A foreign object detection method for high-temperature superconducting maglev train’s track based on YOLOv5

Abstract High temperature superconductor (HTS), by its flux-pinning property and zero resistance characteristics, is widely used in power and magnetic levitation systems. While during the operation procedure of HTS maglev, random appearance of foreign objects on the maglev track, will occupy a key cause of track irregularities and degrade the system’s stability and safety. In order to solve this issue, a foreign object detection system is designed, which comprises unmanned aerial vehicle and the You Only Look Once (YOLO) v5s model. The model is trained to fly the drone along the maglev track and collect image data. The visual model is used for recognition and annotation to detect foreign objects on the track. Drones cruise at a fixed altitude above the track, and high-definition cameras capture video data. The YOLOv5s visual algorithm is used to identify foreign objects attached to the magnetic track, accurately distinguishing between common magnetic and non-magnetic objects near the track. Based on this technology, there is no need to manually inspect the track, reducing the risk of foreign object intrusion caused by human factors and improving the safety and stable efficiency of train operation. This study is expected to provide a foundation for the subsequent operation of HTS high-speed lines and the detection of foreign objects on maglev train tracks.