Performance assessment of a morphological index in fault prediction and trending of defective rolling element bearings

Frequency domain based signal processing methods have been shown to present a quite effective behaviour in the detection of defects, when applied to the analysis of vibration signals, resulting from rolling element bearings. However, these methods typically require some complex and sophisticated analysis, which renders their application cumbersome for applications requiring unskilled personnel or automated fault detection and trending. Parallel, a number of traditional methods exist, such as the root mean square (RMS), the crest factor (CF), the kurtosis (KU), the impulse factor (IF) and the shape factor (SF), requiring only direct processing in the time domain. Alternatively to these methods, a morphological index (MI) for processing vibration signals has been proposed, addressing the issues of how to quantify the shape and the size of the signals directly in the time domain. In this paper, based on a model for the dynamic behaviour of defective rolling bearings, the sensitivity of the MI is assessed, compared to the previous five traditional time domain indices, with respect to the effect of the added noise, the impulse repetition period, as well as the natural period and the damping ratio of the excited resonance, both in the case of an inner and an outer race defect. The results clearly indicate the superiority of the MI over all the other time domain indices compared. This fact is then further verified in three different cases from industrial installations, presenting fault trending analysis of bearings under various defects.