An Adaptive Remaining Useful Life Prediction Method for Hybrid Ceramic Bearing

Ceramic bearings are quickly replacing conventional steel ball bearings in various fields and applications because they exhibit a service life three times longer than that of steel bearings. However, few studies have been reported on prognostics of ceramic bearings. Compared with the degradation process of steel bearing, the deterioration of ceramic bear contains more nonlinearity and uncertainty. The prediction of Remaining Useful Life (RUL) of ceramic bearings remains a new topic. One of the most challenging problems in the prognostic task is the lack of accurate prediction model. Particularly, when the prognostic model is trained from a data set that undergoes different degradation rate compared with the testing data set due to different load condition and service life for each individual component, the prediction often display large discrepancy. In this paper, a particle filtering based method for hybrid ceramic bearing RUL prediction is presented. An integrated adaptive model based on Recursive Least Square (RLS) is proposed to account the parameter variation of the prognostic model. The method is validated using real hybrid ceramic bearing run to failure test data. The validation results prove the effectiveness and high accuracy of the presented methodology.