Rolling element bearing fault diagnosis via fault characteristic order (FCO) analysis

• We report a fault characteristic order (FCO) approach to bearing fault diagnosis under variable speed conditions.
• An amplitude-sum based peak search method is proposed to extract instantaneous fault characteristic frequency (IFCF).
• A fault phase angle (FPA) concept is proposed.
• An IFCF-based resampling algorithm is developed to convert the non-stationary signal into stationary one in the FPA domain.
• The FPA representation is transformed into FCO spectrum for diagnosis.

Order tracking based on time–frequency representation (TFR) is one of the most effective methods for gear fault detection under time-varying rotational speed without using a tachometer. However, for a rolling element bearing, the signal components related to rotational speed usually cannot be directly extracted from the TFR. As such, we propose a new method to solve this problem. This method consists of four main steps: (a) signal filtering via fast spectral kurtosis (SK) analysis – this together with the short time Fourier transform (STFT) leads to a TFR of the filtered signal with clear fault-revealing trend lines, (b) extraction of instantaneous fault characteristic frequency (IFCF) from the TFR using an amplitude-sum based spectral peak search algorithm, (c) signal resampling based on the extracted IFCF to convert the non-stationary time-domain signal into the stationary fault phase angle (FPA) domain signal, and (d) transform of the FPA domain signal into the domain of the fault characteristic order (FCO) and identification of fault type from the FCO spectrum. The effectiveness of the proposed method has been validated by both simulated and experimental bearing vibration signals.