State of charge estimation of vanadium redox battery based on improved extended Kalman filter

Abstract The accurate state of charge (SOC) estimation can protect the battery from overcharging and over-discharging, and it is useful to make an effective dispatching strategy. The extended Kalman filter (EKF) method is used to estimate SOC widely. But it does not consider the SOC constraints. Moreover, the convergence is influenced by the uncertain initial SOC, which may lead to false alarm, unwanted operation of protection, error dispatching and poor robustness of the system. This paper presents an improved extended Kalman filter (IEKF) method to estimate SOC for vanadium redox battery (VRB) by introducing a gain factor. It can be adjusted automatically according to the output error and SOC boundary. To implement IEKF estimator, a VRB state space model is established and its parameters are identified by recursive least square (RLS) method. Then a VRB of 5kW/30kWh experimental platform is built. Finally, the IEKF method is validated and compared with EKF against unknown initial value through the experiments. The results have shown that IEKF method is superior to EKF in terms of accuracy, convergence speed and robustness. And the estimated SOC remains bounded by using IEKF method. It is more suitable for SOC estimation than EKF algorithm in the industrial applications.