摘要
锂离子电池具有无记忆效应、轻量化、环保等特点,因此常作为电动交通工具、电子设备的能源来源,并适用于各种规模的能源存储。在锂离子电池管理系统中,电池的荷电状态(state of charge,SOC)是最关键的指标之一,其准确估计对于实现电池系统的高效能量管理和优化控制至关重要。因此本文提出了一种基于动态噪声自适应无迹卡尔曼滤波的SOC估计方法。首先,通过间歇放电实验获取电池不同SOC下的开路电压,并进一步拟合得到电池的OCV-SOC曲线,接着采用二阶RC等效电路模型对锂离子电池建模,然后通过混合功率脉冲特性工况测试对电池模型参数进行辨识。由于实际应用中锂离子电池为非线性系统且SOC估计精度容易受到噪声的影响,本文在卡尔曼滤波算法的基础上采用无迹变换处理,加入噪声自适应过程,以实现噪声特性自适应估计,动态调整测量噪声与过程噪声,提高算法鲁棒性以及估计精度。最后选取DST与FUDS工况进行验证,结果表明在不同工况下动态噪声自适应无迹卡尔曼滤波算法的估计平均绝对误差、最大绝对误差以及均方根误差相较于自适应无迹卡尔曼滤波、无迹卡尔曼滤波算法均有降低,其平均绝对误差小于0.59%。本文提出的动态噪声自适应无迹卡尔曼滤波算法能够更准确地估计锂离子电池SOC。
Lithium-ion batteries,known for their lack of memory effect,lightweight nature,and environmental friendliness,are widely used as energy sources in electric vehicles,electronic devices,and various scales of energy storage.In a lithium-ion battery management system,the state of charge(SOC)is a critical indicator,and its accurate estimation is essential for efficient energy management and optimal control of the battery system.This paper proposes an SOC estimation method based on the dynamic noise-adaptive unscented Kalman filter(DN-AUKF).The open circuit voltage(OCV)of the battery at different SOCs is first obtained through intermittent discharge experiments and fitted to derive the OCV-SOC curves.The lithium-ion battery is then modeled using a second-order RC equivalent circuit model,with parameter identification conducted via the hybrid pulse power characterization(HPPC)test.Recognizing that the SOC estimation in lithium-ion batteries is a nonlinear process and highly susceptible to operational noise,this study utilizes a traceless transform based on the Kalman filter(KF)to address system nonlinearity,integrates an adaptive factor for noise characteristic estimation,and dynamically adjusts the process noise covariance to enhance algorithm robustness and estimation accuracy.The proposed algorithm is validated under dynamic stress test(DST)and federal urban driving schedule(FUDS)conditions,demonstrating that the DN-AUKF algorithm significantly improves average estimation error,maximum error,and root mean square error compared to the adaptive unscented Kalman filter(AUKF)and unscented Kalman filter(UKF)algorithms under various conditions.The DN-AUKF's average absolute estimation error is less than 0.51%,indicating its superior performance in accurately estimating the SOC of lithium-ion batteries,even under extreme conditions such as low power and high-rate charge and discharge scenarios.
作者
尹康涌
孙磊
李浩秒
郭东亮
肖鹏
王康丽
蒋凯
YIN Kangyong;SUN Lei;LI Haomiao;GUO Dongliang;XIAO Peng;WANG Kangli;JIANG Kai(Electric Power Scientific Research Institute of State Grid Jiangsu Electric Power Co,Nanjing 211103,Jiangsu,China;School of Electrical and Electronic Engineering,Huazhong University of Science and Technology,Wuhan 430074,Hubei,China)
出处
《储能科学与技术》
CAS
CSCD
北大核心
2024年第11期4065-4077,共13页
Energy Storage Science and Technology
基金
国家电网有限公司科技项目(5419-202199552A-0-5-ZN)。
