Enhanced State-of-Charge Prediction for Lithium-Ion Batteries using Informer-LSTM Fusion

State of Charge (SoC) estimation plays a vital role in optimising Battery Management Systems (BMS), particularly for electric vehicles and renewable energy storage applications. However, traditional SoC estimation techniques often struggle to capture modern lithium-ion batteries' complex, non-linear behaviours. This paper introduces a deep learning hybrid model integrating Long Short-Term Memory (LSTM) with Informer model to enhance SoC estimation accuracy. The LSTM component captures short-term temporal dependencies, while the Informer addresses long-term dependencies, improving the overall robustness and accuracy of the system. The model is evaluated on real-world datasets, including UDDS, FUDS, and US06 driving cycles. Experimental findings indicate that the proposed hybrid model outperforms standalone LSTM and Informer models, attaining a Root Mean Square Error (RMSE) of 1.593% and a peak Mean Absolute Error (MAE) of 1.395%. This work demonstrates the potential of combining Informer and LSTM for real-time SoC estimation in battery management systems, offering a robust solution for the future of energy management.