Identification of distributed energy resources in low voltage distribution networks

Electrical profiles of EV and PV systems are allocated within aggregated measurements including conventional electrical appliances. Publicly available data and an experimental dataset including either one or several households are used to train and test classification and regression models. The NILM methods proposed here are based on the usage of conventional machine learning techniques such as kNN, RF, SVM, and MLP.

This provides the proposed algorithms with realistic processing times, a key factor needed to differentiate highly variable DER power profiles from other loads and to update real-time conditions of the electrical system to distribution network operators. <br><br>The results achieved confirm the effectiveness of the methodologies proposed to individually identify DER with outstanding performance metrics for both EV and PV electrical profiles.

This demonstrates the potential of the methods proposed to be implemented as an embedded function of smart meters in customer and low voltage distribution sides to increase observability in distribution networks.<br>