Identification of Distributed Energy Resources in Low Voltage Distribution Networks

The increase of distributed energy resources, namely rooftop photovoltaic (PV) systems and electric vehicles (EV), has brought new technical challenges related to the operation and planning in low-voltage distribution networks. The modification of electrical network dynamics has exposed a lack of observability in this side of the electric system requiring innovative techniques to increase flexibility, reliability, and security of supply in power networks. A supervised nonintrusive load monitoring method commonly used for the identification of traditional electrical appliances connected behind-the-meter in household distribution boards is proposed in this research study. The work is based on a low-complexity, effective machine learning algorithm to identify the presence of PV generation or EV power consumption in aggregated measurements of low-voltage networks. The model is developed in the IEEE European low voltage test feeder. It evaluates several scenarios using data with 1-minute sampling simulated with the tool OpenDSS and with k-nearest neighbour as classification algorithm. The results of the proposed method exhibit high performance for both PV and EV identification and illustrate the potential for these non-intrusive monitoring solutions in supporting the integration of distributed resources in low-voltage power networks.