基于分数阶多新息无迹卡尔曼滤波算法的锂电池SOC估计

Lithium battery SOC estimation based on fractional order multi innovation unscented Kalman filtering algorithm

针对锂电池荷电状态(State Of Charge,SOC)估计时常用的整数阶等效电路模型无法精准反映电池极化反应和提高在噪声干扰下全生命周期SOC的估计准确度问题,在二阶RC等效电路模型的基础上建立分数阶模型,并采用遗传(GA)算法对其进行参数辨识,从而增强参数辨识的鲁棒性。最后在传统的无迹卡尔曼滤波(Unscented Kalman Filter,UKF)算法的基础上引入了多新息理论,提出了一种基于分数阶多新息无迹卡尔曼滤波(Fractional Order Multi Innovation Unscented Kalman Filtering,FOMIUKF)算法来实现对锂电池SOC的实时估计,最后通过搭建仿真模型验证了基于GA分数阶锂电池等效模型的准确性和可靠性,并进行了基于分数阶无迹卡尔曼滤波(Fractional Order Unscented Kalman Filtering,FOUKF)算法、FOMIUKF算法的锂电池SOC估计对比分析,发现FOMIUKF算法估计准确度更高,其估计误差仅为1%。

In response to the problem that the commonly used integer order equivalent circuit model for estimating the state of charge(SOC)of lithium batteries cannot accurately reflect the polarization response of the battery and improve the estimation accuracy of SOC throughout the entire life cycle under noise interference,a fractional order model is established on the basis of the second-order RC equivalent circuit model,and genetic algorithm(GA)is used to identify its parameters,thereby enhancing the robustness of parameter identification.Finally,based on the traditional UKF algorithm,the theory of multiple innovations was introduced,and a fractional order multi innovation unscented Kalman filtering(FOMIUKF)algorithm was proposed to achieve real-time estimation of the SOC of lithium batteries.Finally,a simulation model was built to verify the accuracy and reliability of the GA fractional order lithium battery equivalent model,and FOUKF was used Comparative analysis of lithium battery SOC estimation using FMIUKF algorithm shows that FOMIUKF algorithm has higher estimation accuracy,with an estimation error of only 1%.