A Computational Framework for Identifiability and Ill-Conditioning Analysis of Lithium-Ion Battery Models

Abstract

The lack of informative experimental data and the complexity of first-principles battery models make the recovery of kinetic, transport, and thermodynamic parameters complicated. We present a computational framework that combines sensitivity, singular value, and Monte Carlo analysis to explore how different sources of experimental data affect parameter structural ill-conditioning and identifiability. Our study is conducted on a modified version of the Doyle–Fuller–Newman model. We demonstrate that the use of voltage discharge curves only enables the identification of a small parameter subset, regardless of the number of experiments considered. Furthermore, we show that the inclusion of a single electrolyte concentration measurement significantly aids identifiability and mitigates ill-conditioning.