Phase Identification of Smart Meters Using a Fourier Series Compression and a Statistical Clustering Algorithm

Accurate labeling of phase connectivity in distribution systems is important for maintenance and operations but is often erroneous or missing. In this paper, we present an algorithm to identify which smart meters must be in the same phase using a hierarchical clustering method on voltage time series data. Instead of working with the time series directly, we apply the Fourier transform to represent time series in their frequency domain, remove 98% of the Fourier coefficients, then cluster the remaining coefficients to estimate which meters belong in the same phase. We validate results by verifying they do not change phase in time and by comparing our results to available network-distribution data.