基于AUKF-BP神经网络的锂电池SOC估算

Estimation of the SOC of a battery based on the AUKF-BP algorithm

电池荷电状态(SOC)的估算作为车载电池管理系统(BMS)的核心技术之一,其准确预估可以延长电池使用寿命,确保整车的正常行驶。本文以锂离子电池为研究对象,提出一种基于自适应无迹卡尔曼滤波(AUKF)和BP神经网络相结合的电池SOC估算方法。该方法通过采样策略自适应性提高了UKF的估算精度,并利用训练好的BP神经网络SOC输出值作为UKF的观测量。使用Arbin电池测试平台采集的不同温度下的混合工况和FUDS工况电池测试数据为基础,比较AUKF-BP算法和BP算法的准确性。结果表明,不同温度下的AUKF-BP算法的平均均值误差为0.82%,BP算法的平均均值误差为1.63%,基于AUKF-BP的SOC估计方法具有更高的鲁棒性和准确性。

The estimation of the battery state of charge(SOC)is a core feature of the on-board battery management system(BMS).Its accurate estimation can prolong the service life of a battery and ensure the normal driving of a vehicle.Using lithium-ion batteries as the model,this paper proposes a battery SOC estimation method based on the combination of the adaptive unscented Kalman filter(AUKF)and the BP neural network.This method improves the estimation accuracy of UKF through adaptive sampling and uses the SOC output value of the trained BP neural network for the observation of UKF.Based on the battery test data under mixed working conditions and the FUDS working conditions collected by the Arbin battery test platform at varied temperatures(0℃,25℃,and 40℃),the accuracy of the AUKF-BP algorithm versus the BP algorithm were evaluated.The results indicate that the average mean error of the AUKF-BP algorithm at different temperatures was 0.82%,and the average mean error of the BP algorithm was 1.63%.Overall,an SOC estimation method based on theAUKF-BP algorithm is the most accurate.