Edge-Computing Based Dynamic Anomaly Detection for Transmission Lines

Dynamic anomalies often unpredictably intrude into the transmission corridor, threatening secure operation of the grid. Due to the wide geographic spread of the transmission system, it is often difficult to monitor and respond quickly. Automated dynamic anomaly detection algorithms are needed to promptly detect and respond to these ongoing threats. Challenges include identifying warning zones, anomalies, and the anomaly's motion status. In this work, we use images taken by the surveillance camera on the tower to determine the transmission corridor and warning zones. The anomalies inside the warning zone are identified by an object detection model and tracked using an Intersection Over Union (IOU)-based tracking algorithm. In the event of a threat, an alarm will be generated. Extensive testing with real-world data demonstrates the effectiveness of the proposed framework and its potential to scale through deployment on low-cost edge devices.