电动汽车锂离子动力电池组健康状态估计方法的研究

发布时间：2017-12-30 19:45

本文关键词：电动汽车锂离子动力电池组健康状态估计方法的研究　出处：《青岛科技大学》2017年硕士论文　论文类型：学位论文

【摘要】：作为新一代电动汽车的理想动力源,锂离子动力电池组的健康状态在电动汽车的实际运行过程中起着至关重要的作用。为保证电动汽车在运行过程中的安全性和稳定性,需要对其车载动力电池进行必要的研究和管理。论文以锂离子动力电池组为研究对象,对电动汽车锂离子动力电池组SOH的估计方法进行了研究,经过大量实验和仿真研究了以下内容:(1)首先分析了锂离子电池的特性,包括锂离子电池的特点、结构和工作原理,并给出了锂离子动力电池SOH的评价指标。然后介绍了影响锂离子电池组SOH的因素,从单体电池不一致性和单体电池连接方式两个角度分析了它们对电池组SOH的影响,并重点剖析了单体电池不同的连接方式对电池组SOH性能可靠性的影响。由于电池的内阻既能够与电池电压、电流等其他相关参数联系起来,又能很好体现电池特性的变化,因此本文选择内阻的变化来作为电池SOH的表征参量。(2)在分析了几种单体电池模型后发现,电池等效物理模型可以体现电池的物理特性,也能与其化学特性建立一定联系。因此在该模型基础上,考虑将两个锂离子单体电池并联后作为一个简化的子系统,而N个这样的子系统串联,就可以构成所要研究的动力电池组等效模型,然后再对锂离子电池组进行数学建模及实验分析。接下来选用50AH/3.6V型号的锂离子电池进行电池性能参数的测量试验,依据测量所得的实验数据,采用最小二乘参数辨识方法进行模型参数辨识。最后在Matlab中利用最小二乘拟合工具验证了模型参数辨识结果的可行性。(3)结合粒子滤波算法理论与所建立的电池组模型对电池组内阻进行了跟踪预测,但该算法存在严重的粒子退化现象。为克服这一现象,本文尝试将支持向量机引入粒子滤波算法中,增加粒子多样性以提高对电池组SOH的预测精度。最后对两种算法下的实验结果进行了对比分析,实验表明SVM-PF算法下的跟踪曲线整体波动幅度更小,其平稳性与预测精度明显优于粒子滤波算法,在一定程度上说明了SVM-PF算法对电池组内阻跟踪预测的有效性与优越性。
[Abstract]:As a new generation of electric vehicles, the ideal power source. The healthy state of Li-ion battery pack plays an important role in the actual operation of electric vehicle. In order to ensure the safety and stability of electric vehicle during operation. It is necessary to study and manage the on-board battery. In this paper, the SOH estimation method of Li-ion battery pack for electric vehicle is studied by taking the Li-ion battery pack as the research object. Firstly, the characteristics of lithium ion battery, including the characteristics, structure and working principle of lithium ion battery are analyzed. The evaluation index of SOH of Li-ion battery is given, and the factors influencing SOH of Li-ion battery are introduced. The influence of single cell inconsistency and single cell connection mode on the SOH of the battery pack is analyzed. The effect of different connection modes of single cell on the reliability of battery pack SOH is analyzed, because the internal resistance of the battery can be related to other related parameters such as battery voltage, current and so on. It can well reflect the change of battery characteristics, so this paper chooses the change of internal resistance as the characterization parameter of SOH.) after analyzing several kinds of single cell model, we find out. Battery equivalent physical model can reflect the physical characteristics of the battery, but also can establish a certain relationship with its chemical characteristics, so on the basis of the model. Considering two lithium ion monomer cells in parallel as a simplified subsystem, and N such subsystems in series, we can construct the equivalent model of power battery. Then the mathematical modeling and experimental analysis of the lithium ion battery pack. Then 50 AH / 3.6 V lithium ion battery was selected to measure the battery performance parameters according to the measured experimental data. The method of least square parameter identification is used to identify the model parameters. Finally, the feasibility of the model parameter identification results is verified by using the least square fitting tool in Matlab. Combined with the theory of particle filter algorithm and the established battery pack model, the internal resistance of the battery pack was tracked and predicted. However, this algorithm has serious particle degradation phenomenon. In order to overcome this phenomenon, this paper attempts to introduce support vector machine into particle filter algorithm. Increasing particle diversity to improve the prediction accuracy of battery pack SOH. Finally, the experimental results of the two algorithms are compared and analyzed. Experiments show that the overall fluctuation of the tracking curve under the SVM-PF algorithm is smaller, and its smoothness and prediction accuracy are obviously better than that of the particle filter algorithm. To a certain extent, the effectiveness and superiority of SVM-PF algorithm in the prediction of battery pack internal resistance tracking are demonstrated.
【学位授予单位】：青岛科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U469.72;TP18

【参考文献】