Research On Identification Method Of Subway Short–Wavelength Rail Corrugation Based On CEEMDAN–MPE Algorithm

Rail corrugation is a common way of track damage in the subway system,which is a wave-like periodic vertical change of track geometry failure of the rail.While running through the wave grinding section,severe vibration will be generated on the train,thus affecting the reliability of vehicle components,reducing the comfort of passengers and threatens the safety of the train.The corrugations of subway rails are mostly short-wavelength corrugations.Aimed at improving the low efficiency and low accuracy of current method in shortwavelength rail corrugation detection,the inertial acceleration method is applied to acquire the vibration acceleration signal of the axle box of the operating vehicle.Algorithms based on complete ensemble empirical decomposition with adaptive noise(CEEMDAN)and multiscale permutation entropy(MPE)algorithms are introduced to analyze the signal.In this way,on-line detection of rail corrugation faults comes practical,which provides a basis for track maintenance,and also makes contributions to traveling safety and comfort.Based on model simulation analysis,algorithm design and verification,this thesis mainly carries out the following works:(1)The basic principle of on-line detection of rail corrugation is summarized,a simplified vehicle-rail vertical vibration model consisting of a simplified model of rail corrugation and a vehicle lumped parameter model is established,the model is solved,and the analysis verifies the feasibility of rail corrugation detection based on the vertical vibration acceleration signal of the axle box through theoretical derivation and data analysis.(2)The online detection method of subway short-wavelength rail corrugation based on the CEEMDAN–MPE algorithm is studied,the vibration response characteristics caused by corrugation are analyzed,the multi-scale arrangement entropy,correlation coefficient and energy value are introduced,and the decomposition results are screened and judged;determine corrugation parameters with wavelet Packet time-frequency analysis and frequency domain integration.(3)The vehicle-track coupling dynamics model with flexible rails is established,and different track excitations are applied to simulate,and analyze the vertical vibration on the axle box caused by corrugation under different wave grinding wavelengths,wave depths and train operating speeds.The parameters of the algorithm are determined by analyzing the simulation data.The simulation data of two kinds of corrugation superimposed excitation are analyzed,and the method in this thesis is verified.(4)In order to further verify the correctness and engineering feasibility of the proposed method,the measured signal is selected for analysis.The results show that based on the CEEMDAN–MPE algorithm,the short-wavelength rail corrugation and the occurrence area is able to be identified through the vibration acceleration data of the axle box;The method gains better accuracy in identifying the short-wavelength wave grinding below 80 mm than the EEMD energy value and the wavelet packet energy entropy method,it is capable of extracting the vibration response caused by the irregularity of the short-wavelength wave grinding,and identify the wavelength and depth of the wave grinding more accurately.