Real-time residual life prediction based on kernel density estimation considering abrupt change point detection

Abstract

The existence of abrupt change points in mechanical equipment degradation leads to inaccuracies in the prediction of its residual life. We propose a real-time residual life prediction method based on kernel density estimation (KDE) considering the influence of abrupt change points. First, a non-parametric cumulative sum method is used to detect abrupt change points in the degradation process. Then, integral mean square error is used to determine the abrupt change in the sample number that affects the accuracy of KDE. The weight coefficient is adaptively allocated according to the change in sample numbers relative to the minimum sample number before and after the abrupt change point in real-time monitoring. This method considers abrupt change states in the degradation process and uses KDE, which does not make model structure assumptions or parameter estimations for the degradation process. Finally, the effectiveness of this method is verified using the gear wear degradation data and compared with the real-time residual life prediction method based on KDE without considering the impact of the abrupt change point. The results show that this method can track the dynamic changes of a system more quickly and improve the accuracy of real-time residual life prediction.

Real-time remaining useful life kernel density estimation considering dynamic transition of degradation states

“…In table 2, comparing the Root Mean Square Error (RMSE) between the different models. With the increase of monitoring time, RMSE showed a decreasing trend, and with the increase of monitoring data [26]. The error between the predicted results of the model in this study and the actual remaining useful life was smaller, indicating that the remaining useful life value predicted by the model in this study was closer to the actual residual life value.…”

Multi-sensor information fusion-based prediction of remaining useful life of nonlinear Wiener process

“…To verify the correctness and effectiveness of the proposed method, numerical simulations were conducted using the nonlinear degradation model in equation (3). We set the drift coefficient 2.…”

“…In consideration of industrial big data trends, the identification of current health status according to the characteristics presented by monitoring big data is of great significance for the implementation of system health management [2]. To reduce the occurrence of system failures and decrease the cost of operation and maintenance, it is necessary to effectively monitor and evaluate the health status of a system and accurately predict its remaining useful life (RUL) based on the monitoring information [3]. RUL prediction, as the core content of prognostics and health management, has been garnering significant attention for research in recent years [4].…”

Real-time remaining useful life prediction based on adaptive kernel window width density

“…In prediction models with KDE, most models address the problem of underfitting, where there is little sample data, and overfitting,where there is a lot of sample data, when finding the sample probability density with kernel estimation based on a fixed window width. To address the shortcomings of fixed window widths, Zhang et al [21] introduced the integrated mean square error method into the selection of kernel estimation window widths and improved it to adaptively select window widths. Lin et al [22] proposed an approach for choosing the appropriate kernel bandwidth through the integration of the optimal bandwidth method based on the rule of thumb, the initial bandwidth selection method and the correct bandwidth selection of the state variables.…”