A novel Switching Unscented Kalman Filter method for remaining useful life prediction of rolling bearing

A new Switching Unscented Kalman Filter (SUKF) algorithm is proposed. The corresponding state-space models for each kind of bearing operation state are established, and the UKF algorithm is incorporated into the Bayesian estimation method to calculate the probability of each state at every time and determine the most probable state. The prediction of Remaining Useful Life (RUL) can be carried out once the accelerated degradation stage is detected. In order to make the filtering results of Condition Monitoring (CM) data smoother and avoid misjudgment of status when the degradation speed is negative, the measurement error parameter is selected as the standard deviation of CM data in the degradation stage. The proposed method is applied into the bearing CM data from Intelligent System Maintenance Center of University of Cincinnati. Besides, it is also compared with the traditional Switching Kalman Filter (SKF) algorithm. The results show the effectiveness of the proposed method. (C) 2018 Elsevier Ltd. All rights reserved.