Research On Rail Corrugation Detection Method Based On 1D Laser Displacement Sensor

In recent years,China’s railway transportation has developed rapidly,the total mileage of railway operation has broken through a new high,and the railway maintenance market is entering a period of explosive growth.The investigation shows that the rail corrugation has become one of the main forms of rail damage,and the chord-based measuring method is an important method of rail corrugation detection at home and abroad.However,in the actual testing process,the abnormal situ ation of the chord measuring point is not fully considered.On the one hand,when the testing equipment collects the chord measuring points,the specular reflection on the surface of the rail and the gap between the rails lead to the lack of data during the acquisition process;.On the other hand,due to the existence of rail joints,the collection results of chord measuring points will have abnormal steps.In view of the above abnormal situation in the process of rail corrugation detection based on chord measurement method,this paper carries out relevant research.This paper first introduces the basic flow of the dynamic detection method of rail corrugation,according to the design requirements,gives the overall framework of the rail corrugation detection system,and explains the hardware system and software system in detail.Among them,the hardware system mainly introduces the detection principle of 1D laser displacement sensor and the reasons of the sensor used in this system.Meanwhile,the specific parameters and functions of the relevant circuit elements are introduced in detail.The software part gives the design process and specific implementation method of the rail corrugation detection algorithm.In this paper,an adaptive Bi-LSTM based chord measuring point reconstruction method is proposed to solve the problem of missing chord measuring points in data acquisition.The model can effectively extract the information before and after the string measurement point,and use the trained model to reconstruct the chord measuring point.The model adopts an adaptive output modul e,which solves the problem that the current single-step or multi-step Bi-LSTM prediction model can only The problem of doing fixed-step output.Using the 50 m chord measuring point data,randomly select 1%-15% of 15 kinds of data with different degrees of missing to remove the experimental data.The experimental results show that the data reconstruction effect of this method is better than that of the traditional mean filling and other networks,which provides a new idea for dealing with the missing of the rail surface chord measurement points.In order to solve the deformation problem of data at rail joints in complex rail lines,this paper presents a correction method for abnormal data of rail joints based on slope subsection.On the basis of the existing chord-based measuring method,this paper modeled and simulated the data acquisition process at the rail seam,and pointed out that the measurement method used for the normal rail cannot be used mechanically.The method corrects and reconstructs the abnormal data in sections according to the change of the feature points at the rail gap.Compared with the ordinary chord-based measuring method,the mae,rmse and mse of the corrected measured value and the reference value decreased by 72.54%,82.21% and 96.84%.respectively.,respectively,which provided a technical reference for the detection of the irregularity of the rail joints by the laser displacement technology.