Short-time smoothness spectrum: A novel demodulation method for bearing fault diagnosis

Demodulation of the measured signal to detect faults has been a widely used prominent technique in fault diagnosis of rotating machines. This paper proposes a novel demodulation method called short-time smoothness spectrum (STSS) for bearing fault diagnosis. With a sliding window on the raw signal, the smoothness index (SI) is calculated for each short-time signal to detect the impulsive performance. The proposed STSS can trace and check the change of resonant frequency band along with the time varying, which is able to effectively depict the fault signature. In the proposed study, the STSS was calculated on the spectral reconstruction sequences based on the STFT, which delivers a smoothness spectrogram. This provides a novel idea for solving the demodulation issue to defect the periodic faults. Simulation and comparisons show effectiveness of the proposed method in signal demodulation for bearing fault diagnosis.