Research On Railway Track Irregularity Prediction System And Its Application

In recent years,the construction of railway has been accelerating,and the standard requirements of high-speed railway and heavy haul railway are getting higher and higher.How to make the train run continuously,safely and smoothly according to the specified speed is the primary task of the railway maintenance department.By studying the detection data of the track inspection vehicle and deeply studying the change law of the track line equipment status,it will greatly help the railway maintenance department to carry out the production work scientifically,reasonably,economically and efficiently.First of all,in the preprocessing stage of the data detected by the rail inspection vehicle,there are two problems in the measured data of the rail inspection vehicle: abnormal value and mileage offset.In order to ensure the accuracy of the experiment,we need to preprocess the basic data.The preprocessing methods are absolute mean correction method and trend similarity data offset correction method.Through experiments,it can be seen that the preprocessing of these two kinds of data can find abnormal values and correct them,and can also calibrate the data with offset.Secondly,the paper uses wavelet decomposition and reconstruction theory to solve the problem that the time series of rail inspection vehicle detection data is nonstationary.The wavelet decomposition method is used to process the original data to obtain multi-level stable series,including high-frequency stable series and low-frequency trend series;According to different types of multi-level stationary series,select appropriate prediction models for prediction;After that,the wavelet reconstruction method is applied to recombine the above decomposed stationary series to obtain the prediction data of the original line section state time series.Finally,this paper takes thirty-three batches of TQI detection sequences of Jinshan uplink K200 + 000-k200 + 200 detected by rail inspection vehicles from January 9,2020 to May 10,2021 as the data object,and predicts the detection data of five batches of rail inspection vehicles from May 10,2021 to July 11,2021,and compares them with the real detection data.At the same time,the grey system GM(1,1)model and ARIMA model are used for comparative experiments.Comparing the prediction results of the three models with the actual detection data,it is concluded that the GM-Arima prediction model proposed in this paper has better prediction accuracy.In combination with the prediction model,the maintenance location and maintenance time determination methods are displayed,and the key problems faced by the railway maintenance department such as "what kind of maintenance","when to repair","where to repair" are discussed.