Investigating re-parametrization of electrochemical model-based battery management using real-world driving data

Li-ion batteries in electric vehicles must be utilized more efficiently to lower their economic and environmental cost. To achieve this increase in efficiency, it is of large interest to develop more thorough battery management that can predict internal states in online settings and update usage and control accordingly. Electrochemical models are an important tool in achieving this, and their implementation in battery management systems is the topic of ongoing research. However, electrochemical battery management relies on accurate parametrization and thus requires re-parametrization as a battery ages. We therefore studied viability of re-parametrization for electrochemical model-based battery management. To this end, we performed global sensitivity analysis on selected Doyle–Fuller–Newman model parameters using on-board current measurements. Representative driving data was collected from several types of heavy-duty vehicles. This elucidated which model parameters should be updated periodically to conserve model accuracy and which parameters are sensitive enough to be estimated from the on-board data. Additionally, we studied how parameter uncertainty affects estimation of internal states and highlight how model-based state estimation relying on a beginning-of-life parametrization degrades as electrochemical parameters change with aging.