Modified FasterRCNN-based Detection of Pantograph-Line Arc Through Visualized Monitoring
The stable operation of the high-speed railway pantograph network is of great significance for the power trans-mission to trains.In order to improve the detection accuracy of pantograph arc,the present work made a preliminary at-tempt to design a modified FasterRCNN-based method of identifying pantograph-line arc applicable for visualized monito-ring.The main work entailed the establishment of a FasterRCNN-based model which could extract the arcing area from the pantograph video,and the improvement of image classification and localization performance by combining the VGG16 network and the modified FasterRCNN model.In a simple simulative test,the proposed model achieved 386 true positive detections,5 false detections(including 2 false negative detections),and 12 missed detections,with detection accuracy,precision rate,and recall rate of 99.2%,99.2%,99.5%,respectively,demonstrating the model's potential effectiveness and practicability.
万方数据
维普
