# Low Voltage Customer Phase Identification Methods Based on Smart Meter Data

**Authors:** Alexander Hoogsteyn, Marta Vanin, Arpan Koirala, Dirk Van Hertem  
**arXiv:** [2204.06372](https://arxiv.org/abs/2204.06372)  
**Date:** 2022-04-13  
**Summary page:** https://www.summarizepaper.com/en/paper/2204.06372v1/

## Key points

- Increased deployment of distributed energy generation and integration of new electric loads pose challenges to low voltage distribution systems
- Active management solutions have been proposed in the literature
- Distribution system models need to include phase connectivity of all consumers
- Information on phase connectivity is often unavailable in practice
- Voltage and power measurement-based phase identification methods are implemented and compared
- Voltage data generally yields better results compared to power data from smart meters
- Ensemble learning method can improve accuracy when power data is available alongside voltage data
- Underground cables and lack of color-coding pose additional challenges for phase identification
- Phasor measurement units (PMUs) are expensive and unlikely to be widely available at low voltage levels in the near future
- Smart meters offer a more realistic option for practical implementation
- Mathematical formulation of phase identification methods is presented in Section III
- Performance of different methods under various scenarios is illustrated in Section IV
- Conclusions drawn based on the findings contribute to addressing challenges related to accurate phase identification in low voltage distribution systems

## Summary

The increased deployment of distributed energy generation and the integration of new, large electric loads such as electric vehicles and heat pumps pose challenges to the correct and reliable operation of low voltage distribution systems. In order to address potential problems, active management solutions have been proposed in the literature, which require distribution system models that include the phase connectivity of all consumers in the network. However, in practice, information on the phase connectivity is often unavailable. In this study, several voltage and power measurement-based phase identification methods from existing literature are implemented and compared. The comparison is conducted across different smart meter accuracy classes and smart meter penetration levels using publicly available data. Additionally, a novel method is proposed that utilizes ensemble learning and can combine data from different measurement campaigns. The results indicate that voltage data generally yields better results compared to power data from smart meters of the same accuracy class. Furthermore, if power data is also available, the novel ensemble method can improve the accuracy of phase identification obtained from voltage data alone. The paper highlights that underground cables and lack of color-coding pose additional challenges for phase identification in low voltage distribution systems. Various data-driven approaches have been proposed in literature, including field measurement-based methods using phasor measurement units (PMUs) between consumers and transformers. However, PMUs are expensive and unlikely to be widely available at low voltage levels in the near future. Smart meters offer a more realistic option for practical implementation. Section III presents the mathematical formulation of the phase identification methods used in this study, including the novel ensemble learning method. Section IV illustrates the performance of different methods under various scenarios. Finally, conclusions are drawn based on the findings. Overall, this study contributes to addressing challenges related to accurate phase identification in low voltage distribution systems by implementing and comparing existing methods based on smart meter data.