Research On Acoustic Emission Signal Of Track Vehicle Axle Based On Pulse Response Analysis Of VAR Model

Railway is the main transportation mode of our country,plays a very important role in the national economy,and is the forerunner of the national economy development.Railway passenger and cargo transport volume accounts for the total volume of about 55.In recent years,with the acceleration of urbanization,the net inflow of urban population is increasing,so the railway system is undoubtedly the most popular mode of travel,but the transportation safety is the top priority.The good condition of axle is an important condition for safe running of high speed EMU,railway locomotive and freight car.Therefore,if the Acoustic Emission signal of the axle can be used to judge the real time state of the crack of the axle,the safety of the train can be achieved.Of great significance.In this paper,the vector autoregressive model(VAR)in econometrics is applied to the analysis of acoustic emission signals.Since the establishment of the vector autoregressive model needs to be based on the stationary data,the first step is to test the acoustic emission signal by ADF,to determine that there is no unit root,that is,the data is stable,and then to determine the order of lag and the stability of the model,etc.Finally,the impulse response curve is obtained by impulse response analysis.The representative data of pre-crack,mid-crack and late-stage crack were selected as the reference signal group.The pulse response was analyzed by VAR model,and the pulse response curve was obtained by writing s.Tata software program is used to realize the impulse response analysis of acoustic emission signal big data based on VAR model,which can be used to monitor the axle state of rail vehicle in real time.The impulse response analysis of big data's acoustic emission signal is verified,that is,through the real-time monitoring of thousands of sets of data,the impulse response curve of rail vehicle axle crack is obtained.The pulse response value of some acoustic emission signal data is extracted and the curve of three kinds of cracks is output by MATLAB software.At the same time,the time domain parameter eigenvalue of big data is extracted,and the image is established.The results are compared with the results of pulse response analysis.It can be seen that the results of impulse response analysis are better.Go ahead.By improving the experiment,the length of each group of data is changed,and impulse response analysis is continued to obtain the impulse response curve.The result shows that the data length of the acoustic emission signal has a weak effect on the result of the impulse response analysis,but it is still possible to distinguish the early crack state,the medium-term crack state and the later crack state by the characteristic of the curve.