Interpretable Real-Time Modeling of the Diffusion Overpotential in Lithium Batteries

Abstract

Fractional-order dynamics can form physically-interpretable equivalent-circuit models of the diffusion overpotential in lithium-ion batteries but have complex formulations in the time-domain. Meanwhile, resistor-capacitor circuits have simple implementations but little physical meaning. Thus we propose a discrete-time state-space diffusion model, named ‘receding-horizon diffusion’ (RHD). It combines physical interpretability with computational simplicity. Analogous to theWarburg element in impedance spectroscopy, the RHD constant is explicitly linked with the lithium ion diffusion coefficient. Fivefold validation of the RHD model using simulated and experimental from lithium NiMnCo and NiCoAl cells up to 3 C-rate, temperatures from 0 to 25°C, and wide ranges of states of health and charge. The model has less than 1% modelling error. Ohmic, charge-transfer, and diffusion overpotentials are tracked in real time. The RHD model could be integrated into battery management systems in electric vehicles and used in standard state estimation techniques.