Profile construction and determination of geometric parameters of the rolling surface of a railway wheel using a laser profilometer

The problems arising in the processing of optical images formed from laser signals reflected from the surface of a railway wheel are considered. A design has been developed for the placement of laser emitters and photodetector devices for examining railway wheels in motion. A new algorithm has been developed aimed at improving the accuracy of constructing the profile of the rolling surface of the wheel. To process the recorded optical images in the laser radiation reflected from the surface of the wheel, a new technique based on the use of data filtering has been proposed, which uses a new algorithm that allows removing glare (a cloud of points with false data), outliers (single false points) and leaving only the necessary profile points. The new algorithm is based on the HDBSCAN (Hierarchical Density-Based Spatial Clustering of Applications with Noise) clustering algorithm. The algorithm makes it possible to exclude points forming errors during the construction of the profile of the rolling surface of the wheel. The use of a new design for the placement of optical elements and an algorithm made it possible to increase the accuracy of measurements of key wheel parameters, such as thickness, height and steepness of the ridge, and uniform rolling.