Study On The Automatic Measurement System For The Abrasion Of The Bearing Weight Saddle Based On Computer Vision

The Bearing Weight Saddle, an important accessory of railway trunk, is installed between the freight car bogie and the railway bearing, receiving many kinds of load from different parts. In the train's running abrasion occurs due to the friction between the freight car bogie and the railway bearing. The abrasion beyond the usual level may bring great accidents to the running train. In order to avoid such dangers, abrasion must be measuring termly. There is a special department in railway industry naming "periodical measuring plant" which implements this measuring schedule. But until now, the abrasion is still examined using simple mechanical tools such as measuring templet the end of which is that we cannot get an exact result of the abrasion. So many researchers made up diversified doable measuring project trying to solve this problem. Some of them have even designed automatic measuring equipments based on mechanical frameworks. Since these projects all have the defects like costing too much, low precision and so on, they are not popularized in measuring plants. As a non-touched, rapid and precision measuring means, the computer vision inspection technique is appropriate for measuring those free-form objects like the Bearing Weight Saddle, and also, has many virtues comparing with other mechanical measuring methods.In the thesis, an automatic measuring method of the Bearing Weight Saddle abrasion is introduced based on binocular stereo technique and line structure light. The corresponding measuring system has also been set up which includes four parts, namely, the vision sensor system, the mechanical system, the controlling system and the software. The equable measuring results reveal that the system has high preciseness. This means that this system will be a very solution for automatic measurement of the Bearing Weight Saddle abrasion. Because the measuring criterion issued by National Railway Department isn't vivid enough for measurement, a thorough research on this criterion has been made. The thesis also introduces a special calibration method for vision sensor system in respect that we find the usual calibration ways can not be used in this system. Finally some problems encountered during experiment and the corresponding resolvents are presented. The course of system testing is indispensable in perfecting the automatic measuring system. ...