Maximum Spectral Sparse Entropy Blind Deconvolution for Bearing Fault Diagnosis

Condition monitoring and diagnosis of rolling bearing is a formidable task since the weakened repetitive transient impulses that can represent early faults are usually overwhelmed by background noise and incidental disturbances in the transmission path. Blind deconvolution has gained wide attention because it effectively attenuates the transmission path. However, common deconvolution methods suffer from the serious problem of relying on prior knowledge and are difficult to effectively use in practical engineering. For example, the minimum entropy deconvolution (MED) is susceptible to accidental shock disturbances. The maximum correlation kurtosis deconvolution (MCKD)and multipoint optimal MED adjusted (MOMEDA) rely heavily on fault prior knowledge. The reason is that most of the current indicators suffer from low sensitivity, poor robustness, and reliance on prior knowledge. To address the above problems, the new concept is constructed in this article from the perspective of the squared envelope spectrum called spectral sparse entropy ratio (SSER). It can simultaneously consider the sparsity, complexity, and cyclostationary of the squared envelope spectrum and evaluate the healthy condition of the mechanical system. Based on this indicator, a new blind deconvolution method is established without prior fault knowledge called SSERBD. It can effectively attenuate the influence of transmission paths and successfully diagnose single and compound faults of rolling bearings. Simulation and experimental analysis show that the proposed method has promise in condition monitoring and diagnosis. Compared with common deconvolution methods and health indicators (HIs), it is verified that the proposed method is more advantageous.