Optimised Spectral Kurtosis for bearing diagnostics under electromagnetic interference. (15th June 2016)

• Record Type: Journal Article
• Authors: Smith, Wade A.; Fan, Zhiqi; Peng, Zhongxiao; Li, Huaizhong; Randall, Robert B.
• Is Part Of: Mechanical systems and signal processing. Volume 75(2016)
• Page Start: 371
• Page End: 394
• Publication Date: 2016-06-15
• Subjects: Rolling element bearing -- Bearing diagnostics -- Demodulation band -- Envelope analysis -- Electromagnetic interference; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2015.12.034
• Languages: English
• ISSNs: 0888-3270

Abstract

Abstract: The selection of the optimal demodulation frequency band is a significant step in bearing fault diagnosis because it determines whether the fault information can be extracted from the demodulated signal via envelope analysis. Two well-known methods for selecting the demodulation band are the Fast Kurtogram, based on the kurtosis of the filtered time signal, and the Protrugram, which uses the kurtosis of the envelope (amplitude) spectrum. Although these two methods have been successfully applied in many cases, the authors have observed that they may fail in specific environments, such as in the presence of electromagnetic interference (EMI) or other impulsive masking signals. In this paper, a simple spectral kurtosis-based approach is proposed for selecting the best demodulation band to extract bearing fault-related impulsive content from vibration signals contaminated with strong EMI. The method is applied to vibration signals obtained from a planetary gearbox test rig with planet bearings seeded with inner and outer race faults. Results from the Fast Kurtogram and Protrugram methods are also included for comparison. The proposed approach is found to exhibit superior diagnostic performance in the presence of intense EMI. Another contribution of the paper is to introduce and explain the issue of EMI to the condition monitoring community. The paper outlines the characteristics of EMI arising from widely-used variable frequency drives, and these characteristics are used to simulate an EMI-contaminated vibration signal to further test the performance of the proposed approach. Although EMI has been acknowledged as a serious problem in many industrial cases, there have been very few studies showing its adverse effects on machine diagnostics. It is important for analysts to be able to identify EMI in measured vibration signals, lest it interfere with the analysis undertaken. Highlights: We introduce the problem of contamination of vibration signals with EMI. The nature of such interference is clearly explained. The performance of established diagnostic techniques is shown to be affected by EMI. An approach based on traditional Spectral Kurtosis gives improved diagnostic results. Method is applied to measured and simulated signals with EMI to show improved results.