Lithium-ion battery state of health and failure analysis with mixture weibull and equivalent circuit model

Existing methods for interpreting Electrochemical Impedance Spectroscopy data involve various models, which face significant challenges in parameterization and physical interpretation and fail to comprehensively reflect the electrochemical behavior within batteries. To address these issues, this study proposes a Temperature-Controlled Second-Order R-CPE Equivalent Circuit Model to capture the non-ideal capacitive characteristics of electrode surfaces. Additionally, the study employs a Copula based Joint Mixture Weibull Model and multi-output Gaussian Process Regression to enhance the precision in capturing the distribution of battery electrochemical parameters and predict SoH curves. Experimental validation shows that the model used in this article has an average RMSE error of 8.5%, and the prediction of the SoH curve after the 100th cycle can achieve an average RMSE error of 9.2%. These findings provide insightful implications for understanding the electrochemical complexities and parameter interdependencies in the battery aging process, offering a robust framework for future research in battery diagnostics. [Display omitted] •Constructing the Temperature-Controlled Second-Order R-CPE Equivalent Circuit Model•Using Copula to construct a joint mixture model for parameter variation capture•SM-LMC Multi-Output Gaussian Process Regression for SoH curve prediction•In-depth degradation mechanism analysis based on joint mixture Weibull model Electrochemistry; Energy systems