Weak faults diagnosis of traction motor bearings based on SR-MED-EEMD method

Abstract Rolling bearings play a crucial role in ensuring the security and reliable operation of high-speed trains (HST) as a vital component of their traction system. In response to the difficulty in extracting early fault features of high-speed train traction motor bearings under strong background noise, a method based on sparse representation (SR), minimum entropy deconvolution (MED), and ensemble empirical mode decomposition (EEMD) is proposed. The method reduces the dimensionality of complex signals through SR and then filters the sparse signals through MED to improve the fault signal pulse amplitude. The processed signals are decomposed by using EEMD to remove noise components, further improving the peak signal-to-noise ratio of fault characteristic frequency. The method’s efficacy in detecting faults within rolling bearings has been confirmed through simulation experiments and demonstrated its high precision.