operation and planning. The customers at LV network are usually single phase loads which connect to either phase a, b, or c. However, these phase connection information that distribution network operators have, are not always available or accurate due to lack of data communication or changing of customer connections. The state-of-the-art phase identification methods are relying on using the smart meter data and main challenges their facing are low penetration/coverage of smart meters and high substation load measurement noises. This paper proposes a novel Least Absolute Shrinkage and Selection Operator (LASSO) based data-driven approach to identify customer phase connection in LV distribution network. The proposed method is validated on a real distribution network - with 228 customers and achieved 97% accuracy with only 60% smart meter coverage with a low cost (class 5, i.e., ±5% error) measuring device for substation load measurement.