Scholarlink

<p>The rapid shift to clean energy technologies has propelled the adoption of Electric Vehicles (EVs), necessitating advanced battery state estimation techniques to optimize lithium-ion (Li-ion) battery performance. Traditional estimation approaches struggle to accurately estimate battery states due to the complex and nonlinear nature of Li-ion batteries. This study presents a novel battery state estimation framework for EVs, integrating the Kolmogorov-Arnold Graph Neural Network (KAGN) with the Chaotic Secretary Bird Optimization (CSBO) algorithm to accurately predict the State of Charge (SoC), State of Health (SoH), and State of Temperature (SoT) of Li-ion batteries. The proposed model addresses the nonlinear and dynamic nature of battery behavior using graph-based feature extraction and evolutionary hyperparameter optimization. The KAGN effectively captures complex interdependencies among battery parameters like voltage, current, and temperature, while CSBO enhances convergence, accuracy, and generalization by avoiding local optima and tuning key parameters. The system validates its performance using the NASA Li-ion Battery Dataset and real-time experimental setups, achieves a Mean Absolute Error (MAE) of 0.0065, a Root Mean Square Error (RMSE) of 0.0085, and a high R<sup>2</sup> score of 0.9885, and outperforms existing Deep Learning (DL) models. Additionally, the architecture incorporates an interleaved dual-phase bidirectional DC-DC converter for efficient energy regulation, power distribution, and thermal stability. Through simulation and experimental validation, the KAGN-CSBO framework demonstrates superior accuracy and robustness in predicting battery states under varying operational conditions, enhancing battery safety, lifespan, and energy efficiency.</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

Advanced battery state estimation in electric vehicles using graph neural network and evolutionary optimization

  • 期刊论文

  • 发表日期：

    2026年7月2日

查看全文
  • R. Sundaramoorthi,
  • Chitraselvi S

摘要

The rapid shift to clean energy technologies has propelled the adoption of Electric Vehicles (EVs), necessitating advanced battery state estimation techniques to optimize lithium-ion (Li-ion) battery performance. Traditional estimation approaches struggle to accurately estimate battery states due to the complex and nonlinear nature of Li-ion batteries. This study presents a novel battery state estimation framework for EVs, integrating the Kolmogorov-Arnold Graph Neural Network (KAGN) with the Chaotic Secretary Bird Optimization (CSBO) algorithm to accurately predict the State of Charge (SoC), State of Health (SoH), and State of Temperature (SoT) of Li-ion batteries. The proposed model addresses the nonlinear and dynamic nature of battery behavior using graph-based feature extraction and evolutionary hyperparameter optimization. The KAGN effectively captures complex interdependencies among battery parameters like voltage, current, and temperature, while CSBO enhances convergence, accuracy, and generalization by avoiding local optima and tuning key parameters. The system validates its performance using the NASA Li-ion Battery Dataset and real-time experimental setups, achieves a Mean Absolute Error (MAE) of 0.0065, a Root Mean Square Error (RMSE) of 0.0085, and a high R2 score of 0.9885, and outperforms existing Deep Learning (DL) models. Additionally, the architecture incorporates an interleaved dual-phase bidirectional DC-DC converter for efficient energy regulation, power distribution, and thermal stability. Through simulation and experimental validation, the KAGN-CSBO framework demonstrates superior accuracy and robustness in predicting battery states under varying operational conditions, enhancing battery safety, lifespan, and energy efficiency.