Remaining Useful Life Prediction of Lithium-Ion Batteries Based on Transformer-GRU Parallel Hybrid Network Model

The remaining useful life (RUL) prediction of lithium-ion batteries is a critical task in battery management systems. Accurate prediction is essential for battery failure warning and reliability assurance. However, capacity regeneration and noise in the data can lead to low prediction accuracy of battery life. To address this challenge, this paper proposes a Transformer and gated recurrent unit (GRU) parallel hybrid model for RUL prediction. First, the variational mode decomposition (VMD) algorithm is applied to decompose the raw capacity data, and extract multiple intrinsic mode functions (IMFs), which effectively remove the noise and improve the data quality. Then, the Transformer encoder is used to extract global features, while GRU captures local dynamic changes between adjacent time steps in the time series. The combination of both significantly enhances the model's prediction accuracy and robustness. Finally, experimental results show that the proposed model demonstrates outstanding accuracy in RUL predictions for different batteries. Especially in error metrics such as root mean square error (RMSE) and mean absolute error (MAE), the minimum error can reach 0.0104 and 0.0082, respectively, where it significantly outperforms traditional data-driven models.