Remaining Useful Life Prediction of Lithium-Ion Batteries Based on CEEMDAN-Bi-LSTM Hybrid Model

Accurately predicting the remaining useful life (RUL) is crucial for maintaining the stable operation of equipment, yet it remains a challenging task. Capacity regeneration phenomena can lead to inaccurate RUL predictions. To address this, this paper proposes a prediction method using a CEEMDAN-Bi-LSTM hybrid model. Initially, multiple health features are extracted from charging variation curves. To mitigate the effects of capacity regeneration, the complete ensemble empirical mode decomposition with adaptive noise is applied to decompose the HFs. Subsequently, a bidirectional long short-term memory neural network is introduced to construct the CEEMDAN-Bi-LSTM hybrid model for RUL prediction of lithium batteries. The effectiveness of the proposed model is verified using NASA experimental data by comparing it with several other models. The results demonstrate that the proposed model can significantly enhance the prediction accuracy of lithium-ion battery RUL.