This article presents a processing technique called “Blind Source Separation” initially developed for telecommunication applications and now applied to vibration monitoring of mechanical systems. Experimental results illustrate the potential of blind source separation as a pre-treatment to free the signal from its environment (denoising), so as to improve fault detection and diagnosis. An example of bearing faults detection on an experimental test bench is presented.
Get full access to this article
View all access options for this article.
References
1.
CapdevielleV.ServièreC.LacoumeJL, 1996, Proceedings of European Signal Processing Conference ′96 (EUSIPCO ′96), 2085–2088, “Blind separation of wide band sources: application to rotating machine signals”.
2.
GelleG.ColasM.DelaunayG., 1999, Proceedings of ICA ′99, 109–114, “Separation of convolutive mixtures of harmonic signals with a temporal approach: application to rotating machine monitoring”.
3.
GelleG.ColasM.DelaunayG., 2000, Mechanical System and Signal Processing., vol.14(3), 427–442 “Blind Sources Separation applied to Rotating Machines Monitoring by Acoustical and Vibrations Analysis”.
4.
CardosoJ. F., 1998, Proceedings of the IEEE, Vol.86, 2009–2025, “Blind signal separation: statistical principle”.
5.
ThiNguyen H.L.JuttenC., 1995, Signal Processing, Vol. 45, 209–229, “Blind source separation for convolutive mixture”.
6.
BraunS., 1986, “Mechanical signature analysis: theory and applications”, Academic Press.
