A non-linear time–frequency tool for machinery fault diagnosis under varying speed condition

Vibration signals acquired from rotating machines operating under varying speed condition often exhibit strong frequency modulation. Conventional time–frequency analysis techniques cannot accurately capture the rapid changing dynamic of such signals where the time–frequency representation often suffers from the diffused energy problem. A new technique binding the demodulation technique and synchroextracting transform (SET) is thus proposed in this paper to overcome this problem. A theoretical analysis is also presented in the study to show the underlying mechanism that the SET technique has certain limitations in dealing with strong time-varying signals, whereas the proposed technique can effectively overcome such limitations. Several numerical examples and two sets of bearing experimental data are used to verify the effectiveness of the proposed technique in the analysis of time-varying signals and for bearing fault diagnosis.