Li-ion dynamics and state of charge estimation

This paper focuses on real-time estimation of Li-ion State of Charge (SoC). A first-principles model validated by experimental data from literature is chosen to mimic a real Li-ion cell. Its impedance responses at different SoCs are studied by a simulated electrochemical impedance spectroscopy (EIS). An equivalent circuit model is developed for estimator design in which the parameters (including lumped series resistances R1, lumped interfacial resistances R2 and time constant τ) are derived from system identification and compared with the EIS results. The estimator is designed using extended Kalman filtering (EKF) and is implemented in the first-principles model. It is demonstrated by computer simulation that the SoC during charge/discharge cycles can be estimated with a relative error <3%. The accuracy of SoC tracking is improved if it is jointly estimated along with either R1 or R2 given that these model parameters vary with SoC as revealed by EIS.