Model-Based State of Charge Estimation for Lithium-Ion Batteries: HPPC-Based Parameter Identification and Comparison of Kalman Filters

Creative Commons License

Demirci O., Kuzu Ü., Taşkın S.

Ankara International Congress on Scientific Research-XIII, Ankara, Türkiye, 1 - 03 Mart 2026, ss.841-857, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Basıldığı Şehir: Ankara
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.841-857
  • Manisa Celal Bayar Üniversitesi Adresli: Evet

Özet

Accurate State of Charge (SOC) estimation is essential for the safe, efficient, and reliable operation of lithium-ion batteries in electric vehicles and energy storage systems. This study aims to improve model-based SOC estimation performance through cell-specific parameter identification. A second-order equivalent circuit model was employed, and model parameters were experimentally identified using Hybrid Pulse Power Characterization (HPPC) tests. Capacity and HPPC experiments were conducted over the full SOC range (0–100%), enabling the extraction of ohmic resistance, polarization resistances, and time constants as SOC dependent parameters. Based on the identified model parameters, Coulomb counting, Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF), and Adaptive Extended Kalman Filter (AEKF) algorithms were implemented and comparatively evaluated. SOC estimation performance was assessed using RMSE, MAE, and MAPE error metrics under both experimentally obtained datasets and widely used benchmark datasets from the literature. The results indicate that HPPC-based cell-specific parameter identification significantly enhances SOC estimation accuracy. While the Coulomb counting method was found to be prone to cumulative error under long-term operation, Kalman filter-based approaches provided considerably improved estimation performance. Among the evaluated methods, the UKF achieved the lowest error metrics and demonstrated superior robustness under dynamic load conditions due to its enhanced capability in handling nonlinear battery dynamics. The findings highlight that accurate model parameterization is as critical as filter selection in achieving high performance model-based SOC estimation.