Joint estimation model of SOC-SOH for lithium-ion batteries integrating Transformer and BiLSTM

The accurate estimation of the State of Charge (SOC) and State of Health (SOH) of lithium-ion batteries is crucial for ensuring their reliability and safety in energy storage systems. Conventional estimation methods often fail to account for the strong coupling between SOC and SOH, leading to suboptimal accuracy. In this study, we propose a joint SOC-SOH estimation model that integrates Transformer and Bidirectional Long Short-Term Memory (BiLSTM) networks. The Transformer module captures global temporal dependencies, while the BiLSTM module models local dynamic features, effectively fusing multi-timescale information for robust state estimation. Experimental results on the Oxford battery aging dataset demonstrate that our model achieves high accuracy and robustness, with SOC estimation reaching an R2of 0.989 at SOH 95% and 0.976 at SOH 80%, significantly outperforming conventional single-model approaches. The proposed framework provides a reliable solution for real-time lithium battery health management and lays the foundation for further research in deep learning-based battery state estimation.