Railway Overhead Contact System Point Cloud Classification.

Area of Science:

• Geospatial Engineering
• Railway Infrastructure Monitoring
• Advanced Sensor Technology

Background:

• The overhead contact system (OCS) is critical for high-speed railways, necessitating accurate quality detection.
• Traditional manual OCS detection methods are inefficient, risky, and challenging.
• Vehicle-mounted Light Detection and Ranging (LiDAR) presents a promising alternative for OCS inspection.

Purpose of the Study:

• To develop and validate a novel algorithm for classifying and extracting railway OCS components from LiDAR point clouds.
• To accurately identify contact cables, return current cables, and catenary cables within the OCS.
• To provide a more efficient and precise OCS detection method compared to conventional techniques.

Main Methods:

• Utilized a scale-adaptive feature classification algorithm combined with DBSCAN (density-based spatial clustering of applications with noise).
• Applied the algorithms to vehicle-mounted LiDAR OCS point cloud data.
• Compared the developed method's classification results against TerraSolid for accuracy validation.

Main Results:

• Achieved high average accuracy (99.83% and 99.89%), precision (100% and 99.97%), recall (99.16% and 99.42%), and overall accuracy (99.58% and 99.69%).
• Demonstrated superior performance compared to TerraSolid in OCS point cloud classification.
• Successfully extracted complete OCS components accurately and rapidly from point cloud data.

Conclusions:

• The proposed method provides an accurate and efficient approach for processing railway OCS point clouds.
• This technique offers a valuable new tool for railway component detection with significant engineering application potential.
• Vehicle-mounted LiDAR combined with advanced algorithms enhances OCS inspection quality and safety.