Scholarlink

<p>The accuracy of parameter identification in lithium-ion battery models defines the upper limit of state estimation and is essential for life-cycle management. To overcome the limitations of conventional methods under wide temperature ranges and complex operating conditions, this study proposes a key dynamic parameters online identification approach combining a hybrid electrochemical empirical and circuit modeling with a variable forgetting factor sliding window recursive least squares-chaotic firefly algorithm. The method introduces dynamic weight adjustment and chaotic optimization to enhance convergence and stability. Experimental validation shows that, compared with the forgetting factor recursive least squares algorithm, the proposed method achieves higher accuracy across different temperatures and operating conditions. At -5℃, the mean absolute error, root mean square error, and mean absolute percentage error are reduced by 4.42%, 4.54%, and 4.56% under the Beijing Bus Dynamic Stress Test, and 2.11%, 2.33%, and 2.18% under the Dynamic Stress Test. At 35℃, the reductions reach 8.41%, 9.51%, and 8.51% under BBDST, and 0.68%, 1.18%, and 0.33% under DST. These results confirm the method’s robustness and its potential to support high-precision state estimation and reliable life-cycle management in electric vehicles and energy storage applications.</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 improved variable forgetting factor sliding window recursive least square-chaotic firefly optimization method for key dynamic parameters identification of lithium-ion batteries with hybrid electrochemical empirical and circuit modeling

  • 期刊论文

  • 发表日期：

    2026年2月6日

查看全文
  • Liangwei Cheng,
  • Shunli Wang,
  • Lei Zhou,
  • Hongqiu Xie,
  • Carlos Fernandez

摘要

The accuracy of parameter identification in lithium-ion battery models defines the upper limit of state estimation and is essential for life-cycle management. To overcome the limitations of conventional methods under wide temperature ranges and complex operating conditions, this study proposes a key dynamic parameters online identification approach combining a hybrid electrochemical empirical and circuit modeling with a variable forgetting factor sliding window recursive least squares-chaotic firefly algorithm. The method introduces dynamic weight adjustment and chaotic optimization to enhance convergence and stability. Experimental validation shows that, compared with the forgetting factor recursive least squares algorithm, the proposed method achieves higher accuracy across different temperatures and operating conditions. At -5℃, the mean absolute error, root mean square error, and mean absolute percentage error are reduced by 4.42%, 4.54%, and 4.56% under the Beijing Bus Dynamic Stress Test, and 2.11%, 2.33%, and 2.18% under the Dynamic Stress Test. At 35℃, the reductions reach 8.41%, 9.51%, and 8.51% under BBDST, and 0.68%, 1.18%, and 0.33% under DST. These results confirm the method’s robustness and its potential to support high-precision state estimation and reliable life-cycle management in electric vehicles and energy storage applications.