Online tracking of bearing wear using wavelet packet decomposition and probabilistic modeling: A method for bearing prognostics

Bearings are common and vital elements in rotating machinery. By tracking the condition of a bearing, unscheduled machinery outages and costly damage caused by a bearing failure can be avoided. In this paper, we developed a new scheme based on wavelet packet decomposition and hidden Markov modeling (HMM) for tracking the severity of bearing faults. In this scheme, vibration signals were decomposed into wavelet packets and the node energies of the decomposition tree were used as features. Based on the features extracted from normal bearing vibration signals, an HMM was trained to model the normal bearing operating condition. The probabilities of this HMM were then used to track the condition of the bearing. Experimental data collected from a bearing accelerated life test showed that unlike many of the other commonly used trend parameters whose distinguishing features diminished to normal bearing-like levels as the damage grew, the probabilities of the normal bearing HMM kept decreasing as the bearing damage progressed toward bearing failure. As the bearing approached the end of its life (10% of remaining life), the HMM probabilities dropped dramatically signaling severe damage and imminent bearing failure.