Research on Abnormal Electricity Detection and Diagnosis Model

The abnormal diagnosis report of electricity metering is a prerequisite for on-site inspection and a very important part of the abnormal electricity consumption diagnosis process. This article proposes a measurement anomaly detection and diagnosis algorithm for smart grids based on this requirement. Through in-depth research on the current status and demand of measurement anomaly diagnosis, based on the Faster R-CNN network structure, a model for detecting and diagnosing abnormal electricity consumption is studied. The model includes three functions: abnormal day detection, abnormal data detection, and abnormal time interval detection, thus achieving complete abnormal diagnosis and generation. Through experimental verification, the model performs well in abnormal behavior judgment and abnormal day detection models.