Rolling element bearings' damage is the main cause for unexpected breakdown in rotating machinery. Therefore there is a continuous industrial interest on condition monitoring of bearings targeting towards the development and proposal of robust diagnostic techniques which can detect accurately, robustly and early the generation of the fault. On the other hand, industry is not interested only in the proper damage detection and identification of faults but is mainly targeting towards the robust estimation of the Remaining Useful Life (RUL) of machine elements. The proper estimation of the RUL could be linked directly with the maintenance planning and warehouse organization providing immediately profits in terms of employees health and safety, environmental protection and continuous production. A plethora of diagnostics and prognostics indicators have been proposed during the last decade focusing towards the accurate representation and tracking of the health state of bearings and other machine elements. However, in certain cases (e.g. nonstationary operating conditions), the classic techniques for bearings prognostics (e.g. statistical analysis, frequency analysis and time-frequency analysis) underperform due to the high noise influence or the high machines' complexity. Therefore the classical diagnostics indicators may identify the fault quite late and fail to identify properly the RUL.

In this paper, prognostics indicators based on the measurement of disorder (e.g. entropy) are used in order to track the degradation severity of the machinery. The Spectral Entropy (SE) [1], the Envelope Spectral Entropy (ESE) and the Spectral Negentropy(SN) [2] are used as prognostics indicators in parallel with the Spectra Kurtosis [3] and the RMS. The indicators are estimated on the well-known bearing dataset Prognostia and Particle filtering is used in order to estimate the Remaining Useful Life of bearings. The prognostics indicators are evaluated and compared based on four main criteria: the Monotonicity, the Trendability and the Prognosability as well as on the estimated RUL.

Keywords: Condition Monitoring, Feature extraction, Spectral Entropy, Prognostics, Remaining Useful Life

References:

[1] Pan, Y. N., J. Chen, and X. L. Li. "Spectral Entropy: A Complementary Index for Rolling Element Bearing Performance Degradation Assessment." Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science. Vol. 223, Issue 5, 2009, pp. 1223–1231.

[2] Antoni, J. " The infogram: Entropic evidence of the signature of repetitive transients." Mechanical Systems and Signal Processing. Vol. 74, 1 June 2016, pp. 73–94.

[3] Antoni, J. " The spectral kurtosis: a useful tool for characterizing non-stationary signals." Mechanical Systems and Signal Processing. Vol. 20, Issue 2, February 2006, pp. 282–307.