High precision state of health estimation of lithium-ion batteries based on strong correlation aging feature extraction and improved hybrid kernel function least squares support vector regression machine model.

Feng, Renjun; Wang, Shunli; Yu, Chunmei; Hai, Nan; Fernandez, Carlos

Abstract

The state of health (SOH) of lithium-ion batteries plays a crucial role in maintaining the stability of electric vehicle systems. To address the issue of low accuracy in existing prediction models, this article introduces an enhanced grey wolf algorithm for optimizing the hybrid kernel least squares support vector regression machine used in lithium-ion battery SOH prediction. This research extracted four key health features from the raw data of each battery in the Cycle dataset, which is publicly accessible. Data preprocessing of health features involved Pearson correlation analysis and Hampel filtering techniques. The framework of least squares support vector regression constructs a hybrid kernel function of polynomial kernel function and radial basis function. The integration of differential evolution and the law of survival of the fittest into the grey wolf algorithm enhances its optimization ability. The improved grey wolf algorithm optimizes the parameters of the hybrid kernel least squares support vector regression machine, improving the accuracy and robustness of the model. After data validation, it is known that the optimal average absolute error value predicted by the model can reach 0.32%. This indicates that the proposed method is effective and feasible.

Citation

FENG, R., WANG, S., YU, C., HAI, N. and FERNANDEZ, C. 2024. High precision state of health estimation of lithium-ion batteries based on strong correlation aging feature extraction and improved hybrid kernel function least squares support vector regression machine model. Journal of energy storage [online], 90(A), article number 111834. Available from: https://doi.org/10.1016/j.est.2024.111834

• Journal Article Type: Article
• Acceptance Date: Apr 20, 2024
• Online Publication Date: May 4, 2024
• Publication Date: Jun 15, 2024
• Deposit Date: May 6, 2024
• Publicly Available Date: May 5, 2025
• Journal: Journal of energy storage
• Print ISSN: 2352-152X
• Electronic ISSN: 2352-1538
• Publisher: Elsevier
• Peer Reviewed: Peer Reviewed
• Volume: 90
• Issue: A
• Article Number: 111834
• DOI: https://doi.org/10.1016/j.est.2024.111834
• Keywords: Lithium-ion battery; State of health; Hybrid kernel function; Grey wolf algorithm; Least squares support vector regression machine
• Public URL: https://rgu-repository.worktribe.com/output/2332468