Synchronous analysis of cyclo-non-stationary signals: A comprehensive study with aeronautic applications

• Deterministic/random separation of mechanical signals. • Extension of the synchronous average to non-stationary regime. • Global and local synchronous fitting methods are investigated. Synchronous components are common in measured mechanical signals. They usually consist of a periodic waveform generated by a particular rotating organ. The accurate estimation of such components is of high importance in vibration-based health monitoring, offering the possibility to isolate and investigate the corresponding organ contribution. It is also vital for residual analysis in case where the component of interest is random. Traditionally, the synchronous average was a widely used tool for this purpose, providing almost optimal estimation performance and simplicity. However, the applicability of this technique is confined to the case where signals are recorded under (quasi) stationary regimes, which is restrictive for many applications like aeronautics. This paper studies two extensions of this techniques, namely the global and the local synchronous fitting. Both of these techniques aim to estimate, for a given location in the cycle, an estimation of the mean based on a polynomial fitting. Whereas the former performs this estimation through a global polynomial fit, the latter uses a local approach based on the well-known “Savitzky-Golay filter”. Both techniques are presented, and the resulting filter of the local synchronous fitting operation is provided. The performances of these techniques are investigated through simulations, considering the parameters of both approaches. These methods are then successfully tested in three challenging aeronautic applications. The newly proposed local approach has demonstrated superior performances as compared to the global one. The Matlab code of both techniques is also provided.