An improved parameter identification method considering multi-timescale characteristics of lithium-ion batteries

To monitor and predict battery states, a battery model with accurate model parameters is important to battery management systems (BMS). However, for multi-timescale dynamic characteristics, the precision and adaptability of parameter identification of the Li-ion battery model is unsatisfactory up to now. In this paper, an improved parameter identification algorithm is proposed combining fixed memory recursive least squares (FMRLS) and fading extended Kalman filter (FEKF) which are used to obtain the fast dynamic (FD) and slow dynamic (SD) parameters of equivalent circuit model (ECM) respectively. Open-circuit voltage (OCV) is identified as a component of the SD part because of its slow dynamic nature in this algorithm. Federal urban driving schedule (FUDS) and dynamic stress test (DST) tests with different initial state of charge (SOC) and temperatures were employed for verifications, and the results show that the algorithm can track the battery terminal voltage in time and the root mean square error (RMSE) is as low as 1 mV. Meanwhile, the results reveal that the advanced SOC-OCV tests can be avoided indeed, and model parameters identified by this algorithm have good robustness in different temperatures and high consistency in different operating conditions which are significantly better than conventional algorithms.