A tacho-free technique for bearing fault diagnosis via the oscillatory behaviors based signal decomposition and generalized demodulation

Order tracking is one of the most widely used techniques for machine condition monitoring under variable speed. This method requires the shaft rotating speed beforehand. However, a tacho is not always allowed to be installed in the industry and furthermore the signal components associated to shaft rotating speed cannot be directly extracted from vibration signals for a bearing case. As such, this paper proposes a tacho-free method for bearing fault diagnosis based on the joint application of oscillatory behaviors based signal decomposition (OBSD) and generalized demodulation (GD) to address the problems. The OBSD method, which is capable of distinguishing impulses (low-oscillation component) from sustained oscillation signal components (high-oscillation component) based on the signal oscillatory behaviors, is firstly used to extract the incipient fault-induced impulses. The extracted impulsive-signal together with the short time Fourier transform (STFT) leads to a TFR dominated by fault-revealing trend lines, from which the instantaneous fault characteristic frequency (IFCF) can be estimated. Guided by the extracted IFCF, the GD is subsequently applied to mapping the trajectory of time-frequency component of interest into a linear path parallel to the time axis. The shaft rotating frequency and IFCF as well as their harmonics can then be identified on the order spectra which assemble all spectra of generalized demodulated signals. The experimental result validates the effectiveness of the proposed method for bearing fault diagnosis.