Lithium‐ion batteries play a crucial role in energy storage for mobile electronics, electric vehicles, and renewable energy systems, where accurate state of health (SOH) estimation is essential to ensure system reliability and operational safety. However, achieving high‐precision data‐driven SOH estimation involves overcoming significant challenges in feature extraction, processing, and model training. To tackle these challenges, this article proposes a novel SOH estimation method integrating Savitzky–Golay (SG) filter with Crested Porcupine Optimizer (CPO)‐enhanced Support Vector Machine (SVM). First, representative health features correlated with capacity degradation are systematically extracted from charge voltage, current, and incremental capacity curves using Kendall correlation analysis. Subsequently, SG filter is employed for noise reduction and data smoothing, significantly enhancing the stability and reliability of the input parameters while preserving critical information. Furthermore, to achieve high‐accuracy SOH estimation, a CPO‐SVM framework is proposed. By dynamically adjusting the penalty coefficient and kernel function hyperparameters, this approach effectively addresses the local optimum issue commonly encountered in traditional SVM methods. Finally, the proposed method is validated using datasets from NASA, CALCE, and Oxford, demonstrating superior performance with root mean square error 98.5%. These results highlight substantial improvements in both accuracy and reliability of SOH estimation.
Zheng et al. (Sun,) studied this question.