Data-driven Approaches for Anomaly Detection in Low-Voltage Grid Net Power

This paper examines the anomaly in low-voltage (LV) grids, which are becoming more complex due to the widespread adoption of distributed energy sources such as electric vehicles, electrical heat pumps, and increasing photovoltaic installations. Various data-driven techniques, from statistical analysis to artificial intelligence, are used to analyze data from the grid. The focus of the research is on detecting unexpected changes in energy usage, unauthorized photovoltaic installations, and sudden changes in grid loads, to identify significant deviations from expected patterns. The results show that the suitability of each method depends on the specific objectives of grid operators and effectively differentiates between short-term and long-term anomalies. The anomaly detection capability improves network operators’ understanding of current network dynamics and enables them to better anticipate and manage emerging patterns in the future.