Research On Wheelset Abrasion Detection Method Based On Monocular Vision

With the heavy load of railway freight transportation and the rapid development of passenger transportation in China,the wheelset wear of the main components of the locomotive running part is becoming more and more serious.Excessive wheelset wear directly affects the smooth running and curve passing ability of the vehicle,which seriously threatens the safety of train driving,so it is especially important to detect wheel wear of the vehicle in time and quickly.Nowadays,domestic wear measurement is mainly contact measurement.The inspector is used to measure wheel data.The detection efficiency is low,the labor intensity is large,and the test result is subjectively affected,which can’t meet the needs of the development of high-automatic railway at this stage.Therefore,this paper proposes a set of high-precision,easy-to-manage noncontact wheelset wear measurement methods.After a detailed understanding of the wheelset structure,the tread parameters,the wear type and the wear evaluation criteria of the specific type of wheelset,a suitable tread profile registration method is designed with reference to the wheelset crosssectional contour feature.Firstly,the feature points of the measured wheel pair and the reference contour curve are coarsely matched,then the contour curvature is calculated for accurate registration,and finally the wear amount is calculated according to the parameter definition.This paper focuses on the method of acquiring the tread curve of the wheel by using monocular vision and line structure light.The image acquisition equipment is selected according to the field of view and the pixel,and the relative positional relationship between the devices is determined.The image and data processing analysis is realized by the acquisition card and the host computer triggering and controlling the device.After mastering the camera imaging model and the lens distortion coefficient calculation method,the appropriate calibration method is selected by comparison,the experimental camera parameters are obtained by using the Zhang calibration method,and the line plane light is calibrated to determine the optical plane equation.The acquisition device is used to acquire images and the image is processed and analyzed by the host computer.The image processing steps mainly include image enhancement,image segmentation,filter denoising and strip center line extraction operations.The appropriate image processing algorithm is selected through algorithm comparison analysis.The image is contoured and restored using camera calibration and distortion correction parameters to achieve contour registration and calculation.In order to facilitate human-computer interaction,a set of image processing analysis software is designed using Qt platform.The functions that can be realized mainly include image acquisition result display,image processing,camera calibration,contour registration calculation,and wear data management,etc.The detectable data includes rim thickness.,rim height,tread wear depth,wheel diameter and the same wheel to wheel diameter difference,detection equipment and software operation is simple,high detection accuracy,plays an important role in vehicle wheel alignment inspection and repair.