Parametric estimation of fractional order impedance models to study battery aging

Electrochemical impedance spectroscopy (EIS) is a powerful data-driven technique for estimating the impedance of a Li-ion battery from current and voltage measurements. In classical EIS, the battery impedance is estimated nonparametrically, and the nonparametric impedance estimate is then often interpreted by means of an equivalent electrical circuit model (ECM), whose components relate to the physical processes occurring in the battery. However, by using such an ECM, with a Warburg element to model the diffusion, as an underlying fractional order parametric model, the battery impedance can also be estimated parametrically in the frequency domain. While the nonparametric estimate can only be evaluated in the discrete set of excited frequencies, the parametric estimate can be evaluated in every frequency of the frequency band of interest. Moreover, the parametric estimation is not limited to single sine or odd random phase multisine excitation signals, but also works for all other persistently exciting signals, like for example Gaussian white noise excitation signals. Both the nonparametric and parametric estimation methods are applied to battery cycling measurements, to study the effects of battery aging on the equivalent circuit components.