Scholarlink

<p>Accurate estimation of the state of charge (SOC) in lithium-ion batteries is crucial for battery management systems. Addressing the issue of insufficient accuracy in conventional SOC estimation methods under dynamic operating conditions, this paper proposes a lithium-ion battery SOC estimation method incorporating stress characteristics. To compensate for SOC estimation deficiencies caused by fluctuations in voltage and current characteristics, this paper introduces battery stress variation as a novel feature to characterise SOC degradation. A stress testing experimental platform was constructed, with a series of stress characterisation experiments designed. Savitzky-Golay (SG) filtering was applied to the stress data to obtain optimised, high-quality stress features. Correlation analysis validated the strong relationship between these stress features and SOC. To address the degradation in SOC estimation accuracy caused by the plateau phase in stress variation curves during battery discharge, this paper proposes an LSSVM-AdaBoost-based SOC estimation model. By integrating the advantages of multiple LSSVM models and combining locally optimal models for different charge–discharge phases, this approach significantly enhances lithium-ion battery SOC estimation accuracy. SOC estimation was conducted under Federal Urban Driving State (FUDS) conditions using the constructed experimental platform. Experimental results demonstrate that the root mean square error (RMSE) and mean absolute error (MAE) of the proposed method’s SOC estimates are 0.225 and 0.186, respectively. To validate the effectiveness of stress features, the superiority of the LSSVM-AdaBoost model, and its generalisation capability across different operating conditions, a series of comparative experiments were designed. Results confirm that the LSSVM-AdaBoost model maintains relatively superior estimation accuracy across the entire SOC range. Consequently, the proposed method effectively enhances the accuracy and robustness of lithium-ion battery SOC estimation.</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

SOC estimation of lithium-ion batteries based on LSSVM-AdaBoost modeling with stress characteristic analysis

  • 期刊论文

  • 发表日期：

    2026年3月14日

查看全文
  • Tiezhou Wu,
  • Qi Xie,
  • Lang Mao,
  • Junchao Zhu,
  • Jun Zhang,
  • Jian Kang

摘要

Accurate estimation of the state of charge (SOC) in lithium-ion batteries is crucial for battery management systems. Addressing the issue of insufficient accuracy in conventional SOC estimation methods under dynamic operating conditions, this paper proposes a lithium-ion battery SOC estimation method incorporating stress characteristics. To compensate for SOC estimation deficiencies caused by fluctuations in voltage and current characteristics, this paper introduces battery stress variation as a novel feature to characterise SOC degradation. A stress testing experimental platform was constructed, with a series of stress characterisation experiments designed. Savitzky-Golay (SG) filtering was applied to the stress data to obtain optimised, high-quality stress features. Correlation analysis validated the strong relationship between these stress features and SOC. To address the degradation in SOC estimation accuracy caused by the plateau phase in stress variation curves during battery discharge, this paper proposes an LSSVM-AdaBoost-based SOC estimation model. By integrating the advantages of multiple LSSVM models and combining locally optimal models for different charge–discharge phases, this approach significantly enhances lithium-ion battery SOC estimation accuracy. SOC estimation was conducted under Federal Urban Driving State (FUDS) conditions using the constructed experimental platform. Experimental results demonstrate that the root mean square error (RMSE) and mean absolute error (MAE) of the proposed method’s SOC estimates are 0.225 and 0.186, respectively. To validate the effectiveness of stress features, the superiority of the LSSVM-AdaBoost model, and its generalisation capability across different operating conditions, a series of comparative experiments were designed. Results confirm that the LSSVM-AdaBoost model maintains relatively superior estimation accuracy across the entire SOC range. Consequently, the proposed method effectively enhances the accuracy and robustness of lithium-ion battery SOC estimation.