Online wheel tread defect detection for high-speed trains using bayesian approach
To ensure safety operation and cost-effective maintenance for high-speed train,it is necessary to detect wheel defects soon after their occurrences and then conduct wheel reprofiling for the defective wheelsets.Online wheel tread defect detection by a trackside wheel condition monitoring system can be a promising solution for this problem.This paper presents a fibre Bragg grating(FBG)-based wheel condition monitoring system for wheel tread defects with small radius deviation which are frequently reported in high-speed rail.The trackside monitoring system consists of two FBG strain gauge arrays mounted on the rail foot to measure the dynamic strains of the rails excited by passing wheels.A defect detection algorithm based on Bayesian learning method is then developed for using the online monitoring rail response data to identify the both the potential wheel local defects.A proximity measure technique is utilized to remove the datasets with outliers,which leads to a more appropriate posterior Probability density function(PDF)of monitoring data estimated by Bayesian method.By adopting Chauvenets criterion,the localised anomalies on rail response under the excitation of defective wheel can be identified from the PDF of monitoring data.The proposed method for online wheel tread defect detection is verified by an in-situ test,in which the devised system was installed on a test track and a test train with defective wheels ran through the instrumented zone.It is seen from the test that all the wheel defects were identified and the online detection results agreed well with those from the offline wheel roughness measurement in the depot.
wheel condition monitoring wheel tread defect trackside sensing system FBG strain gauge arrays Bayesian learning method
Xiao-Zhou Liu Yi-Qing Ni
Chinese National Engineering Research Centre(CNERC)on Rail Transit Electrification and Automation(Hong Kong Branch),The Hong Kong Polytechnic University,Hung Hom,Kowloon,Hong Kong
国际会议
The 7th World Conference on Structural Control and Monitoring(7WCSCM)(第七届结构控制与监测世界大会)
青岛
英文
1336-1345
2018-07-22（万方平台首次上网日期，不代表论文的发表时间）