基于数据驱动的多时间尺度锂离子电池状态评估技术研究

发布时间：2018-05-04 11:12

本文选题：锂离子电池 + 数据驱动　；参考：《北京交通大学》2017年博士论文

【摘要】：锂离子电池因其性能优良、绿色环保等优越性成为电动汽车的重要动力源,对锂离子电池的有效管理和应用也随之成为电动汽车发展的关键技术问题。从多时间尺度出发,研究锂离子电池状态评估技术对于确保锂离子电池应用的安全性和高效性而言意义重大。针对锂离子电池在不同使用阶段多时间尺度下面临的管理及应用问题,本论文从以下几个方面对电池状态评估技术展开研究:(1)针对微观时间尺度下的电池SOC估计问题,从数据驱动角度出发,全面分析SOC在线估计的三个步骤:OCV-SOC关系数据获取、电池建模及在线SOC观测。对锂离子电池分别进行基于小倍率充放电法和间隔静置法的OCV测试,对低温、常温和高温下的电池OCV-SOC映射关系及其温度依赖性进行对比分析,基于离线数据驱动方法建立考虑温度效应的电池模型并进行参数辨识,结合无迹卡尔曼滤波方法实现对电池SOC的在线估计。应用统计学中的拟合优度等指标对SOC估计结果进行评估,发现间隔静置测试法适合用于获取OCV-SOC关系数据。(2)针对微观时间尺度下的电池SOP估计问题,从数据驱动角度出发,分析了复杂工况下电池SOP影响因素的统计特性,基于数据驱动方法分别建立了电池SOP参数模型和非参数模型,应用均方根误差、平均绝对误差等统计学指标进行模型对比,对两类模型的适用范围进行讨论,提出参数模型与非参数模型结合的电池SOP估计方法,为BMS的功率分配策略提供依据。(3)针对宏观时间尺度下的电池寿命预测问题,从数据驱动角度出发,分析了电池内阻及容量性能的衰退趋势,通过数据拟合方式对电池容量和循环次数关系进行建模,基于运行数据结合贝叶斯理论对模型参数进行定期更新,实现对电池剩余使用寿命的精确预测,降低对实验室寿命建模结果的依赖性,减少前期寿命测试成本。(4)针对宏观时间尺度下对电池组内失效电池进行识别的问题,从数据角度出发,应用容量增量分析法(Incremental capacity analysis,ICA)对电动汽车的充电过程曲线进行分析,结合数据收缩方法提取IC曲线的特征参数,基于对时间截面上电池特征参数不一致性的分析,提出面向电动车用电池组的失效电池识别模型,模型分类精度高且输入参数可在线获取,符合电池应用的实际情况,工程推广价值高。(5)针对电动汽车退运电池梯次利用时的容量估计及筛选配组问题,从数据驱动角度出发,基于对大批量退运电池容量及内阻的统计特性分析,建立遗传算法与支持向量机相融合的电池容量快速估计模型;基于对退运电池性能数据的分析,结合主观的德尔菲打分和客观的灰色关联度测算,提出电池综合性能评价方法,为电池的梯次利用提供理论依据和工程应用指导。本论文提出的微观时间尺度下基于数据驱动的电池SOC及SOP估计方法、宏观时间尺度下基于数据驱动的寿命预测、失效电池识别方法及梯次利用电池容量估计和性能评价方法,分别解决了锂离子电池在微观和宏观两个时间尺度应用时所面临的关键技术问题,提高了电池利用的安全性和经济性,具有工程应用和推广价值。
[Abstract]:Lithium-ion battery has become an important power source for electric vehicles because of its excellent performance, green environment and other advantages. The effective management and application of lithium ion battery has also become a key technical problem for the development of electric vehicles. From the multi time scale, the state evaluation technology of lithium ion battery is studied to ensure the safety of lithium ion batteries. In view of the management and application problems of lithium ion batteries at different time scales at different stages of use, this paper studies battery status assessment technology from the following aspects: (1) aiming at the SOC estimation questions under the micro time scale, from the point of view of data drive, the analysis of SOC Online The three steps are estimated: OCV-SOC relationship data acquisition, battery modeling and on-line SOC observation. OCV tests on lithium ion batteries based on small rate charge discharge and spaced static method are carried out respectively. The OCV-SOC mapping relationship and temperature dependence of batteries at low temperature, normal temperature and high temperature are compared and analyzed, based on the off-line data driven method The battery model considering the temperature effect and parameter identification, the on-line estimation of the battery SOC is realized with the Untraced Calman filter. The results of SOC estimation are evaluated by the index of the goodness of fit in statistics. It is found that the interval static test method is suitable for obtaining the OCV-SOC data. (2) the battery for the micro time scale SOP estimation problem, from the data driven point of view, this paper analyzes the statistical characteristics of the influence factors of the battery SOP under the complex conditions. Based on the data driven method, the SOP parameter model and the non parametric model are established respectively. The statistical indexes such as the root mean square error and the mean absolute error are applied to the model comparison, and the application scope of the two types of models is carried out. The battery SOP estimation method combined with the parameter model and the nonparametric model is proposed to provide the basis for the power allocation strategy of BMS. (3) in view of the battery life prediction problem under the macro time scale, the decline trend of the battery internal resistance and capacity performance is analyzed from the data driving point of view, and the battery capacity and evidence-based method is carried out through the data fitting method. The relationship of ring times is modeled. Based on the operation data and Bayesian theory, the model parameters are updated regularly to realize the accurate prediction of the remaining life of the battery, reduce the dependence on the laboratory life modeling results and reduce the cost of the preliminary life test. (4) to identify the battery in the battery group under the macro time scale. From the point of view of data, Incremental capacity analysis (ICA) is applied to analyze the charging process curve of electric vehicle, and the characteristic parameters of IC curve are extracted with the method of data contraction. Based on the analysis of the inconsistency of the characteristic parameters of the battery on the time section, the battery group for electric vehicles is proposed. The model of failure battery recognition is of high accuracy and input parameters can be obtained online, which is in line with the actual situation of battery applications. (5) in view of the capacity estimation and selection of the battery cascade utilization of electric vehicles, from the data drive point of view, the system is based on the volume and internal resistance of large batch retreating batteries. Based on the analysis of the fusion of genetic algorithms and support vector machines, based on the analysis of the performance data of the retreated batteries, based on the subjective Delphy score and the objective grey correlation measurement, the comprehensive performance evaluation method of the battery is proposed, which provides the theoretical basis for the cascade utilization of the electric pool and the engineering application. In this paper, a data driven SOC and SOP estimation method based on data driven, data driven life prediction, failure battery identification and cascade battery capacity estimation and performance evaluation are used to solve the two time scale applications of lithium-ion batteries in microcosmic and macroscopic. The key technical problems faced by the company improve the safety and economy of battery utilization, and have the value of engineering application and popularization.

【学位授予单位】：北京交通大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TM912

【相似文献】