Research On Onboard Track Comprehensive Inspection System for Electric Passenger Train
In response to the limitations of track inspection cars occupying"window"resources,low operation frequency,and inability to perform online detection,an onboard track comprehensive inspection system for electric passenger trains has been developed.This system enables online detection and the detection results are analyzed and filtered through artificial intelligence and recognition algorithm technologies,promptly identifying out-of-limit defects and subsequently guiding reasonable maintenance to ensure the safe operation of trains.The system is primarily composed of a track geometric parameter inspection system,a rail profile inspection system,and a track status patrol inspection system.The track geometric parameter inspection system utilizes the inertial trajectory method for real-time collection and analysis of track geometric parameters,achieving a reproducibility deviation of less than 0.5 mm for each parameter.The rail profile inspection system is based on the PLICP(Point-to-Line Iterative Closest Point)method,enabling wear detection with a reproducibility deviation of less than 0.2 mm at 95％confidence.The track status patrol inspection system employs deep learning methods based on convolutional neural networks to analyze damage to rails,fasteners,sleepers,and the track bed,achieving complete coverage and high-definition imaging of inspection objects,with a minimum defect recognition size meeting the detection requirement of 5.0 mm×5.0 mm.The results of verification tests for repeatability,accuracy,and other parameters indicate that the onboard track comprehensive inspection system for electric passenger trains can meet the precision requirements of various inspection indicators.
onboard systemtrack comprehensive inspectiontrack geometric parametersrail profiletrack status
NETL
NSTL
万方数据
