Current intelligent algorithms such as neural networks, fuzzy control, support vector machines, etc. can obtain higher estimation accuracy by describing the nonlinear relationship between SOC and battery voltage, current, and temperature, but they require a large amount of It is difficult to apply the training data as support to the online estimation of the entire vehicle. Extended Kalman Filter (EKF) is an efficient linear filtering and prediction method based on the battery equivalent circuit model. It has been widely used in battery SOC estimation in recent years. As a recursive linear minimum variance estimator, EKF performs real-time estimation by comparing the real-time observation value with the estimated value at the previous moment. It can dynamically track the true value of SOC and is more suitable for electric vehicles with severe current fluctuations. However, EKF can only obtain ideal SOC estimation accuracy if the battery model is accurate. Changes in battery model parameters will bring obvious errors to the estimation.
[0004] The parameter identification of the battery equivalent circuit model generally adopts an offline method, that is, before the battery box is loaded into the vehicle, the battery is subjected to a standard charge and discharge pulse experiment to obtain the relevant battery parameters, and the EKF is used to estimate the SOC. During the process, battery parameters are calculated as fixed constants. However, with the long-term use of electric vehicles, battery aging will cause changes in battery model parameters and nonlinear attenuation of battery capacity. If EKF still uses the initial battery parameters in the calculation process, it will bring serious errors to the estimation. The offline parameter identification method requires disassembling the battery box from the vehicle and using external equipment to conduct charging and discharging experiments on the battery to recalibrate the battery parameters. The disassembly and assembly process is quite cumbersome and difficult to operate.
[0005] In summary, the power battery is a nonlinear and time-varying system. If the extended Kalman filter algorithm always uses fixed battery model parameters as state variables to estimate SOC , as the battery ages, the estimation error will become larger and larger, and it cannot meet the needs of the entire vehicle; if the battery model parameters and battery capacity are recalibrated through offline charge and discharge experiments, the battery box of the entire vehicle needs to be disassembled.