Study Of Fault Diagnosis Of Wheel Thread Flats

The wheel tread flat is referred to a local flat-type wear of the wheel tread,which is caused by the wheel sliding on a rail due to excessive braking force or traction force.It is a kind of wheel polygonal wear and it is a main cause leading to axlebox bearing damage, shaft temperature increasing and the fracture of axles, rails and concrete sleepers. Therefore, the diagnosis of the wheel tread flat has become a key technical problem which must be solved in the development of high speed trains and heavy haul trains.The vibration signal due to wheel tread fault is nonlinear, non-stationary. The traditional signal analysis methods can not be used to effectively analyze it. At present,the traditional fault diagnosis methods of wheel tread flats have low accuracy,while only valid at low speed.When the train is running at a high speed,these methods are ineffective. This article propose a new fault diagnosis method of wheel tread flat through theoretical study. The expirimental test shows that this method can detect wheel flat facult accurately and effectively in low or moderate speed conditions. At high speed,the detection accuracy of this method is reduced.The central part of the new fault diagnosis method is empirical mode decomposition(EMD).It is an adaptive data-driven decomposition method for signals. It can decompose a complex signal into a series of basic intrinsic mode function(IMF) components in the time domain.The empirical mode decomposition method depends on the signal itself, decomposition of different signals will produce different components. Thus EMD is especially suitable for the analysis of non-stationary signal. In order to diagnosis the fault of the wheel thread flat, a new signal analysis method is proposed which combines the EMD, Hilbert-Huang transform and envelope. Firstly, do the EMD of the fault signal is performed, and then the Hilbert envelope of each IMF component is carried out. Finally, the spectrum analysis of each IMF Hilbert envelope is performed to, obtain the EMD envelope demodulation spectrum of the original signal. If the peak frequency of the EMD envelope demodulation spectrum coincides with the wheel rotational frequency, the wheel is considered to have a wheel flat. The expirimental test shows that this method can detect wheel flat facult accurately and effectively in low or moderate speed conditions.Most of the traditional signal analysis methods were implimented by MATLAB.However, the human-machine interface of MATLAB is poor, and MATLAB does not have the data acquisition function. Considered that system shuld be simple and practical, this paper choose one of the American NI company’s software called Lab VIEW to design the system, the software has powerful data acquisition function and signal analysis function, intelligent human-machine interaction, and easy to make a good using interface. This paper combines the PXle-4492 data acquisition card and Lab VIEW to design a detection system for wheel tread flats which has the data acquisition and analysis function by programming. Moreover, it has a good interface.