Rolling bearing fault diagnosis via STFT and improved instantaneous frequency estimation method

摘要

Bearing vibration signals exhibit instantaneous modulation features under variable rotating frequency, making it difficult to identify the characteristic frequency components. As such, a novel rolling bearing fault diagnosis method via short time Fourier transform (STFT) and improved instantaneous frequency estimation algorithm is proposed. The novelty is that it can convert all the trajectories of instantaneous components, e.g., fault characteristic frequency (FCF) and modulation rotating frequency, to linear path in the time-frequency domain. First, a band pass filter is applied to separate the optimal frequency band which is determined by fast kurtogram. Second, the envelope time-frequency representation (TFR) is obtained by jointly adopting Hilbert transform and STFT to the filtered signal. Next, the instantaneous fault characteristic frequency (IFCF) is extracted from the filtered TFR based on improved instantaneous frequency estimation algorithm. Then, the frequency representation is calculated by dealing with the TFR of the envelope. Finally, the proportion of FCF to rotational frequency on the frequency representation is computed and then compared with the theoretical characteristic coefficient. The method is evaluated via the experimental data. The result demonstrates that bearing fault pattern can be identified via the proposed method.