Research On Identification System Of Bearing Ring Surface Defects

Bearing fault of railway vehicle is one of the main troubles in railway transportation, after running a certain mileage, the bearing parts of vehicle need to be inspected in repair factory, fluorescent particle magnetic detection is the main way of bearing detection.Aiming at the existing problems of the manual inspecting in fluorescent magnetic detection, the automatic technology is researched for the identification of the fluorescent magnetic particle detection results. Analyzing and identifying the result of bearing fluorescent magnetic particle detection are based on digital image aquisition and processing. Embedded ARM-Linux system is built, with implemention of the image aquisition through connection with a USB camera, which needs coordination with the stepper-motor-driving rotation of bearing ring to aquisite the image of cylindrical surface.The defects' status of bearing ring surface are identified and determined by image processing. Based on the transplantation and use of OpenCV (open source for computer vision library), the image processing program aquisits image suspicious ingredients by designating the color scope, and expresses the suspicious ingredients against background via a binary image. With pre-processing of this binary image mainly by mathematical morphology method, the area of regional connectivity and round degrees are calculated to identify the crack by comparing with the specified standard. Identification result is effective validating the feasibility of the program.Hardware design and construction of operating system platform are also described in detail for the embedded ARM-Linux system, which is of significant reference value for other designs of intelligent instruments using embedded system.