On-line parameter, state-of-charge and aging estimation of Li-ion batteries

This paper presents an on-line model identification method for Li-ion battery parameters that combines high accuracy and low computational complexity. Experimental results show that modeling errors are smaller than 1% throughout the feasible operating range. The identified model is used in a state observer - an Extended Kalman Filter (EKF) - to obtain an indication about the battery State of Charge (SoC). A novel method to estimate the actual battery capacity on-line, based on the data from the state observer is presented. Based on the real battery capacity, an indication about the State of Health (SoH) can be given. Simulation and experimental results are presented to validate the proposed methodology. Battery capacity estimation errors under 4% are achieved by using only 30 minutes of data (battery voltage and current measurements) acquired during normal driving.