Data Anomaly Traceability Strategy for Low-Voltage Distribution Network Based on Time Series-Causality Learning

Abnormal electricity consumption behavior of users will cause huge economic losses to the distribution network. Many distributed renewable energy sources are integrated into the low-voltage distribution network (LVDN), which makes it difficult to monitor abnormal power consumption. The anomaly traceability strategy aims to find the end user through the analysis of abnormal power consumption data. For LVDNs with incomplete topology and line parameters, this paper proposes a traceability analysis method for abnormal power consumption data based on time series causality learning. Firstly, by mining historical user electricity data, a time series causality model based on Bayesian network is established. Then, based on comparative learning, users with abnormal electricity consumption behavior can be accurately identified. Finally, the effectiveness of the proposed method is verified by actual data.