Remaining Life-span Prediction Based on Adaptive Nonlinear Wiener Process
As a core technology for both prognostics and health management,accurate remaining life-span prediction is of great significance to enhance the safety and reliability of the system.In actual engineering applications,individual differences usually exist among similar products,so that the learned model or parameters can't accurately fit the degradation process of the new equipment by the historical data of the similar products.To address this problem,an adaptive life-span prediction method based on a nonlinear Wiener degradation model is proposed.an adaptive nonlinear Wiener deg-radation model is established considering individual uncertainty and measurement uncertainty at the same,and the online upgrade of parameters is realized by Kalman filtering,expectation maximization algorithm,and Rauch-Tung-Striebel to mitigate the Markovian property of the Wiener process.More-over,the analytical expressions for the distribution probability density function of the remaining life-span are derived through time-space transformation.Both simulated degradation data and the C-MAPSS degradation dataset are employed for experimental validation.The experiment results show that the proposed adaptive nonlinear Wiener degradation model can improve the prediction accuracy by updating the model parameters online.
国家科技期刊平台
NSTL
万方数据
