Research On Train Bearing Fault Diagnosis Based On Acoustic Signals

Rolling bearings are the key components of trains,which have been working in the harsh environment and subjected to heavy alternating loads for a long time,and are very prone to failures,and train bearing failures are one of the most important causes of train failures and even accidents.Therefore,the research on the fault diagnosis of train bearings is of great significance to ensure the safety of train operation.Compared with the traditional rolling bearing fault diagnosis methods,such as vibration signal analysis method,the rolling bearing fault diagnosis method based on acoustic signal has many advantages such as simple installation of equipment and non-contact measurement,but the influence of Doppler effect brings difficulties to the analysis of acoustic signal.In this paper,we study the establishment and analysis of the acoustic signal simulation model of trackside train bearing fault under the influence of Doppler effect,based on the problem that the experimental signal data is not close to the actual problem in the current research of trackside bearing acoustic signal fault diagnosis.Firstly,the causes of Doppler effect were analyzed from two perspectives of acoustics and kinematics according to the different relative positions of the source and receiver,and the Doppler effect was not only related to the relative motion of the source but also related to the relative positions of the source and receiver,and the formula of frequency change before and after the Doppler effect and the formula of frequency deviation rate were derived to lay the theoretical foundation for the establishment of the simulation model.Then,according to the derived frequency equation,a simulation model of the Doppler effect on the sinusoidal signal was first established in Matlab,and the effect of acoustic attenuation was added to the model according to the Stokes-Kirschhoff formula to make it closer to the actual acoustic signal.Finally,the sinusoidal signal at the sound source is replaced by the experimental data of bearing failure at Case Western Reserve University to establish the simulation model of train bearing failure acoustic signal,and the correctness of the model is verified by car experiments.On the basis of the established simulation model of train bearing fault acoustic signal,the correction of Doppler distortion was studied.The frequency skew rate based equal frequency bias resampling correction algorithm was applied,and its feasibility was discussed based on the principle of Doppler effect generation.Based on the characteristics of the bearing fault signal,the best algorithm parameters were determined by several tests,and the bearing outer ring and inner ring fault simulation signals were successfully corrected,but the calculation time was too long.In order to correct the distortion quickly,a fixed frequency offset resampling correction algorithm based on the frequency offset rate curve is also proposed,and its ability to quickly recover the Doppler distortion signal whose frequency changes drastically in a short period of time is verified by the simulated signal.