Scholarlink

There is an increasing demand for modern diagnostic systems for batteries in real-world operation, specifically for estimating their state of health. Online estimation of the capacity of a cell is challenging because of the dynamic nature of cell ageing and the limited parameters available from the cell under operation. Recently, data-driven capacity estimation models based on Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM) have been developed for cells under real-world working conditions in the literature. This research incorporates LSTM-based approaches to efficiently predict the capacity estimation of the remaining battery life or the degradation time. Alterations in dataset formation using statistical approaches to boost the results are performed, proven effective with 0.0067 validation loss and the parameters tuned in the model. This can be used in the real-world by opting for a neural network approach to estimate the remaining battery life instead of manual and traditional methods, as this proves to be more fruitful and uncomplicated for applications in large industries, such as electric vehicles, cell phones, and laptops.

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

Capacity Estimation of Li-Ion Battery Cells Using Deep Neural Networks

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Harshal Hirpara,
  • Shail Shah,
  • Zunnun Narmawala,
  • Jaiprakash Verma

摘要

There is an increasing demand for modern diagnostic systems for batteries in real-world operation, specifically for estimating their state of health. Online estimation of the capacity of a cell is challenging because of the dynamic nature of cell ageing and the limited parameters available from the cell under operation. Recently, data-driven capacity estimation models based on Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM) have been developed for cells under real-world working conditions in the literature. This research incorporates LSTM-based approaches to efficiently predict the capacity estimation of the remaining battery life or the degradation time. Alterations in dataset formation using statistical approaches to boost the results are performed, proven effective with 0.0067 validation loss and the parameters tuned in the model. This can be used in the real-world by opting for a neural network approach to estimate the remaining battery life instead of manual and traditional methods, as this proves to be more fruitful and uncomplicated for applications in large industries, such as electric vehicles, cell phones, and laptops.