A Sound Source Separation Method for Railside Acoustic Fault Detection of Train Bearings

Abstract

The invention relates to a sound source separation method used for train bearing trackside acoustic fault detection. Microphones installed at the two sides of the track are applied to acquire fault sound signals generated by the wheelset bearing in high-speed passing of the train to act as trackside detection signals x(t), and the processing steps of the detection signals are listed as follows: (1) an overcomplete parametric Doppler modulation harmonic atom library Datom={S0(i), i=1,2...n} is constructed; (2) sparse decomposition is performed on the acquired signals x(t) in the constructed overcomplete atom library Datom so as to obtain a projection atom library Datom2={S1(j), j=1,2...m} and the projection coefficient K={k(j), j=1,2...m}; and (3) the atoms Datom3={S2(k), k=1,2...N1} meeting the requirements are selected out of the atom library Datom2 through screening according to the horizontal distance and the vertical distance from the microphones to the wheelset bearing and the reconstruction signal sig is obtained after linear superposition. The method can be used for train bearing fault acoustic fault detection and has the advantages that the in-band noise can be eliminated in comparison with the conventional digital filtering method.

Description

technical field [0001] The invention relates to a high-speed train wheel-set bearing trackside acoustic fault diagnosis technology, in particular to a sound source separation method for a train wheel-set bearing trackside acoustic fault signal using parametric Doppler modulation harmonic atom sparse decomposition and signal reconstruction, Used to separate the wheelset bearing sound signal from the trackside signal. Background technique [0002] When the train is running at high speed, the sound signal sent by the wheel-set bearing contains information closely related to its health status. A microphone is installed on the side of the track to collect the sound signal and the effective fault diagnosis of the wheel-set bearing can be carried out by means of signal processing. It has non-contact Features of monitoring. However, trackside signals are mixed with acoustic signals from other parts of the train, which brings difficulties to effective fault diagnosis. A common meth...

AI Technical Summary

Problems solved by technology

However, this filter can only filter the signal in the frequency domain, and cannot eliminate the in-band noise that is the same as the bearing characteristic frequency band

Therefore, the denoising effect is not good, resulting in inaccurate diagnosis results

Images