Scholarlink

<p>The State of Health (SOH) of lithium-ion batteries is an important parameter of the battery management system and plays a decisive role in the reliability and safety of the batteries. This paper proposes an innovative northern goshawk optimization - hybrid neural network (NGO-HNN) algorithm for highly accurate SOH estimation. First, the convolutional neural network (CNN) layer extracts local features from the original battery data to capture important patterns during the battery charging process. Next, the bidirectional long short-term memory network (BiLSTM) layer learns the long-term dependencies of the battery data from both forward and backward directions to enhance the understanding of the temporal information. Then, the self-attention (SA) weights the output of the BiLSTM to highlight the features most relevant to the SOH estimation. Finally, the NGO algorithm globally optimizes the model’s hyperparameters by simulating the predatory behavior of the northern goshawk, avoiding getting trapped in local optimal solutions and further improving the model’s accuracy and generalization ability. The verification results on the National Aeronautics and Space Administration (NASA) dataset show that, compared with the hybrid neural network (HNN) algorithm, the proposed NGO - HNN algorithm reduces the maximum error (ME) by more than 37.27% in the single - battery verification and by more than 15.86% in the multi - battery cross - validation. This research provides an efficient and reliable solution for the SOH estimation of lithium-ion batteries.&#xa0;</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 innovative northern goshawk optimization - hybrid neural network algorithm for highly accurate state of health estimation of lithium-ion batteries

  • 期刊论文

  • 发表日期：

    2025年11月18日

查看全文
  • Liang Zhang,
  • Donglei Liu,
  • Shunli Wang,
  • Yurong Zhou,
  • Carlos Fernandez

摘要

The State of Health (SOH) of lithium-ion batteries is an important parameter of the battery management system and plays a decisive role in the reliability and safety of the batteries. This paper proposes an innovative northern goshawk optimization - hybrid neural network (NGO-HNN) algorithm for highly accurate SOH estimation. First, the convolutional neural network (CNN) layer extracts local features from the original battery data to capture important patterns during the battery charging process. Next, the bidirectional long short-term memory network (BiLSTM) layer learns the long-term dependencies of the battery data from both forward and backward directions to enhance the understanding of the temporal information. Then, the self-attention (SA) weights the output of the BiLSTM to highlight the features most relevant to the SOH estimation. Finally, the NGO algorithm globally optimizes the model’s hyperparameters by simulating the predatory behavior of the northern goshawk, avoiding getting trapped in local optimal solutions and further improving the model’s accuracy and generalization ability. The verification results on the National Aeronautics and Space Administration (NASA) dataset show that, compared with the hybrid neural network (HNN) algorithm, the proposed NGO - HNN algorithm reduces the maximum error (ME) by more than 37.27% in the single - battery verification and by more than 15.86% in the multi - battery cross - validation. This research provides an efficient and reliable solution for the SOH estimation of lithium-ion batteries.