SOC fusion estimation for lithium battery using dual OCV-SOC relationships combined with fractional order extended Kalman filter

Abstract Using a single or averaged OCV-SOC (open circuit voltage VS. state of charge) relationship obtained under charging or discharging condition will deteriorate battery SOC estimation of model-based methods. To address the issues, this paper proposes a novel SOC fusion estimation method considering charging and discharging OCV-SOC relationships. This method fully considers the actual working state of batteries in electric vehicles, which is beneficial for improving SOC estimation results. The SOC is respectively estimated using OCV-SOC relationship under discharging and charging conditions combined with fractional-order extended Kalman filter (FOEKF) algorithm. Both SOC estimation results using charging and discharging OCV-SOC relationships are obtained and then fuse them by the weighted summation for getting the final SOC. The fusion weights are established using the predicted voltage error. The US06 Highway Driving Schedule and Beijing Dynamic Stress Test (BJDST) are used for verifying the proposed method. The SOC root-mean square errors (RMSEs) compared to the regular technique that uses a single OCV-SOC relationship obtained under discharging condition decrease from 1.65% and 1.70–0.71% and 0.70% under the above two tests. The applicability of a certain OCV-SOC relationship for estimating SOC under all driving conditions is limited. Using multiple OCV-SOC relationships obtained under different conditions for SOC fusion estimation can effectively improve estimation accuracy. We have demonstrated the correctness and effectiveness of this paper’s ideas using BJDST test and US06 test.