Study Of The Online Automatic Non-destructive Detecting System Of The Cracks In Track Wheels

The wheel plays a crucial role in track, so the railroad department always pays much attention to detect crack and determine its location in track wheel. Presently the methods used in home and overseas to detect crack in track wheels are basically static non-destructive test (NDT) techniques, such as vortex, magnetic powder and ultrasonic. But in recent years, it is a new trend in developed countries to develop in-service or on-line NDT techniques to dynamically detect the cracks in track wheels. According to the characteristics and requirements to realize on-line NDT of track wheels, the functions are thoroughly researched on of all parts that consist of a detecting crack system to be designed in this dissertation.The effects of macro and micro damages that contribute to the growth of cracks in track wheel are first carefully analyzed, and according to the theory of fracture mechanics the law of crack growth is formulated, then a method to determine the intervals of detection is put forward based on the design theory of damage tolerance. In the process of plastic deformation, crack initiation and growth, and impact, the characteristics of acoustic emission (AE) signals and the factors that affect them are also elaborately discussed. And the propagating equations of P-wave, S-wave and R-wave are deduced mathematically from the correlation between stress and strain in elastics mechanics. The components and their propagating characteristics of Lamb wave are mainly dealt with, and the causes and features of dispersion effect are further expounded. All these features are very important in the crack detectingWhen to design the hardware, not only because transient AE signals have rich ultrasonic modes, but are transmitted by media and thus maybe reflected, attenuated and dispersed, the broad-band AE transducer is used to capture the transient signals, and the entire signal-collecting channel is designed to have wider band-pass and higher signal-to-noise ratio. The digital signal processor (DSP) is also used to process the signals with high-speed and real-time.Furthermore, owing to possible great noise signals mixed with AE signals and the overlap of their spectrums after being filtered, an adaptive noise compensator is designed to eliminate the interference of the background noise and therefore improve signal-to-noise ratio of the designed system. A digital filter is also needed to further suppress unwanted signals. In processing the AE signal, the analysis tool of time and frequency-wavelet package is adopted, in order to confirm the characteristics of time and frequency in strange points of the signal.All results from every DSP, such as time and frequency values of strange points and characteristic values presenting the AE signal, are carried by CAN field bus, and finally reach receiving buffer zone of industrial control computer. The data are subsequently inputted into BP neural network to nonlinearly calculated the stress intensity factor range, represented by AE signals. Then the crack propagation rate and its size are calculated according to Paris formula of fracture mechanics. In the meantime, all data, including received data, calculated results from neural network, dynamic rotational speed of the track wheel, crack propagation rate and its size, are all stored in database, to make analysis and inquiry efficient.