Scholarlink

<p>State of Charge (SoC) needs to be SoC needs to be accurately measured in order to guarantee the safety, reliability, and functionality of electric powered systems including electric vehicles and energy storage units especially when the Lithium-Ion (Li-Ion) battery is involved. The nonlinear and time varying characteristics of the Li-Ion batteries pose challenges to conventional SoC estimation methods, which provide lower accuracy with changing load profiles and change in environmental conditions. In order to address such constraints, this paper presents a hybrid SoC estimation model, combining a Nonlinear Auto-Regressive Moving Average with exogenous inputs (NARMAX) model and an Adaptive Sunflower Optimization Algorithm (ASOA). The algorithm uses Principal Component Analysis (PCA) to decrease the number of dimensions and find significant features between measurements of voltage, current, and temperature in order to obtain the appropriate variables and derived indicators. The modelling process of the NARMAX-ASOA methodology will improve the precision of modelling as it can consider the complex interdependences throughout the time and parameters within the model are optimally optimized. The results demonstrate the effectiveness of the proposed solution of the performance evaluation, which are the root mean square error (RMSE) 1.12, the mean absolute error (MAE) 0.87, the mean absolute percentage error (MAPE) 1.53, and the <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\:{R}^{2}\)</EquationSource> </InlineEquation>of 0.994. Generally, the results confirm that the framework is a viable and sound method in managing Li-Ion batteries in the real world.</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

Machine learning-driven SoC estimation for Li-Ion batteries using optimized nonlinear predictive modelling

  • 期刊论文

  • 发表日期：

    2026年4月29日

查看全文
  • Bhanu S. Vijayanand,
  • Shrishail C. Prabhakar,
  • R. Jayaganthan

摘要

State of Charge (SoC) needs to be SoC needs to be accurately measured in order to guarantee the safety, reliability, and functionality of electric powered systems including electric vehicles and energy storage units especially when the Lithium-Ion (Li-Ion) battery is involved. The nonlinear and time varying characteristics of the Li-Ion batteries pose challenges to conventional SoC estimation methods, which provide lower accuracy with changing load profiles and change in environmental conditions. In order to address such constraints, this paper presents a hybrid SoC estimation model, combining a Nonlinear Auto-Regressive Moving Average with exogenous inputs (NARMAX) model and an Adaptive Sunflower Optimization Algorithm (ASOA). The algorithm uses Principal Component Analysis (PCA) to decrease the number of dimensions and find significant features between measurements of voltage, current, and temperature in order to obtain the appropriate variables and derived indicators. The modelling process of the NARMAX-ASOA methodology will improve the precision of modelling as it can consider the complex interdependences throughout the time and parameters within the model are optimally optimized. The results demonstrate the effectiveness of the proposed solution of the performance evaluation, which are the root mean square error (RMSE) 1.12, the mean absolute error (MAE) 0.87, the mean absolute percentage error (MAPE) 1.53, and the \(\:{R}^{2}\) of 0.994. Generally, the results confirm that the framework is a viable and sound method in managing Li-Ion batteries in the real world.