Singular Spectrum Decomposition and Iteration Sparsity-Oriented Morphological Demodulation for Bearing Fault Diagnosis

The singular spectrum decomposition (SSD) can decompose the original signal into several singular spectrum components (SSC). Therefore, the optimal SSC contains more fault information. The optimized characteristic frequency intensity coefficient is used to select the SSC. After that, to further strengthen the impulses in SSC, the sparsity operation and morphological analysis are combined in this paper. The sparsity can suppress the noise in the signal and enlarge the impulses. A loop is constructed and the kurtosis is used to select the optimal sparsity analysis result. The closing morphological analysis is used to supplement the sparsity operation and make the impulses more obvious. At the end, the slice bispectrum is used to remove the residual Gaussian noise. The simulation and experimental cases (including sound and vibration signals) from the bearing test rig are used to demonstrate the feasibility and effectiveness of the proposed method.