Research On Nondestructive Monitoring Methods Of High Speed Railway Based On Vibration Signal

In recent years, high-speed rail network has been developing rapidly, which has been changing thepattern of the existing transportation and has eased the tension of rail transport effectively. The newdesigned high-speed rail infrastructure meets the requirements of speed of at least250km/h.Compared with the traditional train, the high-speed train runs faster, and the traffic gap is rathershorter. Meanwhile, the start-up and braking of the high-speed train do greater damage to the rail,heavy transportation also increases the burden of the track, making it more susceptible to fatigue andfracture. The currently method used for rail flaw detection is based on ultrasonic testing of car-basedand large-scale detection vehicle, which hasn’t kept up with the development of high-speed railway.Testing car works during the train traffic gap, which has shortcomings of slowness, low efficiency,and poor security. Large-scale detection vehicle is widely used in foreign countries, but it also hassome unfavorable factors, for example, the vehicle occupies track, the detection speed which istypically less than100km/h is limited, and most of the detection vehicle can only detects the surfacedamage, but has nothing to with internal cracks in rail head and other kinds of defects. In this context,we propose a new real-time health monitoring method of rail, which uses rail vibration signals toanalysis rail damage. In the theoretical simulation, ANSYS software was used to establishnon-destructive model and models with internal injury, and do dynamics simulation on the models.Analysis vibration response signal by wavelet transform, wavelet packet decomposition, andempirical mode decomposition (EMD), to extract the damage information. We found that EMDmethod was the best way. According to the signal characteristics, this paper proposed anti-symmetricextension method, using EEMD (Ensemble Empirical Mode Decomposition) method to suppress theend effect and modal aliasing, and it’s able to make a preliminary determination of damage. Underlaboratory conditions, built the vibration signal testing system, using homemade pendulum hammerrail model, using ICP accelerometer sensor to collect the vibration signals and transmit the signalsthrough the USB data acquisition card to a computer equipped with LabVIEW program for storage,then import the results into MATLAB for damage analysis. Experiments show that EMD method canidentify the damage.