Font Size: a A A

Research And Application Of Rail Longitudinal Displacement Monitoring System Based On AI Image Recognition

Posted on:2022-06-30Degree:MasterType:Thesis
Country:ChinaCandidate:R FanFull Text:PDF
GTID:2492306563463704Subject:Control Engineering
Abstract/Summary:PDF Full Text Request
Transportation industry exerts far-reaching influence on national economic development in China,in which railway transportation dramatically promotes its economic growth.In China,the railway lines are mainly the continuously welded rail track in temperature-controlled stress response,which not only saves large quantity of rail materials and energy consumption,but also,most importantly,reduces the strikes between rails and trains,further prolonging the service time of trains and rails and reducing the cost for rail maintenance.However,it is easy to have track buckling for the rail in summer and broken rail in winter,particularly in northen China with huge temperature difference,which are very dangerous for security of trains and easy to cause accidents.Therefore,it is very necessary to monitor rail displacement.However,since traditional approach for monitoring couldn’t achieve real-time supervision and requires huge amount of people and resources,a new monitoring system for displacement,which could realize real-time monitoring of rail displacement and changes,is designed based on current conditions for the monitoring of the rail displacement.Major research work of this paper is as follows.(1)A positioning algorithm based on ArucoTag is proposed.During positioning,4×1ArucoTag is adopted and the center of the tag is confirmed as the origin of the coordinate for rail substrate.In the algorithm,matrix operation is conducted between the location and pose estimation of ArucoTag under every camera coordinate system,which is recognized during positioning,and the location and pose estimation of rail substrate under ArucoTag coordinate system,which is calculated through mathematical operations.The result of matrix operation will be under mean filtering to get the final location and pose estimation of rail substrate under camera coordinate system through mathematical operations.Offset value of the displacement of the tag in three directions will be obtained by extracting the translation vectors in every location and pose estimation and then conducting difference operation.(2)Displacement monitoring system is successfully designed and applied,which includes hardware for data collection and software for monitoring platform.The hardware for data collection uses STM32 controller and is equipped with EOS(Embedded Operation System)to process tasks efficiently in accordance with pre-set time intervals.Industrial-grade camera is used,which could identify ArucoTag fixed on rails through improved image positioning algorithm to collect data of rail displacement and transmit the data to STM32 controller by placing camera on Raspberry Pi.Other monitoring data of railway lines will be collected through various types of sensors.STM32 controller transmits the collected data to the monitoring platform for supervision through NB-Io T,the wireless transmission module,thus setting up the displacement monitoring platform based on Fast API framework,which consists of map positioning and data visualization based on Baidu Map API.(3)The algorithm is verified and analyzed,the result demonstrates that the algorithm improves the accuracy of test with good stability.the system is tested on railway and the result shows the system is in excellent performance.
Keywords/Search Tags:Rail Displacement, ArucoTag, Data Visualization, Controller
PDF Full Text Request
Related items