Nonlinear Squeezing Wavelet Transform for Rotor Rub Impact Fault Detection

Classical time-frequency analysis methods can depict time-frequency structure of non-stationary signals. However, they may have a shortage in extracting the weak components with small amplitude hidden in complex signals, because their time-frequency representation coefficients are proportional to the amplitude or energy of a signal. In this paper, we present a new data analysis method, called nonlinear squeezing wavelet transform, to extract the weak feature of highly oscillating frequency modulation for rotor rub-impact fault. The time-frequency representation of the proposed method is independent of the signal amplitude and only relevant to the signal phase, thus it can be used to characterize the time-frequency pattern of non-stationary multi-component signals, especially for weak components detection. The experiments on simulated signals verify the effectiveness of the proposed method in weak signal detection. Finally, the validity of this method is demonstrated to extracting the feature on a real rotor system with weak rub-impact fault.