José Ignacio Olalde-Verano Sascha Kirch Clara Pérez-Molina Sergio Martin

Abstract

The state of health (SOH) of a Li-ion battery is a critical parameter thatdetermines the remaining capacity and the remaining lifetime of the battery. Inthis paper, we propose SambaMixer a novel structured state space model (SSM)for predicting the state of health of Li-ion batteries. The proposed SSM isbased on the MambaMixer architecture, which is designed to handle multi-variatetime signals. We evaluate our model on the NASA battery discharge dataset andshow that our model outperforms the state-of-the-art on this dataset. Wefurther introduce a novel anchor-based resampling method which ensures timesignals are of the expected length while also serving as augmentationtechnique. Finally, we condition prediction on the sample time and the cycletime difference using positional encodings to improve the performance of ourmodel and to learn recuperation effects. Our results proof that our model isable to predict the SOH of Li-ion batteries with high accuracy and robustness.