TFDS Lateral Trouble Detection Algorithm Research

With the increasing impact of railway freight transportation on the economy of our country,to ensure the safety of freight transport becomes much more urgent.Trouble of moving Freight car Detection System(TFDS)is a detection system designed to ensure the safety of freight transportation.The system currently works in the man-machine operating mode.Most of the detection work needs to be done by a fault inspector.To increase the proportion of "machine" detection in the detection system in order to realize automatic fault detection,can greatly improve the efficiency and cost savings of freight fault detection.According to the structural characteristics and fault characteristics of the portion part of freight car,the corresponding side fault localization and block key loss detection algorithms have been proposed.The main research work in this dissertation includes the following several respects:1.For the localization of the side frame key,the locking plate and the bearing saddle of the side portion of the freight train,based on its position in relation with the wheels,Hough transform circle detection is used to detect the side fault.For the locking plate and the bearing saddle coarse and fine positioning has been used.For the side frame key,after the completion of the coarse positioning and fine positioning operation,accurate positioning operation will be performed.2.For the identification of the loss of the side frame key,according to its characteristics of a triangle shape boundary,a recognition method based on the code feature Fourier description is proposed.First,due to its low gray level,low contrast and unclear boundary contour,the target image is preprocessed to enhance the contour boundary,segment the image track target image boundary.furthermore,the boundary of the image is represented by the chain code and the number of the shapes is counted to ensure the starting point of digital boundary.Finally,the Fourier Descriptors will be calculated by Fourier transform and processed by normalization.The distance calculation formula on the target image is used to diagnose defects.3.Design and implementation TFDS of the side fault detection system.By this system,the Hough transform circle detection on the location of the side fault and the code feature Fourier Descriptors recognition on the defect of the block key could be validated.