A Hybrid Approach for RUL Prediction of Li-ion Batteries Using GRU-Based Additive Attention

Abstract

The reliability and the safety of battery-powered systems rely in major part on the accurate prediction of the Remaining Useful Life (RUL) of lithium-ion batteries. In this context, we propose a deep learning approach that combines Gated Recurrent Units (GRU) with additive attention mechanism to enhance the RUL prediction. The GRU network captures temporal dependencies in battery degradation data, while the additive attention mechanism selectively focusses on the most relevant temporal features, improving prediction accuracy. The model is validated using publicly available datasets. Experimental results prove that our GRU-Attention model outperforms traditional methods, providing a robust solution for battery health management.