Fault feature extraction of rolling bearings based on TVF-EMD and SSA noise reduction

Aiming at the difficulty of extracting weak early fault features of rolling bearings, a novel method based on time-varying filtering empirical mode decomposition (TVF-EMD) adaptive fusion of mode components and singular spectrum analysis (SSA) noise reduction was proposed. Firstly, in order to reduce the nonlinearity and non-stationarity of the fault signal, the bearing signal is decomposed into a series of intrinsic mode functions (IMF) by TVF-EMD. Secondly, in order to overcome the deficiency of excessive IMF components after TVF-EMD decomposition, a composite sensitive modal decision factor (CSMDF) was constructed using the curtosis, complexity and fractal dimension of IMF. The CSMDF was used to arrange the IMF components in descending order, and the IMF components were fused successively according to the increasing principle of the composite sensitive modal decision factor. Finally, the optimized fusion component is denoised by SSA, and the denoised component is analyzed by Hilbert envelope spectrum to realize the feature extraction of bearing faults. The performance of the proposed method is analyzed experimentally by simulating fault signals and two measured fault signals. The experimental results show that the proposed method has good ability of sensitive feature screening and fusion and noise reduction, and can extract the early bearing fault characteristics more accurately, and realize the accurate identification of bearing fault types under noisy environment.