Rail Wear Measurement Based On Structured Light

With the rapid development of high-speed railways in my country,high-speed trains have been operating under high-speed,heavy-load and high-density operating conditions for a long time,making the rail surface wear increasingly serious,which may cause serious safety accidents.Therefore,the detection and analysis of the rail profile and its Wear measurement has gradually become one of the very important inspection tasks of the railway engineering department.The measurement technology based on structured light is the research focus and hot spot in the field of visual measurement today,and has been widely used in various fields of industrial measurement.Therefore,the rail wear measurement method based on structured light has become one of the hot topics in the safety inspection of railway infrastructure..This article focuses on how to achieve accurate measurement of rail wear in a dynamic,high-speed,and high-noise driving environment,which has important theoretical significance and application value.The main tasks of this paper are:1.The traditional laser sensor for rail profile detection uses a non-contact test measurement method.The laser sensors are installed on both sides to dynamically scan the inner and outer sides of the rail,resulting in a large system,a complex structure,and a high price.At the same time,when the collected images are stitched together,the processing of discrete points will lose precision to a certain extent,resulting in loss of wear information.For this reason,this paper adopts the measurement method of single-side installation of laser sensors,and proposes a light strip center extraction algorithm that can resist noise.The algorithm uses a density clustering algorithm to deal with outliers,making the algorithm robust to noise.2.The feature point selection of the traditional profile alignment method cannot handle the special situation that the collected image appears at the end of the rail head,which will cause the feature point matching to fail,and thus cannot complete accurate alignment.For this reason,this paper proposes an effective identification and alignment method for special rail types.This method uses the feature of no wear at the end of the rail head to align the profile and uses the straight part below the rail waist to correct the results.In the experiment A good alignment effect has been achieved.3.Aiming at the problem that the collected profile data needs to be aligned with the standard profile rotation,this paper proposes a noise-robust two-stage improved ICP(Iterative Closest Point)algorithm.A removal operation for interference matching noise is added between the two-stage ICP matching algorithm,which improves the matching accuracy of the ICP algorithm,and can obtain high-accuracy wear measurement results.