Remaining Useful Life Prediction Considering Operating Condition Change Based on Regression and Empirical Mode Decomposition

This paper proposes remaining useful life prediction method considering the operating conditions and instantaneous noise of the lithium-ion battery. With the change of the operating conditions and instantaneous noise in training data, the regression model cannot accurately predict state of health degradation. Thus, proposed method preprocesses training data by empirical mode decomposition in order to eliminate temporary noise. Moreover, training data reset is performed to extract the latest tendency by excluding the data before the change in operating condition based on regression results. The feasibility of the proposed algorithm was verified through the cycling dataset, and the result shows that the accuracy of the RUL estimation can be improved by proposed method than traditional regression.