Online Tracking of Bearing Wear using Wavelet Packet Transform and Hidden Markov Models

Abstract

In this work, a new method was developed based on wavelet packet decomposition and hidden Markov modeling (HMM) for monitoring bearing faults. In this new scheme, vibration signals were decomposed into wavelet packets and the node energies of the decomposition were used as features. An HMM was built to model the normal bearing operating condition based on the features extracted from normal bearing vibration signals. The probabilities of this HMM were then used to monitor the bearing condition. Experimental data collected from a bearing accelerated life test clearly showed this new method's superiority over classical methods