Resersch On Target Catching Of Uncoupling Automation System For Hump Yard

Our country territory is vast, and also large number and variety of goods transport,coupled with the rapid economic development all the factors have proposed a severe test forrailway traffic, so build a large-scale marshalling station to transit goods is imperative. Atpresent, China has build Xin Fengzhen, Chengdu bei and other large marshalling stations, andusing the comprehensive automatic system in marshalling station to improve efficiency, butthe uncoupling operation in the vehicle disintegration process is still completed manually. Inorder to further improve the efficiency of disintegration and marshalling, amelicrate workerconditions, and ensure the safety of the workers, there is an automatic uncoupling deviceinstead of manual operation.Reserch of uncoupling automation system in this thesis is only limited in the part ofpreoptic obtained, the binocular vision technology is applied in the process of automaticuncoupling to measure the distance betweenthe coupler and equipment, the3D coordinatesthat have been obtained can be used to control the robot to complete uncoupling operations.Although the coupler type of freight trains in China is complex, but if catching goal couplercan be transformed into getting3D information of uncoupling handle, the difficult ofdesigning system can be creatly reduced.Large marshalling have large amount of work, and continuously working day and night,so the image acquisition background is complex and changeable. Because of image captureand transmission process is closely related with the outdoor environment and illuminationchanges, the quality of image should be improved for using the methods of histogramequalization and image smoothing that having been improved, eliminate the effects of lightand noise, and then use the improved edge detection method to cat up the image, so that it canreduce the amount of information and reserve the features. In order to distinguish the positionof handle in the complicated image background, the template matching method can be used inthe thesis for processing image. To ensure the efficiency of the disintegration, the uncouplingoperation must be achieved in the situation that the vehicles and equipment are in the state ofmotion, at this time, the handle position in the image may be appeared the situation withtranslation and stretching. So in the thesis, the template maching that based on the geometriccharacteristics was used to match the region of interest in the image, and have optimized thealgorithm by adding termination condition of calculating similarity measure, improvingsearch strategy, using the method of least squares to improve matching accuracy.Purpose of binocular vision technology is to get the information of3D coordinate foruncoupling kandle. Firstly, getting the mapping relationship between image coordinate andworld coordinate from using camera calibration with neural network which is optimized by improved genetic algorithm. Secondly, using the method that combined template matchingwith feature matching, and then giving stereo matching to the region that contains informationfor uncoupling kandle of the image, with the edge point as a feature point in the region,calculating the parallax with the corresponding relation that one point has projected in twoimaging plane, so that it can get the depth information of handle. Finally, through mappingrelationship that is built by neural nerwork calibration,2D coordinates of uncoupling handlein the world coordinate system can be got, and then the3D information can be got after thedepth information is calculated.The research results of this thesis is to get the3D imformation of uncoupling handlewhich is connected with goal coupler through binoculat vision technology, the uncouplingautomation system can use the information to control manipulator to complete the uncouplingoperation.