Scholarlink

<p>Traditional state of charge (SOC) estimation methods, such as Thevenin equivalent circuit models, face limitations in capturing the dynamic behavior of lithium-ion batteries, particularly in terms of impedance and internal electrochemical processes. This paper presents a fractional-order equivalent circuit model that improves on the Thevenin model by substituting an ideal capacitor with a fractional-order capacitor, enabling more accurate representation of battery reactions. A fractional-order particle swarm optimization algorithm with dynamic inertia weights is used for parameter identification, enhancing the model’s ability to reflect actual battery dynamics. The fractional-order unscented Kalman filter (FUKF) manages nonlinearities and uncertainties in SOC estimation, while an integrated sliding mode observer boosts robustness against disturbances and model inaccuracies. Experimental results show that the improved FUKF (IFUKF) achieves superior SOC estimation accuracy. Under UDDS conditions, it reaches a mean absolute error of 0.65%, a root mean square error of 0.69%, and a maximum error of 0.88%. Under DST conditions, the mean absolute error is 0.71%, the root mean square error is 0.73%, and the maximum error is 0.97%.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An improved unscented Kalman filter for SOC estimation of lithium-ion batteries based on fractional-order model

  • 期刊论文

  • 发表日期：

    2025年3月27日

查看全文
  • Yingying Wang,
  • Jie Ding,
  • Taotao Tu

摘要

Traditional state of charge (SOC) estimation methods, such as Thevenin equivalent circuit models, face limitations in capturing the dynamic behavior of lithium-ion batteries, particularly in terms of impedance and internal electrochemical processes. This paper presents a fractional-order equivalent circuit model that improves on the Thevenin model by substituting an ideal capacitor with a fractional-order capacitor, enabling more accurate representation of battery reactions. A fractional-order particle swarm optimization algorithm with dynamic inertia weights is used for parameter identification, enhancing the model’s ability to reflect actual battery dynamics. The fractional-order unscented Kalman filter (FUKF) manages nonlinearities and uncertainties in SOC estimation, while an integrated sliding mode observer boosts robustness against disturbances and model inaccuracies. Experimental results show that the improved FUKF (IFUKF) achieves superior SOC estimation accuracy. Under UDDS conditions, it reaches a mean absolute error of 0.65%, a root mean square error of 0.69%, and a maximum error of 0.88%. Under DST conditions, the mean absolute error is 0.71%, the root mean square error is 0.73%, and the maximum error is 0.97%.