A spectral coherence cyclic periodic index optimization-gram for bearing fault diagnosis

The identification of optimal frequency band (OFB) sensitive to faults is crucial for bearing fault diagnosis. In this paper, a spectral coherence cyclic periodic index (SCCP) optimization-gram (SCCPgram) is proposed. First, the vibration signal is mapped into a two-dimensional plane containing spectral and cyclic frequencies by spectral coherence theory. Second, a novel feature indicator SCCP is developed, which fully explores the merits of autocorrelation in revealing the cyclic information hidden in noise, and converts the periodic pattern of cyclic frequency direction in spectral coherence into a visual representation, thus unveiling the fault components. Then, considering SCCP as a metric, the SCCPgram is designed based on 1/3-binary tree filter bank to select the OFB. Finally, the spectral coherence is integrated over the OFB to generate improved envelope spectrum. The proposed method is validated by simulation and experimental data, and results show that SCCPgram has better performance compared with other state-of-art methods.