Reseach on Super-layer Image Detection Method for Rail Flaw

The rail damage status has a significant impact on the safe operation of the rail transit system. In order to ensure the safe operation of the rail transit system, it is necessary to detect the rail damage status. Ultrasonic flaw detection of rails has developed rapidly due to its advantages of safe, fast and accurate detection. At the same time, intelligent image recognition technology has achieved unprecedented results with the interest of deep learning technology and has been applied to ultrasonic flaw detection of steel rails. At present, some people use deep learning image recognition technology to intelligently identify structural damage in the B-scan image. However, due to the overlap and occlusion of the reflection point images in the B-scan image, the geometric feature information is lost, which reduces the recognition accuracy. To solve the above problems, this paper proposes a rail ultrasonic flaw detection method that combines super-layer image and deep learning image recognition technology. This method can effectively avoid the problem of overlapping and occlusion of reflection points and effectively improve the recognition accuracy.