A new angular resampling algorithm for the bearing fault diagnosis under the time-varying rotational speed

Abstract

Order tracking is one of the most effective algorithms to eliminate the effect of time-varying rotational speed on the rotary machines. However, this algorithm is not suitable for the faulty rolling bearing unless the peak time of the fault-induced impulse is set as zero which cannot be met in the real engineering. The traditional resampling process will cause uneven intervals between the adjacent impulse peaks in the angular domain and then affect the envelope analysis-based diagnosis result. To solve this problem, a new resampling algorithm with three parts is proposed: (a) linearly fitting the instantaneous rotational speed measured by the tachometer, (b) resampling the vibration signal from the time domain to the angular domain using the traditional method, (c) calculating the envelope deformation amount and then compensating the resampled result. The effectiveness of the proposed method has been validated by both the simulated and experimental bearing vibration signals.