Optimization-based reduction and Padé approximants for lithium-ion battery cell models with degradation

This works studies the model reduction of electrochemical lithium-ion battery cell models. Battery cell modeling is crucial to predicting battery behavior under different conditions and enables battery management system (BMS) control algorithms to ensure efficient and reliable operation. The finite difference method (FDM), commonly used for describing nonlinear battery dynamics, requires a high number of states to accurately capture the battery behavior. When computational resources are limited, model reduction is necessary to increase computational tractability and runtime. The literature presents a variety of reduced-order modeling techniques, though these techniques do not commonly incorporate degradation dynamics. To address this gap, this study explores model reduction with the inclusion of degradation effects. Two methods are used to accomplish this. The first uses optimization-based techniques to identify reduced-order models, and the second extends the Padé approximation technique to incorporate degradation terms found in the boundary conditions of the single-particle model (SPM). The models produced are compared for different operating conditions of a lithium-ion battery cell. Simulated case studies indicate that both reduction techniques are able to capture the cell dynamics with degradation only using a few states.