Scholarlink

<p>With the rapid growth of the electric vehicle industry, it has become a critical challenge to accurately predict the remaining useful life (RUL) of lithium-ion batteries. This paper introduces a novel RUL prediction framework called Mamba-MoE. It integrates Mamba with a mixture of experts (MoE) and other innovations to enhance accuracy and robustness. Our approach integrates wavelet threshold denoising (WTD), a state-space model (SSM), and MoE for long-term sequence modeling. We also introduce an interpretable hyperparameter optimization method to refine performance further, ensuring adaptability across diverse battery datasets. Extensive experiments on three datasets demonstrate that Mamba-MoE significantly outperforms existing methods in terms of accuracy and robustness. This work advances RUL prediction through an efficient, interpretable, and scalable framework, achieving an average R<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11581_2025_6607_Article_IEq1.gif" Format="GIF" Height="10" Rendition="HTML" Resolution="72" Type="Linedraw" Width="8" /> </InlineMediaObject> <EquationSource Format="TEX">\(^2\)</EquationSource> <EquationSource Format="MATHML"><math> <mmultiscripts> <mrow /> <mrow /> <mn>2</mn> </mmultiscripts> </math></EquationSource> </InlineEquation> exceeding 0.9833 across all datasets and providing valuable insights for real-world battery health management.</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 enhanced prognostication of lithium-ion batteries degradation trajectory and remaining useful life based on Mamba-MoE model

  • 期刊论文

  • 发表日期：

    2025年8月6日

查看全文
  • Fengyi Wang,
  • Minghan Bao,
  • Qiang Ru

摘要

With the rapid growth of the electric vehicle industry, it has become a critical challenge to accurately predict the remaining useful life (RUL) of lithium-ion batteries. This paper introduces a novel RUL prediction framework called Mamba-MoE. It integrates Mamba with a mixture of experts (MoE) and other innovations to enhance accuracy and robustness. Our approach integrates wavelet threshold denoising (WTD), a state-space model (SSM), and MoE for long-term sequence modeling. We also introduce an interpretable hyperparameter optimization method to refine performance further, ensuring adaptability across diverse battery datasets. Extensive experiments on three datasets demonstrate that Mamba-MoE significantly outperforms existing methods in terms of accuracy and robustness. This work advances RUL prediction through an efficient, interpretable, and scalable framework, achieving an average R \(^2\) 2 exceeding 0.9833 across all datasets and providing valuable insights for real-world battery health management.