ABSTRACT Lithium‐ion batteries, crucial for electric vehicles and energy storage systems, encounter capacity degradation and safety risks over time, necessitating precise state of health (SOH) estimation for reliable operation and risk management. Traditional methods often rely on single‐view incremental capacity (IC) curves, which inadequately leverage the full range of available data. A multi‐view SOH estimation method based on variational slow features (VSFs) is presented in this paper. Using cubic spline interpolation and Kalman filtering for preprocessing, multi‐view IC curves are calculated and generated. Five health indicators (HIs) are then extracted from these multi‐view IC curves, and a detailed correlation analysis highlights the HI extracted from the enhanced view IC curves with the highest correlation. To further optimise the use of these HIs, a novel technique employing VSF is proposed. Robust features are extracted from the multi‐view HIs using a Siamese‐Variational Autoencoder, enhancing the precision of SOH assessment. A two‐stage training process captures the dynamic nature of capacity degradation. The method's effectiveness is demonstrated through experiments on the CALCE open dataset, achieving high accuracy and robustness with a root mean square error of 1.019e−2 and an R 2 value of 0.987, confirming its suitability for battery health monitoring.
Yang et al. (Thu,) studied this question.