Residual life prediction of lithium battery based on hybrid model of WOA–VMD and attention mechanism

Abstract

Abstract In order to improve the long-term reliability of lithium battery and ensure the safe operation of the system, a forecasting method combining whale optimization algorithm (WOA), variational mode decomposition (VMD) and the Gated Recurrent Unit model with Attention mechanism (GRU-Attention) was proposed. WOA–VMD was used to decompose the battery capacity data into the intrinsic mode component and residual component, which were, respectively, predicted by GRU-Attention, and finally effectively integrated to obtain accurate capacity prediction results. Through the experiments of the Li-ion battery data set of the Advanced Life Cycle Engineering Center of the University of Maryland, the proposed Li-ion battery combination prediction model has high prediction accuracy and stability. This method is a novel combination of advanced technologies and algorithms, and its effectiveness in improving the long-term reliability of lithium batteries has been verified by experiments. This research is of great significance for promoting the development and application of lithium battery technology.