Synchrosqueezing transform based general linear chirplet transform of instantaneous rotational frequency estimation for rotating machines with speed variations

Instantaneous rotational frequency is a key point symptom in the rotating machinery condition monitoring. How to acquire an accuracy instantaneous rotational frequency is a hot topic in research areas. Time-frequency analysis can be considered as a positive tool to characterize the variable speed tendency. However, limited by Heisenberg uncertainty principle or introduced cross-terms, the basically time-frequency analysis methods often generate smearing time-frequency representation, which seriously weaken its in industrial acquirement. During this paper, Synchrosqueezing transform based general linear chirplet transform method is proposed to characterize variable speed tendency of the non-stationary signal in strong time-varying conditions. The described method by means of the principle of general linear chirplet transform and synchrosqueezing transform, which could break the limitation of linear time-frequency analysis and enhance time-frequency resolution. Numerical and experimental investigations are provided to testify the high performance in addressing time-varying signal by the novelty method.

I. Introduction

Signal processing methodologies have the ability to reveal the interested characteristic of mono- and multi-component of rotating machinery [1]-[6]. Fast Fourier transform (FFT) as the most famous and classical method to extract fault characteristic frequency (FCF) in stationary conditions [7]. However, the stationary conditions cannot meet the real industrial; thus the mentioned methods do not adequate to analysis real data.