A comprehensive review of data-driven electrochemical impedance spectroscopy (EIS) approaches for Li-ion battery SoH and RUL estimations

Electrochemical Impedance Spectroscopy (EIS) is a non-invasive diagnostic technique that has significant potential in determining the state of Li-ion batteries. Among the most critical parameters for battery management are state of health (SoH) and remaining useful life (RUL), which provide key insights into battery degradation and longevity. Traditionally, battery health estimation relied on signals like voltage, current, and temperature data. Among different health estimation techniques, data-driven approaches have gained prominence for their ability to extract deeper insights from data. This review presents a comprehensive overview of EIS-based data-driven techniques employed in the literature for SoH and RUL estimation. The paper first categorises various feature engineering methods used to extract meaningful performance indicators from EIS spectra. It then classifies the data-driven estimation techniques and compares the performance of different algorithms. By synthesising existing work, this review provides a deeper understanding of the most commonly used machine learning (ML) and deep learning (DL) algorithms, key performance metrics, and directly and indirectly derived features from EIS spectra that are strongly correlated with battery health. It aims to guide future research in selecting appropriate modelling strategies and to highlight the emerging role of EIS-based, data-driven battery prognostics.