Scholarlink

<p>Accurate prediction of the reversible capacity and remaining useful life (RUL) of Li-ion batteries (LiBs) is crucial for effective battery management strategies. This paper combines a bimodal decomposition algorithm with a Long Short-Term Memory (LSTM) network for LiB capacity and RUL prediction. First, battery health features and capacity degradation trends are decomposed with the Complementary Ensemble Empirical Mode Decomposition (CEEMDAN) and subsequently clustered using the k-means method to identify modal components. Then high-frequency components are further decomposed using Variational Mode Decomposition (VMD), optimized with the Symbiotic Organisms Search (SOS) algorithm. These processed features are then fed into the LSTM model to construct their relevance with respect to capacity and RUL. Finally, comparative experiments using the NASA dataset demonstrate that the proposed method outperforms other mainstream approaches, particularly in capturing the capacity regeneration interval, showcasing its superior prediction accuracy.</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

Bimodal decomposition and LSTM network-based joint prediction of reversible capacity and remaining useful life for Li-ion batteries

  • 期刊论文

  • 发表日期：

    2025年5月8日

查看全文
  • Suoqing Yu,
  • Liping Zhang,
  • Liyong Ni,
  • Shuzhen Feng

摘要

Accurate prediction of the reversible capacity and remaining useful life (RUL) of Li-ion batteries (LiBs) is crucial for effective battery management strategies. This paper combines a bimodal decomposition algorithm with a Long Short-Term Memory (LSTM) network for LiB capacity and RUL prediction. First, battery health features and capacity degradation trends are decomposed with the Complementary Ensemble Empirical Mode Decomposition (CEEMDAN) and subsequently clustered using the k-means method to identify modal components. Then high-frequency components are further decomposed using Variational Mode Decomposition (VMD), optimized with the Symbiotic Organisms Search (SOS) algorithm. These processed features are then fed into the LSTM model to construct their relevance with respect to capacity and RUL. Finally, comparative experiments using the NASA dataset demonstrate that the proposed method outperforms other mainstream approaches, particularly in capturing the capacity regeneration interval, showcasing its superior prediction accuracy.