Improvement And Application Of LMD Algorithm In Track Irregularity Data Processing Of High Speed Railway

Local mean decomposition(LMD)is a time-frequency analysis method for processing non-stationary or weak stationary signal data.This method can adaptively decompose multi-scale complex natural signal from high frequency to low frequency into the sum of several product functions(PF)according to the characteristics of the signal.At present,LMD algorithm is widely used in mechanical fault detection,seismic wave analysis,feature information extraction,biomedical and image signal analysis.The processing of track irregularity data is an important work for analyzing track space geometric state,evaluating track quality and track maintenance.Dynamic inspection data is a kind of weak stable signal data.LMD algorithm has unique advantages in processing non-stationary or weak stable signal data.Therefore,it is no doubt to expand the application of LMD theory in the field of Track Irregularity Analysis and processing It is a valuable research content.In this paper,the LMD algorithm is deeply studied,and the problems existing in the LMD algorithm are improved,and it is applied to the denoising pretreatment and multi-scale analysis of track irregularity data.The main research results are as follows:In this paper,the overall theory of local mean decomposition is deeply studied.Based on the theory of LMD algorithm,the shortcomings of LMD algorithm itself are taken as the research object.A lot of in-depth studies are carried out on the endpoint effect and the selection of moving average step size in LMD algorithm,and the corresponding complete solutions are proposed.The endpoint effect leads to divergence at both ends of the smoothing local mean function and envelope estimation function,and gradually diffuses to the interior with the number of iterations,resulting in distortion of the final decomposition results.A data sequence extension method based on slope matching of sampling points is proposed to suppress the endpoint effect in the LMD decomposition process.On this basis,the endpoint effect evaluation index based on slope RMS is proposed.Through the analysis of simulation data processing,it is proved that this method can effectively suppress the endpoint effect and improve the decomposition accuracy of LMD,and the effective value of sampling point slope can be used as an evaluation method of endpoint effect.In addition,aiming at the selection of moving average step size in LMD decomposition process,this paper proposes a moving average step size selection method based on golden section rate,which takes the maximum distance between adjacent extreme points of signal as the selection object,carries out golden section on it,and selects the smaller part of the segmentation point as the moving average step size,which is used as the smoothing step size.The simulation results show that compared with the traditional step size selection method,this method has higher computational efficiency and more complete local details,which shows that the improved LMD has better decomposition results.In order to denoise the track irregularity data effectively,a combined denoising model combining improved LMD algorithm and wavelet soft threshold denoising is constructed.Based on the average frequency of a series of PF components decomposed by LMD,the frequency jump degree between adjacent pf components is calculated,and the criterion of dividing order of noise components is constructed,According to the degree of frequency jump,the PF component dominated by noise is determined and denoised by wavelet soft threshold.Through the analysis of simulation experiments,this method can better retain the useful information in the noise component,and is superior to the traditional denoising methods in accuracy and stability.The improved LMD decomposition method is used to decompose the track irregularity data in multi-scale,identify the range of different wavelengths in the dynamic inspection data,extract the hidden regular wavelengths in the track irregularity,make a correct judgment on the track deformation,accurately extract,analyze and reasonably use the various wavelength information contained in the dynamic inspection data,so as to realize the detection and location of local high-frequency track diseases.