General synchroextracting chirplet transform: Application to the rotor rub-impact fault diagnosis

Abstract

The rub-impact fault between the rotor and the stator surface is a common and serious failure in rotating machinery. The vibration signals acquired during the rotating machinery operation will carry the time-varying characteristics of rapid fluctuating instantaneous frequency (IF) when the healthy condition of rotor system changes. The proposed method is termed as the general synchroextracting chirplet transform which is based on the classical chirplet transform and employs the second-order synchroextracting transform to concentrate the time-frequency representation (TFR). Moreover, an adaptive demodulation parameters determination method based on kurtosis is proposed. The proposed method is non-sensitivity to noise, allows for signal reconstruction, and retains certain processing ability for multicomponent signals. The effectiveness and practical value of the method are verified by the numerical signals and the vibration signals of the heavy oil catalytic cracking machine which has the rotor rub-impact fault.