Bearing degradation assessment based on entropy with time parameter and fuzzy c-means clustering

Bearings are one of the most crucial elements in rotating machine. The condition of bearings decides the operation of machine. Consequently, it is necessary to study the assessment of bearing degradation in order to develop condition-based maintenance. This paper improves an indicator based on entropy which is calculated by wavelet packet decomposition and auto-regressive model. By introducing time parameter, the indicator solves the problem of instability in the initial stage of operation and it is less influenced by the operational conditions. Then, fuzzy c-means clustering can evaluate the process of degradation. Moreover, it can provide the threshold adaptively and help to repair by unit replacement. To ensure the applicability, the data of this paper comes from two laboratories, FEMTO-ST Institute and Intelligent Maintenance System Center. The result indicates that the method is effective to assess bearing degradation process