Scholarlink

<p>Lithium-ion batteries (LIBs) are widely used in electrochemical energy storage systems due to their high performance. However, aging over time makes it essential to estimate the state of health (SOH) of LIBs proactively to ensure a safe and reliable energy supply. Accurately predicting SOH is challenging, particularly with real-world noisy battery data. This study proposes a novel robust algorithm, named as&#xa0;the Mixture of Gaussian and Laplacian Extreme Learning Machine (MoGL-ELM), for improving SOH estimation of LIBs. The MoGL-ELM model enhances resilience and modeling capabilities in the presence of both Gaussian noise and non-Gaussian noise characterized as Laplacian noise. By leveraging the Expectation Maximization (EM) algorithm, the model optimizes the objective function, significantly improving estimation accuracy compared to traditional single-core Extreme Learning Machine (ELM) techniques. The effectiveness of the proposed algorithm was validated by using three benchmark datasets, the National Aeronautics and Space Administration (NASA), the Center for Advanced Life Cycle Engineering (CALCE) at the University of Maryland, and a second-life battery dataset. Results demonstrate better performance and generalization of the proposed model, achieving twice the estimation accuracy of existing machine learning (ML), deep learning (DL), and ELM methods. For noisy input data, the model achieved an average root mean square error (RMSE) of 1.87% and a mean absolute percentage error (MAPE) of 0.81% for different datasets under study. These results highlight the MoGL-ELM algorithm’s potential in advancing the reliability and performance of LIBs in various applications, contributing to the development of robust energy storage 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

State of health estimation for lithium-ion batteries using a hybrid Mixture of Gaussian and Laplacian extreme learning machine algorithm

  • 期刊论文

  • 发表日期：

    2025年4月15日

查看全文
  • Pallabi Kakati,
  • Devendra Dandotiya,
  • Rajiv Ranjan Singh

摘要

Lithium-ion batteries (LIBs) are widely used in electrochemical energy storage systems due to their high performance. However, aging over time makes it essential to estimate the state of health (SOH) of LIBs proactively to ensure a safe and reliable energy supply. Accurately predicting SOH is challenging, particularly with real-world noisy battery data. This study proposes a novel robust algorithm, named as the Mixture of Gaussian and Laplacian Extreme Learning Machine (MoGL-ELM), for improving SOH estimation of LIBs. The MoGL-ELM model enhances resilience and modeling capabilities in the presence of both Gaussian noise and non-Gaussian noise characterized as Laplacian noise. By leveraging the Expectation Maximization (EM) algorithm, the model optimizes the objective function, significantly improving estimation accuracy compared to traditional single-core Extreme Learning Machine (ELM) techniques. The effectiveness of the proposed algorithm was validated by using three benchmark datasets, the National Aeronautics and Space Administration (NASA), the Center for Advanced Life Cycle Engineering (CALCE) at the University of Maryland, and a second-life battery dataset. Results demonstrate better performance and generalization of the proposed model, achieving twice the estimation accuracy of existing machine learning (ML), deep learning (DL), and ELM methods. For noisy input data, the model achieved an average root mean square error (RMSE) of 1.87% and a mean absolute percentage error (MAPE) of 0.81% for different datasets under study. These results highlight the MoGL-ELM algorithm’s potential in advancing the reliability and performance of LIBs in various applications, contributing to the development of robust energy storage systems.