In this paper, the technique of laser ultrasonic rapid detection of rail surface defects is studied, the interlaced laser ultrasonic defect detection imaging scheme is designed: the laser ultrasonic signal is excited and detected on both sides of the rail, the image is fusion-registered by algorithms such as filtering and image registration to obtain a complete rail surface inspection image to display defect characteristics, which solves the problem that laser ultrasonic is not sensitive to defects. According to the theory of thermal bomb and the proposed detection scheme, a finite element model is established to simulate the propagation process of laser ultrasonic signals in the rail, and the detection signal with surface defects information is obtained. In order to verify the effectiveness of the proposed method, a series of experiments were carried out to obtain the rail detection image with surface defects, and the influence of the laser spot size on the detection image results was analyzed. The experimental results show that the proposed laser ultrasonic imaging detection method can quickly obtain the detected image and effectively display the defect characteristics. The laser spot size has a significant influence on the detection result. When the laser spot is small, the effect of the detection image can be improved. The proposed method provides a reference for further establishing the actual rail inspection system.
Due to the large number of points in the point cloud, the complexity of registration is quite high. To solve this problem, a registration method based on backpropagation (BP) neural network and random sphere cover set (RSCS) is proposed in this study. For the two point clouds to be registered, each is simplified based on the BP neural network. In order to avoid losing a large number of key points in the simplification process, a fixed RSCS algorithm is used for each point cloud to replace the key points with the super-point (SP) sets, and then the SP sets are combined with the simplified data. The iterative closest point (ICP) algorithm is used for fine registration. The point cloud is simplified by BP neural network and fixed RSCS, which reduce the number of points for the subsequent fine registration. Therefore, the time and space complexity can be effectively reduced. Experimental results show that the proposed method effectively improves the computational efficiency while maintaining almost the same precision details, which is of great significance for the registration of point clouds with a large number of points.