Scholarlink

<p>Accurate prediction of the Remaining Useful Life (RUL) of lithium-ion batteries is critical for energy storage management. This study proposes a hybrid prognostic framework integrating Non-dominated Sorting Genetic Algorithm II (NSGA-II) with Multi-Layer Perceptron-based Particle Filtering (MLP-PF). The approach combines physics-based degradation modeling with data-driven learning to capture both deterministic and stochastic battery behavior. NSGA-II optimizes key degradation parameters, improving adaptability under multiple objectives, while MLP-PF models nonlinear temporal dynamics and uncertainty propagation. The proposed framework is validated using battery degradation datasets. The results demonstrate a reduction in prediction error, achieving a Root Mean Square Error (RMSE) of 0.012–0.018 and Mean Absolute Error (MAE) below 2.5%, outperforming conventional methods by approximately 15–25%. Additionally, the model provides reliable uncertainty bounds with improved prediction stability. These findings confirm the effectiveness of the proposed hybrid approach for accurate RUL estimation. The framework is suitable for real-time battery health monitoring and supports improved decision-making in battery management systems.</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

Integrated NSGA-II and MLP-PF model for enhanced lithium-ion battery prognostics

  • 期刊论文

  • 发表日期：

    2026年5月14日

查看全文
  • R. Sivanand,
  • M. V. S. Prem Sagar,
  • A. Elaiyaraja,
  • S. Suganya,
  • Priya S. Lakshmi,
  • V. Saravanan

摘要

Accurate prediction of the Remaining Useful Life (RUL) of lithium-ion batteries is critical for energy storage management. This study proposes a hybrid prognostic framework integrating Non-dominated Sorting Genetic Algorithm II (NSGA-II) with Multi-Layer Perceptron-based Particle Filtering (MLP-PF). The approach combines physics-based degradation modeling with data-driven learning to capture both deterministic and stochastic battery behavior. NSGA-II optimizes key degradation parameters, improving adaptability under multiple objectives, while MLP-PF models nonlinear temporal dynamics and uncertainty propagation. The proposed framework is validated using battery degradation datasets. The results demonstrate a reduction in prediction error, achieving a Root Mean Square Error (RMSE) of 0.012–0.018 and Mean Absolute Error (MAE) below 2.5%, outperforming conventional methods by approximately 15–25%. Additionally, the model provides reliable uncertainty bounds with improved prediction stability. These findings confirm the effectiveness of the proposed hybrid approach for accurate RUL estimation. The framework is suitable for real-time battery health monitoring and supports improved decision-making in battery management systems.