A Remaining Useful Life Prediction Method in the Early Stage of Stochastic Degradation Process

Abstract

In order to achieve more accurate predicted RUL in the early stage of degradation, a novel remaining useful life (RUL) prediction method for the stochastic degradation process is proposed. Technically, modeling the degradation process as a Wiener process (WP) whose drift increment is a weighted sum of kernel functions can flexibly depict the nonlinear degradation trend. Introducing a long short term memory (LSTM) network can capture the long-term dependencies of the offline experimental and online observed degradation data to forecast the future degradation increment. Then, based on the degradation model, a numerical approximate distribution of the RUL is derived to quantify the uncertainty of the predicted RUL. Finally, a practical case study of lithium-ion batteries is provided to demonstrate the high accuracy of the proposed methods for RUL prediction especially in the early stage of degradation.