Three-dimensional Profile Reconstruction Of Railway Wheel-rail Based On Moire Profilometry

Railway safety is a major event related to the national economy and people’s livelihood.Regular inspection of railway key parts such as wheel and rail can provide a strong guarantee for the safe operation of the railway.Using structured light three-dimensional reconstruction method to obtain the three-dimensional profile of wheel-rail surface quickly and accurately is one of the important means for defect detection.As a three-dimensional measurement method of structured light,moire profilometry has the advantages such as single frame reconstruction,high speed,no contact and high precision,so it is suitable for dynamic reconstruction.In this thesis,moire profilometry is applied to the railway field to reconstruct the three-dimensional profile of wheel tread and rail.The digital sinusoidal gratings are generated by computer and projected to the surface of object.The moire fringe patterns are obtained by fringe superposition and frequency domain filtering.Finally,the phase information of the object is extracted from moire fringes,which provide a new method of wheel-rail three-dimensional reconstruction.The main research results of this thesis are as follows:1.In this thesis,the simulation of key steps of moire profilometry are realized,and the results of moire profilometry and other structured light three-dimensional reconstruction methods are compared.It is proved that moire profilometry has the advantages of both high precision and single frame reconstruction,which provides experimental support for its application in railway dynamic high-precision three-dimensional reconstruction in the future.2.In this thesis,the effects of different filtering methods on the extraction of moire fringes are compared,which proved the importance of frequency domain filtering for moire profilometry.It is necessary to select suitable filtering methods according to different environmental conditions in order to get a relatively good reconstruction results.At the same time,moire profilometry is used to reconstruct objects with noise,which proves the anti-noise ability of moire profilometry.3.The experimental platform for three-dimensional profile reconstruction of wheel tread and rail is built,and the depth of the measurement system is calibrated.The three-dimensional profile reconstructions of the wheel tread,wheel tread with hexagonal nut and rail are realized,which proves that the moire profilometry can be used in the three-dimensional profile reconstruction of wheel-rail in railway.4.In the process of three-dimensional reconstruction of moire profilometry,it is often need to select different filtering methods to extract moire fringes according to different conditions to achieve a relatively good reconstruction results.Therefore,in this thesis,deep learning and moire profilometry are combined.By using the deep neural network to train the simulation data set,the moire fringes can be extracted directly from the deformed fringes,which replaces the frequency domain filtering step in the traditional moire profilometry.The experimental results show that the moire fringe information in the deformed fringes with or without noise can be well extracted.In addition,the deep neural network also has a good performance in phase unwrapping with noise.It provides a new idea for the combination of deep learning and moire profilometry.