14. Automated Visual Inspection System for Vehicle Underbody
Periodic visual inspections of rolling stock do not require disassembly of the vehicles, but they require manpower due to the short inspection intervals and the reliance on manual checks. Under these circumstances, we developed a system that automatically inspects the bogies and underbody of vehicles, which have many inspection points, for any visible abnormalities.
The system captures images of the underbody of passing vehicles using an imaging device installed at the entrance of the depot, or similar locations (Figures 1 and 2). By using a line-scan camera, the system can obtain high-resolution continuous images. By automatically recognizing the vehicle number displayed on the carbody from the captured images, the system can manage the images by vehicle number without the need to install RFID tags or similar devices on the vehicles. Additionally, the diagnostic algorithm, which suppresses the effects of disturbances such as direct sunlight and rain and evaluates differences from the normal appearance as an anomaly score (Figure 3), enables the detection of unspecified abnormalities regardless of the time of day or weather conditions.
As a result of verification through the imaging of shunting vehicles for up to 21 months, we confirmed that the system can diagnose 17 out of 24 simulated abnormalities (Table 1) with 0% false negatives and less than 1% false positives.
Automating the manual visual inspection with this system contributes to labor savings in rolling stock inspections.
Other Contents
- 9. Design Method for Post-installed Anchor Joint Members in the Reconstruction of Concrete Structures
- 10. Reinforcement Method for Preventing Fatigue Cracks at Steel Girder Support Sections
- 11. Method for Identifying Potentially Critical Locations of Loose Bearing Based on On-board-measured Track Geometry
- 12. Extension of Rail Replacement Cycles Considering Fatigue/Soundness
- 13. Support Method for Extending Inspection Periods Based on Statistical Analysis of Equipment Inspection Records
- 14. Automated Visual Inspection System for Vehicle Underbody
- 15. Autonomous Train Operation System
- 16. General Purpose Real-time Algorithm for Generating Driving Curves for Driver Advisory System
- 17. Method for Updating On-board Databases Using Public Communication Networks
- 18. Labor-saving for Generating Crew Schedule to Enable Workforce Efficiency and Reduce Labor Burden
- 9. Design Method for Post-installed Anchor Joint Members in the Reconstruction of Concrete Structures
- 10. Reinforcement Method for Preventing Fatigue Cracks at Steel Girder Support Sections
- 11. Method for Identifying Potentially Critical Locations of Loose Bearing Based on On-board-measured Track Geometry
- 12. Extension of Rail Replacement Cycles Considering Fatigue/Soundness
- 13. Support Method for Extending Inspection Periods Based on Statistical Analysis of Equipment Inspection Records
- 14. Automated Visual Inspection System for Vehicle Underbody
- 15. Autonomous Train Operation System
- 16. General Purpose Real-time Algorithm for Generating Driving Curves for Driver Advisory System
- 17. Method for Updating On-board Databases Using Public Communication Networks
- 18. Labor-saving for Generating Crew Schedule to Enable Workforce Efficiency and Reduce Labor Burden
