Strain feature-assisted state of health estimation for lithium-ion batteries

Accurate state of health (SOH) estimation of lithium-ion batteries is crucial for safety. The accuracy of estimation largely depends on the relevance of health features (HFs). However, conventional electrical features-based methods face challenges in capturing cell inconsistencies in batteries and are prone to electromagnetic interference, hindering estimation accuracy. Therefore, this paper proposes a strain features-assisted SOH estimation framework that enhances accuracy by integrating electrical and strain features. Firstly, an aging experiment is conducted on eight cells, and a dataset comprising strain and electrical signals is constructed. We systematically summarize electrical signal characteristics, identifying 11 features. An innovative method for extracting strain features is proposed based on an in-depth analysis of signal properties. Subsequently, Pearson correlation analysis is employed to quantitatively identify key features which are highly correlated with SOH. Finally, strain features are integrated with five groups of electrical features and input into a Gaussian Process Regression (GPR) for SOH estimation. The results demonstrate that integrating electrical and strain features reduces the average root mean square error (RMSE) by 25.34% compared to using electrical features alone for SOH estimation. This highlights the effectiveness of incorporating strain features, consistently improving accuracy regardless of the specific combination of electrical features.