State of Health Estimation of Lithium-Ion Battery Based on Novel Health Indicators and Improved Support Vector Regression

Accurate estimation of the state of health (SOH) is a critical function of battery management system (BMS), essential for ensuring the safe and stable operation of lithium-ion batteries. To improve estimation precision, this paper proposes a novel health indicator (HI) construction method and an improved support vector regression (SVR) approach. First, the convolution operation is applied to discharge voltage data to extract new HIs that characterize battery aging; their correlations are then verified. Second, principal component analysis (PCA) is employed to reduce input dimensionality and computational burden. Third, to address the challenge of SVR parameter selection, an improved sparrow search algorithm (ISSA) is proposed for parameter optimization. Finally, the proposed method is validated using both the NASA dataset and a laboratory experimental dataset, with comparisons against existing approaches. The results show that the method achieves accurate SOH estimation under various aging conditions, demonstrating its effectiveness, robustness, and practical potential.