Phase Identification of Smart Meters by Spectral Clustering

Abstract

Phase identification is important for the safe operation of distributed power grid, and a data-driven phase identification approach has been proposed in this paper, in which the OpenDSS (Open distribution system simulator) is used to perform power flow analysis on a real distribution feeder-European Low Voltage Test Feeder. The spectral clustering is used to identify phases of users/smart meters. Verification by European Low Voltage Test Feeder in different settings. The phase identification accuracy of proposed method is over 95% when the voltage sampling period is no more than 1 hour in the case of complete data set. Results indicate that the proposed method performs better than the constrained k-means method when the sampling period exceeds 15 minutes.