Condition based monitoring by vibration sensors is a widely spread technique for monitoring the status and condition ("health") of rotating machines. In most cases the monitoring is based on the ability to isolate specific elements of the vibration signals, generated by the different rotating components.

The generated signals are propagating through various transmissions paths of the machine, that distort the original signals, hence affect the assessment of the machine condition. While these effects usually ignored by most vibration analysis techniques, first steps towards mitigating this problem have taken place during the last years. These techniques used pre-whitening methods, which usually served to separate the signal from its background as well as to reduce the transmission path effects.

In this paper we aim to go a step further in transmission path restoring through deepening our understanding of their effects on the vibration signals. We start by comparing three main pre-whitening methods: liftering low quefrencies at the Cepstrum (Cepstrum-liftering), adaptive clutter separation (ACS), and pre-whitening using auto-regressive (AR) models. First, it is shown that signal's pre-whitening by AR model has large errors and therefore is less adequate for transmission path restoration process. A theoretical approach to adjust parameters of ACS and Cepstrum-liftering techniques and examine them through quantitative methods is proposed.

The ability to utilize ACS and Cepstrum-liftering based techniques to restore the originals signals out of the measured signal is explored. Finally, the restoration quality is tested by simulation outcomes and real measured data.

References:

[1] R. Klein, Comparison of methods for separating excitation sources in rotating machinery, The International Conference SURVISHNO 7, 2013.

[2] R.B Randall, B. Peeters, J. Antoni, S. Manzato, New cepstral methods of signal pre-processing for operational modal analysis, ISMA2012-USD2012, Leuven, Belgium, pp. 775-764, 2012.

[3] R.B. Randall, W.A Smith, Cepstrum-based operational modal analysis revisited: A discussion on pole-zero models and the regeneration of frequency response functions, Article in Mechanical Systems and Signal Processing (MSSP), 2016.

[4] R.B. Randall, W.A Smith, New cepstral techniques for operational modal analysis, Conference Paper, 2017.

[5] N. Sawalhi and R.B. Randall, Spectral Kurtosis Enhancement using Autoregressive Models, ACAM2005 conference, Melbourne, 2005.

[6] R. Castiglione, J. Antoni, L. Garibaldi, Separation and identification of structural modes in largely underdetermined scenarios using frequency banding, Journal of Sound and Vibration 414 pp. 192-217, 2018.

Keywords: vibration signal, transmission-path, pre-whitening, adaptive clatter separation (ACS), Cepstrum-liftering, AR model, restoration process.