The Research Of Typical Fault Image Recognition Method Based On Geometric Model Auxiliary In TFDS

In recent years, with steady and robust economic development, the railway as themain artery of national economy has undergone leapfrog development. Therefore, astable, accurate and efficient fault detection method is of great importance. At present,traditional manual fault detection is mainly applied in train inspection in China.However, manual fault detection is easily affected by many inevitable factors, such asthe weather, occupational level and fatigue degree of the workers. Since traditionalmanual fault detection fails to meet the increasingly higher requirements for safeoperation of trains, Trouble of moving Freight car Detection System (TFDS) emergesand is aimed to eliminate uncertain factors in manual fault detection. With TFDS, faultdetection is accomplished automatically by computers or man-machines, thus theefficiency and reliability of fault recognition are greatly improved. TFDS plays animportant role in improving train inspection quality, railway running safety andtransport efficiency.TFDS involves many vehicle types, the types of component are complex, the faultshave different characteristics and are very strictly recognized. Aiming at thesecharacteristics, this paper mainly studies TFDS typical fault image recognition methodbased on geometric model auxiliary, and combined with the feature of the shape andgrayscale of image, the matching algorithm of geometric moment invariants is appliedto detect the freight cars fault image. In this paper, machine vision is used to replacemanual inspection to achieve the rapid detection of TFDS typical fault.Firstly, the image recognition method of basic primitive is studied, which the basicprimitive includes component parts of circles, lines, etc. in CAD model. On this basis,with the help of linear geometric relationships, the complex primitive which is consistedof ellipse, rectangle, trapezoid and other polygons assisted positioning or faultrecognition is presented.Secondly, due to the influence of illumination and occlusion interference, it isdifficult to separate component, which only rely on its grayscale characteristics.Combined with components apparent geometric contours, TFDS typical fault imagerecognition method based on geometric model auxiliary is proposed and used to achieveside frame key fault image recognition in TFDS.In addition, in order to overcome the shortage of fault image recognition methodbased on geometric model auxiliary, the matching algorithm of geometric momentinvariants is successfully applied to recognize side frame key fault image. Theexperiments had achieved good effect. Moreover, both the matching algorithm ofgeometric moment invariants and fault image recognition method based on geometricmodel auxiliary are analyzed and compared.Finally, TFDS fault dynamic image recognition system has been developedindependently. The system is suitable for a variety of TFDS typical faults on-line recognition. Featured by the high efficiency, reliability and practicability, the system hasachieved good experimental effect.In this paper, the research of typical fault image recognition method based ongeometric model auxiliary in TFDS provides a new solution for solving the TFDStypical fault image automatic recognition, the proposed method lays the foundation forengineering applications in automatic fault recognition of freight cars.