Lithium-Ion Battery’s State of Health Prediction: A Deep Learning Framework

This paper presents a comprehensive study on predictive maintenance of lithium-ion batteries in electric vehicles (EVs) using data-driven approaches. The study involves collecting data from four individual battery cells, each subjected to various charging and discharging parameters. After preprocessing the data, we apply feature extraction techniques to extract relevant features. Subsequent data analysis guides the development of machine learning (ML) and deep learning (DL) models on the combined dataset of the four cells. A crucial aspect of this study involves addressing measurement noise inherent in cellwise data. Through innovative techniques, we mitigate the effects of measurement noise, improving the accuracy and robustness of our models. The proposed DL models demonstrate remarkable efficiency in handling noise, leading to superior predictive performance in estimating State of Health (SoH) as degraded capacity. The findings of this research offer valuable insights into predictive maintenance strategies for EV batteries, providing a pathway towards optimized battery management. The methodologies and techniques presented herein contribute to advancing battery health monitoring systems, thereby enhancing battery lifespan and performance in EV applications.