Cyclostationarity in Rotating Machine Vibrations

The vibrations of rotating machinery are produced by a combination of periodic and random processes due to the machine's rotation cycle and interaction with the real world. The combination of such components can give rise to signals which have periodically time-varying ensemble statistics and are best considered as cyclostationary. In this paper, second-order cyclic statistical methods are described and several applications of these to machine vibration analysis are introduced. The second-order techniques are compared with traditional (purely stationary) spectral analysis and synchronous averaging (first-order cyclic-analysis). This comparison is made for data collected from a rotating machine subjected to bearing faults and the applications are demonstrated. 1 Introduction The condition monitoring of rotating machinery often involves the use of signal processing algorithms on measured vibrations. As most systems will have to work in noisy industrial environments, some form of statist..