System For Locomotive Pantograph And Roof Status Detection

Locomotive pantograph and roof state detection is one of the main contents of locomotive on-line monitoring system.Traditional manual detection methods are complicated and have potential safety problems,so it is gradually being replaced by a variety of automatic detection methods.The detection performance and efficiency are improved.This thesis focuses on locomotive pantograph and its roof state detection system based on image processing,and mainly studies pantograph status detection,roof state detection,design and implementation of detection system:1)Study on pantograph state detection.First,the pantograph abrasion detection includes pantograph image preprocessing,slide area positioning,slide edge detection,upper and lower edge positioning and abrasion value calculation.As the pantograph slide in the image is darker than other area,the integral map skill is used to locate the pantograph slide area.Because of the irregular edge of the slide,a single pixel-based directional tracking algorithm is proposed to locate the upper edge of the slide.During the detection process,perspective transformation is used to correct geometric distortion of pantograph pictures,image enhancement methods are used to improve image quality.Canny edge detection is used to obtain the edge line of pantograph slide image so as to locate the pantograph catenary and the upper and lower edges of the slide.Then calculate the thickness of slide and draw the abrasion curve.According to coordinate results after extracting corners of pantograph heads with Harris corner detection,compare with the standard images and calculate offset of pantograph centerline.The experimental results show that by using these methods the slide area can be quickly located,edges of pantograph slide can be detected accurately,pantograph abrasion value and centerline offset value can be calculated precisely.2)Study on roof status detection.According to the condition of long locomotive body,wide roof and non-uniform approach,the roof status detection method based on area obtain camera and line scan camera roof image acquisition is studied respectively.In order to collect the images of the line scan roof image,a linear calibration based on vehicle-speed is used to correct roof images.The roof images collected by line scan cameras are spliced by the multi-resolution matching method based on the pyramid data structure to finally complete the real roof generating.For the image acquisition of area obtain roof image,the SURF algorithm which takes advantages of overlap region feature points is used to splice the image of the frame one by one.Finally,detections of foreign objects are completed based on the improved symmetric differential algorithm.The experimental results show that roof images which are obtained by linear scan camera is superior to the area one.3)According to the standards and requirements of rail transit and actual conditions of the installation site,the pantograph and the roof status detection system of the locomotive is designed and implemented.Field tests show that the pantograph abrasion detection error is less than 0.6mm,also can quickly generate locomotive roof images and detect foreign objects.At present,the system has been installed and commissioned at the Nanxiang Depot of Shanghai Railway Bureau.