| I&T Solution |
Intelligent Railway Track Crack Detection System Based on SWIR Detector and High-Speed Camera
(REF: S-1702) |
| Trial Project |
|
| Solution Feature |
- Utilizing SWIR (Short-Wave Infrared) detection technology enables the identification of limited and undetectable details that are difficult or impossible to perceive using visible light, making it suitable for internal structure inspection.
- Utilizing a high-definition and ultra-high-speed camera capable of capturing images at a rate of 1,594 frames per second with 510,000 pixels, the necessary images are taken from the front, outer, and inner sides of the track.
- Utilizing a dual-channel sampling approach that combines a SWIR (Short-Wave Infrared) detector and an ultra-high-speed camera, along with a multi-parameter recognition algorithm, enhances the system's accuracy and stability.
- Maintain clock synchronization with the railway system, issue timestamps and alarm content upon detecting an abnormality, and utilize those timestamps to retrieve the location of the train within the railway system.
- Utilizing edge computing to conduct real-time, on-site computational analysis of collected data, combined with cloud services, allows the reporting of alarm data to the cloud, thus reducing the burden of data transmission.
|
| Trial Application and Expected Outcome |
- Preparing the experimental setup involves key components such as a high-speed camera and a SWIR (Short-Wave Infrared) detector. A simulated scenario is then constructed, involving a moving vehicle traveling at 80 kilometers per hour alongside a steel plate
- In the constructed simulated scenario, data is collected and then divided into training, validation, and testing sets. The training set data undergoes preprocessing and labeling for subsequent analysis.
- Selecting an appropriate algorithm for validation, with independent computations from the dual channels, the system integrates to obtain an identification score for railway cracks. Adjustable parameters allow for adaptation to different sensitivity requirements
- Deploying the experimental setup and edge computing device in a real-world environment, we evaluate the accuracy of railway crack detection and the precision of location reporting. Based on the evaluation, the solution is optimized and adjusted accordingly
- Improving the solution, we refine the system architecture and supplement it with cloud services. Following these enhancements, we conduct a pilot run in a real-world environment
|
| Additional Solution Information |
智能鋼軌裂縫偵測方案-Robocore 0502.pdf
|
| Info on I&T Solution Provider |
| Solution Provider | : | ROBOCORE TECHNOLOGY LIMITED | | Address | : | 香港沙田科學園 科技大道西 19W大樓 211室 | | Contact Person | : | Zhou Lijuan |
| Position | : | Project Manager | | Tel | : | 94266627 | | Email | : |
lucy@robocore.ai |
|
