Scholarlink

<p>The state of charge (SOC) of lithium-ion batteries is vital for efficient energy management and prolonging battery lifespan. To improve the accuracy of SOC estimation for lithium-ion batteries, this paper proposes an improved genetic algorithm (IGA) and a nonlinear autoregressive particle filter (NARXNN-PF: nonlinear autoregressive neural network with exogenous inputs integrated with particle filter) for parameter identification and SOC estimation, respectively. Based on the dual-polarization model, parameter identification is achieved by minimizing terminal voltage errors while accounting for uncertainties in initial conditions and measurement errors. Using the accurately identified model parameters, the NARXNN-PF is applied for online estimation. The SOC predictions generated by the NARXNN serve as prior information for the particle filter. During particle weight updates, the predictive capability of the NARXNN is leveraged to refine particle weights, optimizing their distribution and thereby enhancing the algorithm’s overall accuracy and robustness.</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

Based on NARXNN-PF state of charge estimation for lithium batteries

  • 期刊论文

  • 发表日期：

    2025年1月28日

查看全文
  • Chen Haizhong,
  • Hou Huiheng,
  • Liu Feng,
  • Shen Xin

摘要

The state of charge (SOC) of lithium-ion batteries is vital for efficient energy management and prolonging battery lifespan. To improve the accuracy of SOC estimation for lithium-ion batteries, this paper proposes an improved genetic algorithm (IGA) and a nonlinear autoregressive particle filter (NARXNN-PF: nonlinear autoregressive neural network with exogenous inputs integrated with particle filter) for parameter identification and SOC estimation, respectively. Based on the dual-polarization model, parameter identification is achieved by minimizing terminal voltage errors while accounting for uncertainties in initial conditions and measurement errors. Using the accurately identified model parameters, the NARXNN-PF is applied for online estimation. The SOC predictions generated by the NARXNN serve as prior information for the particle filter. During particle weight updates, the predictive capability of the NARXNN is leveraged to refine particle weights, optimizing their distribution and thereby enhancing the algorithm’s overall accuracy and robustness.