Battery state-of-charge estimation based on sigma point Kalman filter

In this paper, the nonlinear operating performance of the valve regulated lead acid (VRLA) battery is investigated, and a comprehensive battery model is established based on 2^nd-order Randle model and ampere-hour counting concept; then a multidimensional system states estimation method is proposed based on sigma point Kalman filter (SPKF), and battery state of charge (SOC) is included in the states; finally, based on the data collected from experiments, battery SOC is estimated offline using SPKF algorithm by adopting 1^st-order Thevenin model and 2^nd-order Randle model, respectively. SOC estimation results show that, compared with Thevenin model, the model error can be suppressed more effectively by using Randle model in the comprehensive model, also battery SOC can be estimated more accurately.