A fastener integrity detection method for unmanned aerial vehicle track fine inspection

By using a vision-guided standard representation method for the overall layout of rail fasteners and the HRFDet fastener inspection network architecture, the occlusion and shadow problems in fastener inspection during UAV railway inspection were solved, achieving uniform distribution and fine inspection of fasteners, and improving the integrity and accuracy of the inspection.

CN119693625BActive Publication Date: 2026-07-24BEIJING UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
BEIJING UNIV OF TECH
Filing Date
2024-12-17
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

In traditional UAV railway inspection, the rail fastener detection algorithm cannot take into account the inherent geometric relationship between fasteners, resulting in fasteners being missed due to obstruction or shadows. This cannot guarantee the completeness and accuracy of the inspection and poses a safety hazard.

Method used

We employ the visual prior-guided standard characterization method SVP-RFR for the overall layout of rail fasteners and the end-to-end fastener detection network architecture HRFDet. By designing adaptive anchor frames and adaptive loss functions, we achieve overall layout and fine detection of fasteners, reducing the problem of missed detection caused by occlusion and shadows.

Benefits of technology

It achieves uniform distribution and precise inspection of rail fasteners, reduces the impact of occlusion and shadow, provides inspection coverage and integrity indicators, provides statistical data reference for drone inspection, and improves the integrity and accuracy of inspection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119693625B_ABST
    Figure CN119693625B_ABST
Patent Text Reader

Abstract

The application provides a fastener integrity detection method for unmanned aerial vehicle track fine inspection. The health condition of the rail fastener directly affects the stability of the rail position. The problem of missing detection of the fastener caused by obstruction or shadow inevitably exists, and the specific position and statistical data of the missing detection cannot be given. The application is established on a designed special standard embedding space, has good geometric interpretability, can be closest to all rail fastener wholes in the image, and makes all fasteners uniformly distributed on both sides of the rail in the rail direction. By using the fastener layout representation method, an end-to-end heuristic knowledge guided fastener whole detection network architecture is proposed, which can ensure the integrity of the fastener detection and realize the perfect detection of the rail fastener in the unmanned aerial vehicle aerial image of the track area from coarse to fine, and maximally reduce the missing detection problem of the fastener caused by the thick shadow or obstruction of the trees, contact network rods or rails.
Need to check novelty before this filing date? Find Prior Art