Vibration data pre-processing techniques for rolling element bearing fault detection

Research output: Chapter in Book/Report/Conference proceedingConference paper

6 Citations (Scopus)

Abstract

Real world vibration measurements often contain a lot of signal components originating from different machine elements. Detecting and recognizing the signal of a single element is thus a complicated endeavor because of the masking presence of other element’s signals in the measured data. Of particular interest to the industry is the early detection of faults in rolling element bearings. Normally the signal strength of a faulty bearing is however weaker than the ones of more prominent signals coming from other mechanical elements like gearboxes, rotating shafts and/or mechanical loads. These stronger deterministic signals form discrete components in the frequency domain. A number of techniques have been developed to filter out these components in order to detect the fault signals of the bearing elements more easily, e.g. time-synchronous averaging (TSA), self-adaptive noise cancellation (SANC), discrete/random separation (DRS) and
cepstral editing procedure (CEP). The latter technique has proven to be particularly promising in filtering out deterministic harmonic content in the measured signal while enhancing the bearing fault signals that can be treated as cyclostationary. If the shaft speed variation is limited,
the cepstral editing procedure can remove selected discrete frequency components without order tracking. In the cepstral domain the narrow impulses from the deterministic components can be removed or liftered by applying a notch lifter on the cepstrum. Since manually choosing which parts to lifter is arduous, an automated CEP method (ACEP) is developed that detects the cepstral peaks without user interaction and that tries to outperform the traditional cepstral editing procedure. In a general condition monitoring scheme, going from measured signal to fault detection &
diagnosis, this ACEP method is an interesting precursor to other analysis techniques like spectral kurtosis and envelope analysis.
Original languageEnglish
Title of host publicationProceedings of the international conference on sound and vibration
Pages307-314
Number of pages8
Publication statusPublished - 11 Jul 2016
EventInternational congress on Sound and Vibration - Greece, Athens, Greece
Duration: 10 Jul 201614 Jul 2016

Conference

ConferenceInternational congress on Sound and Vibration
Abbreviated titleICSV
Country/TerritoryGreece
CityAthens
Period10/07/1614/07/16

Fingerprint

Dive into the research topics of 'Vibration data pre-processing techniques for rolling element bearing fault detection'. Together they form a unique fingerprint.

Cite this