Research On Diagnosis Method Of High-speed Train Axle Box Bearings Based On Optimal Empirical Wavelet Transform

In recent years,with the rapid development of China’s economic construction,higher requirements have been put forward for the railway carrying capacity,which is closely related to the running speed and safety of trains.Increasing the speed of trains and ensuring the safety of trains is a necessary requirement for the railway system’s further development.As the core component of the high-speed train running part,the axle box bearings have strict geometric accuracy requirements.However,due to the alternating stress acts on each rigid contact surface,it is easy to cause bearing failure failures,which affect the safety of the train.The real-time condition monitoring of the axle box bearings can be realized by installing a test sensor on the outside of the axle box to collect vibration signals,but the collected signals contain interference components such as background noise.Therefore,extracting the bearing fault information covered by interference from the measured signal is a key step to realize the fault diagnosis of the axle box bearings of the high-speed trains.The empirical wavelet transform has a complete mathematical calculation framework and has good adaptability to nonlinear and non-stationary signals.Aiming at the shortcomings of the frequency band division of the traditional empirical wavelet transform,an empirical wavelet transform optimization idea based on the frequency spectrum trend is proposed.The empirical mode decomposition can extract the low frequency components of the frequency spectrum,and the mathematical morphology filter can extract the fault center frequency position of the frequency spectrum.Based on this,two methods for estimating the spectrum trend are proposed,and two optimized method,EEWT and MEWT,are obtained.The simulation signal test verifies that both methods can realize the effective diagnosis of bearing faults,and can greatly improve the boundary concentration and frequency band overdivision.According to the fault distribution characteristics of the bearing signal envelope spectrum,the harmonics identification index is proposed as the evaluation parameter of the frequency band fault information,and it is applied to the effective frequency band selecting of the actual engineering signal,which improves the practicability of the EEWT method and the MEWT method.The harmonics diagnosis capabilities of the two algorithms are verified through bench test signals and real vehicle signals.The results show that both methods are suitable for single fault and compound fault feature extraction of high-speed train axle box bearings.