A new rolling bearing fault diagnosis method based on GFT impulse component extraction. (15th December 2016)

• Record Type: Journal Article
• Authors: Ou, Lu; Yu, Dejie; Yang, Hanjian
• Is Part Of: Mechanical systems and signal processing. Volume 81(2016)
• Page Start: 162
• Page End: 182
• Publication Date: 2016-12-15
• Subjects: Fault diagnosis -- Rolling bearing -- Graph Fourier Transform -- Impulse component -- Component extraction; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2016.03.009
• Languages: English
• ISSNs: 0888-3270

Abstract

Abstract: Periodic impulses are vital indicators of rolling bearing faults. The extraction of impulse components from rolling bearing vibration signals is of great importance for fault diagnosis. In this paper, vibration signals are taken as the path graph signals in a manifold perspective, and the Graph Fourier Transform (GFT) of vibration signals are investigated from the graph spectrum domain, which are both introduced into the vibration signal analysis. To extract the impulse components efficiently, a new adjacency weight matrix is defined, and then the GFT of the impulse component and harmonic component in the rolling bearing vibration signals are analyzed. Furthermore, as the GFT graph spectrum of the impulse component is mainly concentrated in the high-order region, a new rolling bearing fault diagnosis method based on GFT impulse component extraction is proposed. In the proposed method, the GFT of a vibration signal is firstly performed, and its graph spectrum coefficients in the high-order region are extracted to reconstruct different impulse components. Next, the Hilbert envelope spectra of these impulse components are calculated, and the envelope spectrum values at the fault characteristic frequency are arranged in order. Furthermore, the envelope spectrum with the maximum value at the fault characteristic frequency is selected as the final result, from which the rolling bearing fault can be diagnosed. Finally, an index KR, which is the product of the kurtosis and Hilbert envelope spectrum fault feature ratio of the extracted impulse component, is put forward to measure the performance of the proposed method. Simulations and experiments are utilized to demonstrate the feasibility and effectiveness of the proposed method. Highlights: The path graph is first introduced into the time series analysis in a manifold perspective. A new adjacency matrix weight is defined. A new rolling bearing fault diagnosis method based on GFT impulse component extraction is proposed.