An improved complementary ensemble empirical mode decomposition method and its application in rolling bearing fault diagnosis

Although ensemble empirical mode decomposition (EEMD) can suppress the modal confusion phenomenon in the EMD method to a certain extent, the added white noise cannot be completely neutralized. The complementary EEMD (CEEMD) adds white noise with opposite signs to the analysis signal in pairs, which greatly reduces the reconstruction error. Aiming at the problems of modal confusion and the difficulty in accurately extracting fault features of rolling bearings, a CPCEEMD (CEEMD-PE-CEEMD) bearing fault diagnosis method is proposed, which fully combines the CEEMD algorithm and the advantages of signal randomness detection based on permutation entropy (PE). After the abnormal components of the CEEMD are detected by permutation entropy, CEEMD of the remaining signals is conducted directly. Intrinsic mode function (IMF) components with large correlation coefficients are selected for Hilbert envelope spectrum analysis, and fault features are extracted from the envelope diagram. By analyzing the simulation signal and the measured bearing signal, the results show that the proposed method has a good decomposition effect, results in a certain inhibition effect on the modal confusion in the EMD process, and can effectively extract the characteristic information of the rolling bearing fault signal, which is feasible. (C) 2021 Elsevier Inc. All rights reserved.

Automated summary

The study introduced a CPCEEMD bearing fault diagnosis method that combines the advantages of the CEEMD algorithm and signal randomness detection based on permutation entropy to address the issues of modal confusion and difficulty in fault feature extraction. The proposed method showed good decomposition effect, certain inhibition effect on modal confusion in the EMD process, and effective extraction of characteristic information from rolling bearing fault signals in simulation and measured signals.