基于GA和PSO的电动客车锂离子电池SOC估计

SOC estimation of Li-ion battery in electric bus based on GA and PSO

以广州某巴士企业电动公交在实际运行和停车充电状态下的电压、电流和荷电状态(SOC)的数据,分别建立基于支持向量回归机(SVR)的锂离子电池放电和充电的SOC估计模型,并利用网格搜索法(GS)、遗传算法(GA)和粒子群算法(PSO)进行参数优化,对比估计精度和拟合优度。处理放电阶段数据时,基于PSO优化后的SOC估计模型误差为2.39%,拟合优度为0.913,均优于其他算法;处理充电阶段数据时,基于GA优化后的SOC估计模型误差为0.16%,拟合优度为0.990,优化效果最好。针对不同阶段的SOC估计,可采用不同的算法来优化估计模型,以提高精度和拟合优度。

The voltage,current and state of charge(SOC)data in the case of actual operation and charging state while stopping of electric bus from a Guangzhou bus company were selected,the SOC estimation model for Li-ion battery based on support vector regression(SVR)under discharge and charge was built.The grid search(GS),genetic algorithm(GA)and particle swarm optimization(PSO)were used to optimize the parameters,the estimation accuracy and goodness of fit were compared.When processing the data in discharge stage,the model estimation error after optimization based on PSO was 2.39%,the goodness of fit was 0.913,better than other algorithms.When processing the data in charge stage,the estimation error of the model optimized based on GA was 0.16%,the goodness of fit was 0.990,which showed the best optimization effect.For SOC estimation at different stages,different algorithms could be used to optimize the estimation model respectively to improve the accuracy and goodness of fit.