Indicator-assisted Sparse Morphological Decomposition for High-speed Bearing Diagnosis

This paper proposes an indicator-assisted sparse morphological decomposition method in according with the morphological characteristics of the signal, which can effectively separate discrete frequencies and accurately identify impulses for the high-speed bearing signal. The spectral flatness is introduced to constrain the sparsity of Fourier dictionary for separating the discrete frequencies. Then, the kurtosis is applied to constrain the sparsity of the convolutional dictionary for identify fault impulses. The performance of the proposed method is verified through an experiment of the high-speed bearing.