Hierarchically identifying 31 electrochemical model parameters only via battery single dynamic current profiles

Accurate parameters identification is pivotal for lithium-ion battery modeling, monitoring and control. Electrochemical model has clearly physical interpretation and embodies high accuracy, while non-destructive identification for full model parameters only using dynamic data still lies as an intractable issue. To fill this research gap, here we simplify classical pseudo-two-dimensional model as parabolic polynomial approximation model, and innovatively develop a hierarchical identification framework only via single dynamic current profiles. Our framework, without the help of sensitivity analysis, enables hierarchical identification for stoichiometric ratios, capacity-related parameters, and the remaining parameters. By statistics, total 31 model parameters can be non-destructively identified via our framework, which almost covers full parameters involved in popular electrochemical models. The cross-validation based on three typical dynamic scenarios demonstrates that our framework significantly benefits accurate battery electrochemical modeling and also has well scenario compatibility, where the root-mean-square-error and mean-absolute-percentage-error between measured and modeled voltage can be easily limited below 11 mV and 0.3% respectively. Moreover, through qualitative comparison with existing state-of-the-art identification methods, our framework empowers non-destructive identification for more model parameters with less scenario requirements. Those findings enlighten a novel and promising solution for electrochemical model parameter identification via real-time sampled dynamic data.