The envelope-based cyclic periodogram. (June 2015)

• Record Type: Journal Article
• Authors: Borghesani, P.
• Is Part Of: Mechanical systems and signal processing. Volume 58/59(2015)
• Page Start: 245
• Page End: 270
• Publication Date: 2015-06
• Subjects: Cyclostationarity -- Cyclic power spectrum -- Envelope analysis -- Cyclic periodogram; Structural dynamics -- Periodicals; Vibration -- Periodicals; Constructions -- Dynamique -- Périodiques; Vibration -- Périodiques; Structural dynamics; Vibration; Periodicals; 621
• DOI: 10.1016/j.ymssp.2014.11.009
• Languages: English
• ISSNs: 0888-3270

Abstract

Abstract: Cyclostationary analysis has proven effective in identifying signal components for diagnostic purposes. A key descriptor in this framework is the cyclic power spectrum, traditionally estimated by the averaged cyclic periodogram and the smoothed cyclic periodogram. A lengthy debate about the best estimator finally found a solution in a cornerstone work by Antoni, who proposed a unified form for the two families, thus allowing a detailed statistical study of their properties. Since then, the focus of cyclostationary research has shifted towards algorithms, in terms of computational efficiency and simplicity of implementation. Traditional algorithms have proven computationally inefficient and the sophisticated "cyclostationary" definition of these estimators slowed their spread in the industry. The only attempt to increase the computational efficiency of cyclostationary estimators is represented by the cyclic modulation spectrum. This indicator exploits the relationship between cyclostationarity and envelope analysis. The link with envelope analysis allows a leap in computational efficiency and provides a "way in" for the understanding by industrial engineers. However, the new estimator lies outside the unified form described above and an unbiased version of the indicator has not been proposed. This paper will therefore extend the analysis of envelope-based estimators of the cyclic spectrum, proposing a new approach to include them in the unified form of cyclostationary estimators. This will enable the definition of a new envelope-based algorithm and the detailed analysis of the properties of the cyclic modulation spectrum. The computational efficiency of envelope-based algorithms will be also discussed quantitatively for the first time in comparison with the averaged cyclic periodogram. Finally, the algorithms will be validated with numerical and experimental examples. Highlights: Review of traditional estimators of the cyclic power spectrum. Derivation of a new estimator of the cyclic power spectrum based on envelope analysis. New generalised form for envelope-based estimators. Numerical and experimental (bearing signals) validation.