A Statistical Approach to Guide Phase Swapping for Data-Scarce Low Voltage Networks

Abstract

Phase swapping, which rebalances the unbalanced three-phase low voltage (LV, 415 V) networks, improves network efficiency by reducing capacity waste and energy losses. A key challenge against phase swapping is that the majority of LV networks are data scarce, i.e., there is a general lack of data in LV networks. In light of this, this paper proposes a new statistical approach to develop phase swapping guidance for data-scarce LV networks with neither time-series network measurements nor customer metering data. First, given a set of data-rich LV networks (with time-series phase currents data collected at LV substations throughout a year), typical load profiles and their weights in each of the three phases are extracted by applying a nonnegative matrix factorization method. Then, phase swapping guidance are developed for data-rich LV networks along with their rebalancing potentials (rebalancing potentials refer to the reduction of phase imbalance degree). Second, a rapid screening model is developed to efficiently identify the data-scarce LV networks with high rebalancing potentials. Phase swapping guidance are then developed for these data-scarce networks with high rebalancing potentials. Case studies reveal that the statistical approach produces effective phase swapping guidance, which reduce the phase imbalance degrees for 99% of the LV networks and the maximum reduction is 35%. Validation results show that the average reduction of the phase imbalance degree for data-scarce networks is only 14.3% less than that for data-rich networks.