Parameter estimation of the Doyle-Fuller-Newman model for Lithium-ion batteries by parameter normalization, grouping, and sensitivity analysis

Using electrochemistry-based battery models in battery management systems remains challenging due to the difficulty of uniquely determining all model parameters. This paper proposes a model parameterization approach of the Doyle-Fuller-Newman (DFN) model, by first reparameterizing the DFN model through normalization and grouping, followed by a sensitivity analysis and a parameter estimation procedure. In the parameter estimation procedure, we show the influence of the number of estimated parameters, as well as the influence of the data length of the identification data, on the obtained model accuracy. Additionally, the model with parameters obtained using the proposed parameterization approach is compared to a model whose parameters have been obtained using cell teardown. Finally, the consistency and accuracy of the parameter estimation procedure is analyzed by applying the estimation routine to a synthetic cell, represented by a DFN model with randomly chosen parameters. The results of this analysis show that the parameter estimation approach using current/voltage data can lead to a significantly better output accuracy, while it might not lead to physically meaningful parameters. This motivates the need for an approach that combines both and where cell tear-down can assist the parameter estimation using current/voltage data in achieving physically meaningful parameters.

Automated summary

This paper presents a new approach for parameterizing battery models, which includes reparameterizing, sensitivity analysis, and parameter estimation to achieve more accurate models. The study shows that using current/voltage data for parameter estimation can improve accuracy, but may not result in physically meaningful parameters.