Periodically correlated random processes: Application in early diagnostics of mechanical systems

The covariance and spectral characteristics of periodically correlated random processes (PCRP) are used to describe the state of rotary mechanical systems and in their fault detection. The methods for estimation of mean function, covariance function, instantaneous spectral density and their Fourier coefficients for a given class of non-stationary random processes on the basis of experimental data, namely: the synchronous averaging, component, least squares method and linear filtration methods are considered. The first and second order periodicity detection methods are used for vibration signals analysis. A method for mechanical system fault identification and classification based on a harmonic series representation is developed. Examples of fault detection in rolling/sliding bearings and gearboxes are given.