Diagnosis Of Track Structure Anomaly Based On The High Order Statistical Characteristics Of Train Vibration Response

High-speed railway has become an important part of transportation in China.With the growth of its service period,various diseases have begun to appear in the track structure,which is the basis of train operation.The performance degradation of the track system caused by these diseases will inevitably affect the safety and comfort of driving.Although the existing methods of rail inspection and mobile inspection can carry out regular detection of track structure,their practicability and economy restrict the real-time detection of track anomaly.The dynamic response on the operating train has the advantages of convenient and fast testing,and there will be obvious non-gaussian mutation in the abnormal track caused by the train passing diseases.If the non-gaussian detection method of the signal is used,it is expected to be able to determine the abnormal time point quickly and efficiently.The high-order statistical method is effective in non-gaussian signal processing.Based on this,this paper intends to establish a track anomaly irregularity detection framework based on the high-order statistical characteristics of train vibration response by using the high-order statistical theory,so as to achieve the purpose of rapid diagnosis of track anomaly.The main contents of this paper are as follows:(1)This paper systematically introduces the basic theory of non-gaussian feature extraction of signals based on higher-order statistical theory,and gives the definition and basic properties of higher-order statistics such as higher-order cumulant,higher-order moment spectrum,higher-order cumulant spectrum,and higher-order time-frequency distribution,as well as the advantages of higher-order statistics in processing non-gaussian,non-stationary and nonlinear signals.(2)The train-track model is established based on the multi-body dynamics software.By superplacing the typical abnormal track irregularity waveform on the track random irregularity spectrum as the wheel rail excitation,the sensitivity of the dynamic response of each part of the train to track anomalies is compared,the feasibility of diagnosing common track structure anomalies based on the train vibration acceleration response is verified,and the bogie is determined to be the optimal measuring point for the train vibration response.(3)Three commonly used high-order statistical features,namely,high-order accumulator,bispectrum and wigner time-frequency tri-spectrum,are selected to extract the segmented features of the vibration response of the train under various abnormal conditions,so as to realize the location of the abnormal time and point,and the influence of such factors as the order of high-order statistics and the segmented length of data on the recognition effect of abnormal signals is studied.A framework of track anomaly unevenness detection based on piecewise high order statistical characteristics of train vibration response is proposed through several simulation calculations and induction analysis.(4)Based on the rail abnormal disease situation given by the railway department,taking the measured train vibration response of high-speed emu through the disease site as the analysis object,the validity of the rail abnormal irregularity detection framework based on the highorder statistical characteristics of train vibration response is further verified.