The main focus of the battery management system is the estimation of the battery's State of Charge (SOC), which is an indicator to determine the driving range of an electric vehicle. The Extended Kalman Filter (EKF) algorithm is the most promising for SOC estimation when the system is running. The EKF state estimation algorithm is sensitive to the process noise covariance matrix Q and measurement noise covariance matrix R. Inappropriate noise covariance matrices reduce the accuracy and cause divergence in state estimation. This paper uses the Sunflower Optimization algorithm (SFO) to find the optimal values of the noise covariance matrices before applying EKF for online SOC estimation. It is clearly stated that the iterative SFO does not affect the instantaneous response of EKF in the online estimate because the SFO is only performed once to determine the optimal values. The effectiveness of the proposed identification is examined through the constant discharge rate test and dynamic driving profile. The proposed algorithm's performance indices, such as maximum error, Mean Absolute Error, Mean Square Error, and Root Mean Square Error of SOC estimation, are low compared to the trial-error method, genetic algorithm, and adaptive extended Kalman filter. Besides accuracy, the proposed method quickly converges even when the initial SOC is inaccurate. The feasibility and benefits of the proposed SFO-EKF algorithm are demonstrated by qualitative and quantitative findings on five challenging datasets. To assess the performance of the SFO-EKF for SOC estimation in constant-current discharge and extreme drive cycle situations, three public datasets and two datasheets are employed. The simulation results show that the proposed method has high accuracy and a better convergence rate in estimating SOC under static and dynamic operating conditions.
Read full abstract