A New Attention-based Method For Estimating Li-ion Battery State-of-Charge

This paper examines the performance of the inverted Transformer (iTransformer) model, a cutting-edge attention-based architecture, in estimating the state of charge (SoC) of Li-Ion batteries in electric vehicles (EVs). To enhance battery longevity and optimize driving range in battery electric vehicles (BEVs), developing an accurate estimation technique for the SoC that can be used in the battery management system has become the primary interest of various automotive research studies. The iTransformer is a recent version of the Transformer model that demonstrated promising prediction accuracy and enhanced computational efficiency in various forecasting problems. This paper involves training the iTransformer on a real-world Li-Ion battery dataset that includes standard driving cycles at various temperatures and comparing the prediction accuracy with other architectures. Research results obtained from the iTransformer indicated an RMSE of 0.06% and MAXE of 0.9%. Future work will focus on experimenting to refine the model and expanding its applicability to a broader range of battery technologies.