A Remaining Useful Life Prediction approach for Li-ion batteries Based on Variational Mode Decomposition and SVM

Abstract

This paper proposes a remaining useful life (RUL) prediction model for li-ion batteries which combines variational mode decomposition (VMD) and support vector machine (SVM). First, the battery capacity degradation data is decomposed into the trend degradation sequence and other fluctuation sequences through VMD. Then build SVM regression models for different modes. Finally, all the prediction results are added to get the final RUL prediction value. The experiment verifies the effectiveness of the method through NASA lithium-ion battery aging test data, and compared with the single SVM regression model and the Gaussian process regression model, the VMD-SVM method obtained more accurate prediction results.