The Fault Diagnosis And Software Implementation Of Train Angle Cock

In order to meet the needs of the situation, improve traffic safety, China hasindependently developed a set of the TFDS system(Trouble of moving Freight carDetection System). The system integrates multiple technologies, makes up thedefect of inspection practices of the traditional train, improves the quality andefficiency of train detection and guarantees the running safety of train.Designing a high-precision and high-efficiency fault diagnosis algorithm forAngle cock images is the difficulty of this article. Through in-depth analysis andresearch, this paper has designed two sets of high accuracy, fast and good faultdiagnosis algorithm which have been named by the bottom fault diagnosisalgorithm and side fault diagnosis algorithm. Each fault diagnosis algorithmcontains fault location algorithm and fault detection algorithm. They accuratelycomplete accurate positioning and fault judgment of Angle cock of the bottom andside images. In addition, in this paper, the final algorithm which is a combination ofthe two sets of fault diagnosis algorithm, completes the fault diagnosis of Anglecoke and provides a good support for TFDS.In this paper, the design and implementation of the algorithm make full use ofdigital image processing which contains mirror, zoom, Hough transform, erosionand dilation and other classic approach. In addition, this paper improves the imagebinarization using a combination of the local threshold and the global threshold,strengthening the image segmentation effect. Otherwise, this paper also improvesgray operator using grayscale mapping and windows grayscale mapping approachtargeting the edge, makes the results more accurate positioning. In the templatematching processing, this paper uses image compression method improve greatlythe efficiency of the algorithm.In order to meet the demand of design and test, this paper designs two sets ofman-machine interface. Batch extracting image interface extracts Angle cockimages with simple, rapid Characteristics, and enables separate side and bottompictures. Data displaying and algorithm testing interface has realized the algorithmdesign and after the design of program testing, ensuring the accuracy of thealgorithm. On the data display method, in this paper, the improvement was carriedout. The improved method has less code, strong legibility and easy operation.To solve well the solution between a host program and dynamic invocation, thispaper presents a improved method of DLL encapsulation which is novel, convenientdebugging, easily invoked by TFDS system automatic recognition platform. Paper finally introduces testing methods of the Algorithm, they solve the testingof a single image and batch images, ensure the validity of the test algorithm. Thefinal test result is less than8%, non-response rates to0%, meeting the requirementsof the TFDS system.