| In the traditional train approach warning system,most of the signal acquisition of sound vibration is achieved through contact sensors.However,one of the salient weak points of the system is that in rainy or foggy days rails would get rusted.Thus,the sensors cannot be attached to the surface of rails,which leads to the failure of signal acquisition.The laser is a non-contact detection method that leaves no traces.Designing a laser non-contact train approach warning system by using the directionality,coherence and monochromatic characteristics of the laser.(1)This paper firstly models the noise generated by wheel-rail vibration.Based on the Remington wheel-rail rolling acoustic model,three vibration methods including the Eurail beam vibration,the overall vibration of rails and the lateral vibration of the two-point fixed rod,and a new wheel-rail noise model is established.The mathematical model is used to analyze the frequency of the train-rail vibration in all stages so that an inherent frequency will be acquired.By contrast,it is concluded that when the wheel-rail vibration is below 1000 Hz,the wheel-rail noise is generated by the overall vibration of the wheel-rail.(2)The Matlab software is used to analyze the wheel-rail sound signal in both time and frequency domains.Besides,the signals and warnings of train approaching will be affirmed through the detection of the short-term average energy of the wheel-rail noise signal and the amplitude change of the power spectrum at 300 Hz.(3)This paper designs an improved spectral subtraction method.In order to obtain a smooth signal spectrum,the entire signal is segmented by 30 ms to form a number of short-term power spectra.Then,based on the short-time power spectrum,the improved spectral subtraction is used to denoise the wheel-rail noise and get its pure sound.(4)A hardware system is set up based on the laser acquisition device,preamplifier circuit and MPS010601 data acquisition card.Then design simulation software to verify,use virtual instrument platform LabVIEW to build a simulation software system which includes wheel-rail noise acquisition,analysis,identification and warning functions.To analyze the spectrum characteristic of the wheel-rail noise in both time and frequency domains.Using the recognition algorithm and related data parameters,a virtual instrument system for train approaching warning is constructed. |
