Bearing spall is the foremost cause of failure in rotating machineries, which can lead to catastrophic failure when it is not repaired properly. Many researches have been studied for the bearing prognostics that predict bearing's remaining cycle before the maintenance, but they largely depend on each case, and there are a lot of challenges to be solved for practical prognostics. In this paper, a new method based on entropy changes at specific frequencies is proposed for more robust results. Degradation feature is extracted from decomposed signals into frequency domain, and important attributes to predict the remaining cycle are found. This method is demonstrated using the real test data provided by FEMTO-ST institute. The results show that bearings can be used 56～78% of their whole life in average.
展开▼
