Research on 3D Inspection Method for Surface Scratches on High-Speed Train Axles and Development of Inspection System

In order to achieve automatic detection of scratches on the surface of high-speed train axles, a three-dimensional detection system for high-speed train axle surface scratches was established. The algorithms for scratch defect image recognition, three-dimensional morphological detection, and scratch prediction based on multi-source data fusion were studied for the system. First, the characteristics of scratch defects on the surface of high-speed train axles were analyzed, and the scratch defects were quickly identified and located. Then, the high-precision line laser sensor was used to detect the three-dimensional morphology of the scratches located by the algorithm. After that, the width and depth of the detected scratches were measured. Finally, the system was used to conduct data verification on standard samples. The experimental results show that the scratch recognition accuracy of the algorithm is 98.62%, and the accuracy of the scratch detection is $\pm 0.1 \mathrm{~mm}$. The system can basically meet the requirements of stability, reliability, and high precision for the three-dimensional detection of scratches on the surface of highspeed train axles.