Predicting the Remaining Useful Life of Lithium-ion Batteries with Mode-Gating Network

Accurately predicting the Remaining Useful Life (RUL) of lithium-ion batteries is crucial for individual electronic devices. However, due to the complexity of battery aging principles and the phenomenon of capacity regeneration, accurately predicting the RUL of lithium-ion batteries remains a challenge. To address this, this paper proposes an innovative Mode-Gating Network (MGN) for accurately predicting the RUL of lithium-ion batteries. Firstly, Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) is employed to perform multi-scale decomposition on battery capacity degradation data for obtaining Intrinsic Modal Functions (IMFs) and a residual. Then, Transformer networks are used to predict each IMF and the residual. Meanwhile, the IMFs are input into a gating network to merge the predictions of the Transformer networks for obtaining the final RUL prediction. Experiments on the NASA dataset indicate that the prediction accuracy of MGN is significantly superior to the existing advanced methods, demonstrating its effectiveness and robustness.