Remaining useful life prediction of rolling bearing using fractal theory

The remaining life of rolling bearing is typically difficult to predict using existing methods due to the nonstationary characteristics of rolling bearing vibration signal. To improve the prediction accuracy, the parameters that can indicate the degradation trend of rolling bearing are defined and can improve remaining useful life (RUL) prediction using the method of empirical mode decomposition based on differential (DEMD) and gray Markov model. In comparison to the method of generalized mathematical morphology particle characteristics (GMMP), the results demonstrate that the fractal spectrum and its parameters can be used to predict the degradation trend of rolling bearing and to increase the accuracy of degradation trend estimation via gray Markov model. Mathematical morphology fractal spectrum parameters can reflect more slight fluctuations than GMMP method, and the root mean square percentage error (RMSE) is reduced by approximately 4%. New parameters and methods are introduced for RUL prediction of rolling bearing. (C) 2020 Elsevier Ltd. All rights reserved.