Estimation of the synchronous average under varying rotating speed condition for vibration monitoring
Résumé
The decomposition of a vibration signal into deterministic components characterized by first order cyclostationarity and random components characterized by second order cyclostationarity is of great value in condition monitoring. This issue has found solutions only when the rotation speed is stationary. However under varying rotation speed, the recorded signals are no longer cyclostationary. Indeed the vibration amplitude is related to the varying operating speed. In such a case, resampling the signals in the angular domain transforms forces them to become cyclo-non-stationary. This paper introduces a new approach based on a decomposition of the vibration signal into a set of cyclo-non-stationary components that form a basis of functions. Each basis function captures the long-term evolution of the synchronous average over consecutive cycles. We demonstrate the efficiency of this approach to extract deterministic components under varying rotating speed condition.