Study On Analytical Method Of Track Geometrical Irregularities For High-speed Railway

The irregularity status of the railway track directly affects the stability and safety of trains,as well as the comfort of passengers.In recent years,high-speed railway has developed rapidly in China.The long-term operation of high-speed railway will cause certain damage to the track structure and then make the track in an irregular status,which will pose a potential threat to the safe running of trains.The track irregularity data,measured by the high-speed railway track inspection vehicles,is the original data to record the status of the high-speed railway line.And the signal processing method can be effectively used to extract useful information from track irregularity data,thereby helping railway workers to grasp the track irregularity status in time,which is conducive to the implementation of track maintenance and repair work to ensure the safe operation of high-speed railway.Therefore,based on the track irregularities data of high-speed railway,this paper studies the signal processing method for track irregularities.The main contents are as follows:(1)The disease of the arch on the track slab is one of the reasons for the track irregularity.The traditional track quality index(TQI)method based on the time domain analysis method is not enough to deal with the detection of the track slab structural disease.Different from the traditional method,the time-frequency analysis method is used to analyze the data of track longitudinal irregularity,and an evaluation algorithm of track slab structure status based on the wavelet energy spectrum is proposed in this paper.The algorithm determines the track slab disease by detecting the time-frequency spectrum peak of the mileage-wavelength.The experimental results and field check results prove that the algorithm can effectively detect the track slab disease and its location.At present,the algorithm has been applied in practice.(2)As a reference for the track spectrum of a certain line,the standard track spectrum can be used to compare with the track spectrum of the line to judge the status of the line in time.Firstly,a seventh-order polynomial fitting model of the track spectra is proposed,which is suitable for fitting the track spectra of ballasted and ballastless tracks of high-speed railway.Furthermore,the calculation method of track standard spectra: average spectrum and 70% spectrum,80% spectrum,90% spectrum based on cumulative distribution is proposed.Based on the method,the standard spectra ofdifferent foundations under rails(turnouts,bridges,tunnels)of ballasted and ballastless track are calculated.Then it lays the foundation for further study of the standard spectra by comparative analysis of these standard spectra.(3)At present,track irregularity data is measured by track inspection vehicles.However,the track inspection vehicles are not only expensive to use but also limited in number,so they can only patrol each line for inspection,which leads to a longer inspection cycle.On the contrary,vehicles operating every day can collect car-body vibration response data at any time,so using car-body vibration acceleration data collected on operating vehicles to inverse track irregularity data becomes the current research direction.The paper designs time-frequency features based on wavelet decomposition,compares a variety of machine learning algorithms,and proposes an inversion algorithm for track longitudinal irregularities based on KNN(K-Nearest Neighbor)regression.Then the influence of factors such as train speed,ambient temperature and track usage time on the inversion performance is analyzed,and the applicable conditions of the inversion algorithm are proposed.The experimental results show that the algorithm can accurately invert the longitudinal irregularity data in the wavelength range of 4?42m.