Maximum correntropy criterion-based blind deconvolution and its application for bearing fault detection. (15th March 2022)

• Record Type: Journal Article
• Authors: Liu, Chuliang; Tan, Jianping; Huang, Zhonghe
• Is Part Of: Measurement. Volume 191(2022)
• Publication Date: 2022-03-15
• Subjects: Bearing fault diagnosis -- Blind deconvolution -- Autocorrentropy -- Cross correntropy -- Maximum correntropy criterion; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2022.110740
• Languages: English
• ISSNs: 0263-2241

Abstract

Abstract: The distortion of signal caused by transmission path always brings great difficulties to feature extraction for bearing fault diagnosis. It is thus an important work to eliminate the influence of transmission path and recover impulsive features. A novel method called maximum correntropy criterion-based blind deconvolution (MCCBD) is presented in this paper and the strong analytical connection which holds for MOMEDA and MCCBD is revealed. Maximum correntropy criterion (MCC) is used as an objective function to reduce the influence of outliers and the coefficients of the optimal deconvolution filter can be obtained. Moreover, the proposed method adopts an autocorrentropy-based strategy to automatically evaluate the impulse period, hence it can enhance the fault impulses buried by noise without any prior knowledge. Finally, the superiority of MCCBD is exhibited by comparing with some existing methods on the simulated signal and two experimental datasets. Highlights: A deconvolution method is proposed under the framework of adaptive systems learning. The strong connection which holds for MOMEDA and MCCBD is revealed. A new fault period detection method based on autocorrentropy is proposed. The effectiveness of MCCBD is verified by simulated signal and experimental data.