Scholarlink

<p>The article proposes an online terminal voltage collapse detection algorithm for lithium-ion batteries based on terminal voltage and cell temperature. In contrast to the model-based approach, battery modeling and state of charge estimation are not required in the proposed scheme. In addition, the proposed scheme is free from massive training data and labeling. Moreover, it relies on dynamic mode decomposition and advanced Kalman filtering (KF) techniques, including standard KF, ensemble KF, and robust KF, and a novel residual-corrected adaptive linear estimator (RCALE). Three methods of noise estimation, namely autocovariance least squares, innovation-based adaptive estimation (IAE), and Bayesian recursive noise estimation, are employed in enhancing the robustness of the estimator to thermal variability and aging. The experiments are validated using actual battery dataset under temperature and aging effects, where the dataset obtained by NASA. The experiments indicate that the proposed RCALE and standard KF combined with IAE achieve over 98% detection accuracy, outperforming the current state-of-the-art methods, such as machine learning and model-based approaches.</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

Lithium battery failure detection algorithm based on data-driven modeling and novel residual-corrected adaptive linear estimator

  • 期刊论文

  • 发表日期：

    2026年6月29日

查看全文
  • Ali Qahtan Tameemi,
  • Jeevan Kanesan,
  • Anis Salwa Mohd Khairuddin

摘要

The article proposes an online terminal voltage collapse detection algorithm for lithium-ion batteries based on terminal voltage and cell temperature. In contrast to the model-based approach, battery modeling and state of charge estimation are not required in the proposed scheme. In addition, the proposed scheme is free from massive training data and labeling. Moreover, it relies on dynamic mode decomposition and advanced Kalman filtering (KF) techniques, including standard KF, ensemble KF, and robust KF, and a novel residual-corrected adaptive linear estimator (RCALE). Three methods of noise estimation, namely autocovariance least squares, innovation-based adaptive estimation (IAE), and Bayesian recursive noise estimation, are employed in enhancing the robustness of the estimator to thermal variability and aging. The experiments are validated using actual battery dataset under temperature and aging effects, where the dataset obtained by NASA. The experiments indicate that the proposed RCALE and standard KF combined with IAE achieve over 98% detection accuracy, outperforming the current state-of-the-art methods, such as machine learning and model-based approaches.