Scholarlink

<p>Accurate state of charge (SoC) estimation for lithium-ion batteries is challenged by nonlinear electrochemical dynamics, parametric uncertainties, and sensor noise. This paper proposes an attractive ellipsoid super-twisting sliding mode observer (AE-STSMO) to address these issues. The method integrates the super-twisting algorithm (STA) to mitigate chattering and accelerate convergence, with the attractive ellipsoid method (AEM) for optimal gain selection via linear matrix inequalities (LMI). Lyapunov stability analysis confirms that the estimation error is uniformly ultimately bounded. Experimental validation using the CALCE dataset under dynamic driving schedules (DST, BJDST, FUDS, and US06) and varying temperatures (0–<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(45^{\circ }\)</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mn>45</mn> <mo>∘</mo> </msup> </math></EquationSource> </InlineEquation>C) demonstrates the observer’s efficacy. The AE-STSMO achieves a root mean square error (RMSE) consistently below 0.17%, exhibiting superior thermal robustness and accuracy compared to standard filtering and machine learning benchmarks.</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

The State of Charge Estimation Method for Lithium Battery Using Attractive Ellipsoid Super-Twisting Sliding Mode Observer

  • 期刊论文

  • 发表日期：

    2026年6月3日

查看全文
  • Ngoc Quan Dinh,
  • Kien Trung Nguyen

摘要

Accurate state of charge (SoC) estimation for lithium-ion batteries is challenged by nonlinear electrochemical dynamics, parametric uncertainties, and sensor noise. This paper proposes an attractive ellipsoid super-twisting sliding mode observer (AE-STSMO) to address these issues. The method integrates the super-twisting algorithm (STA) to mitigate chattering and accelerate convergence, with the attractive ellipsoid method (AEM) for optimal gain selection via linear matrix inequalities (LMI). Lyapunov stability analysis confirms that the estimation error is uniformly ultimately bounded. Experimental validation using the CALCE dataset under dynamic driving schedules (DST, BJDST, FUDS, and US06) and varying temperatures (0– \(45^{\circ }\) 45 ∘ C) demonstrates the observer’s efficacy. The AE-STSMO achieves a root mean square error (RMSE) consistently below 0.17%, exhibiting superior thermal robustness and accuracy compared to standard filtering and machine learning benchmarks.