锂离子电池容量损失预测及健康状态估计研究

发布时间：2018-03-05 04:12

本文选题：锂离子电池　切入点：容量损失预测　出处：《哈尔滨工业大学》2017年博士论文　论文类型：学位论文

【摘要】：动力电池的梯级利用技术,作为一项重大的前沿应用科学研究,其内涵为通过对电池的机制诊断、寿命预测、状态估计等技术手段将不再满足电动汽车动力性能要求的退役电池,降级应用至能量、功率密度要求较低的通信基站、分布式储能等领域,进而延长动力电池的使用寿命,降低电动汽车的使用成本,同时避免废弃物排放对环境造成的污染。但目前对于动力锂离子电池特性及其老化过程的研究一般仅停留在电池容量损失率小于20%的阶段,相关研究成果并不能被简单推衍至电池的整个寿命周期,因而无法在全寿命周期内保证对电池容量损失机制的可靠诊断与对电池梯级替换时间点的有效预测。此外,作为电池退役时间点的在线判断依据,现有车用电池健康状态(SOH)估计方法也面临着温度适用范围与电动汽车实际工作环境不匹配的问题。为了解决以上问题,本文从对锂离子电池的容量损失机制诊断研究入手,分别对电池在全寿命周期内的容量损失预测问题及车用阶段的电池SOH在线估计问题进行了有针对性的研究,其主要研究内容如下:首先,针对现有用于容量损失机制诊断的开路电压(OCV)老化模型在电池老化中后期,对OCV曲线拟合逐渐失真,进而导致诊断结论不可靠的问题,提出了基于电极电势曲线非均匀压缩特性的OCV老化改进模型。该模型通过建立电池固有电极坐标系与可用电极坐标系之间的非均匀压缩转换函数,表征了电极材料单一粒子尺寸变化与多粒子尺寸分布情况变化对电极电势曲线上单相区和两相共存区占比情况的影响,从而使得OCV老化模型在整个寿命周期中保持了良好的拟合精度。通过电池在1C电流倍率、常温25℃循环老化条件下的全寿命OCV特性实验表明,相较于现有模型,本模型将整个寿命周期内的OCV拟合RMS误差由11 m V减小至2 mV。其次,针对现有容量损失预测模型仅适用于表征电池老化前期的容量变化规律,进而导致无法对电池在全寿命周期内梯级利用替换时间点进行合理规划的问题,建立了基于扩散应力分布理论的可循环锂损失-活性材料损失(LLILAM)复合容量损失模型。该模型在现有LLI损失模型的基础上,利用球形粒子的扩散应力分布理论,建立了电极粒子脱嵌锂过程中由往复变化应力所引起的材料疲劳断裂效应模型,表征了耦合老化条件(工作温度与电流倍率)与电池LAM损失速率之间的定量关系,进而获得了对电池在全寿命周期下容量损失轨迹的预测能力。通过电池在1C电流倍率、40℃条件下与0.5C电流倍率、25℃老化条件下的验证实验表明,与现有模型相比,本模型在5%的容量损失率预测误差容限内,将容量损失预测模型的适用范围由容量损失率小于20%的阶段扩展至电池的整个寿命周期。此外,针对现有电池健康状态(SOH)估计方法局限于室温工作条件,导致与电动汽车实际工作环境不匹配的问题,提出了一种适用于宽温度范围的锂离子电池健康状态在线估计方法。该方法利用固态电解质界面膜生成过程中,可循环锂消耗所引起的欧姆内阻增加量与电池容量损失量之间的函数关系,结合温度变化对电池欧姆内阻各组分阻值的作用规律模型,从原理上突破了现有电池SOH在线估计方法温度适用范围的局限。实验证明本方法在与现有方法具有同等估计精度(误差小于5%)的前提下,将SOH估计的温度适用范围由20℃~30℃拓宽至-10℃~50℃。
[Abstract]:By using the technology of power battery cascade, as a major application of cutting-edge scientific research, its connotation is through the mechanism of diagnosis, the battery life prediction, state estimation technique will no longer meet the retired battery electric vehicle power performance requirements, application to downgrade energy, low power density requirements of communication base station, distributed storage etc., and prolongs service life of the battery, reduce the cost of electric vehicles, while avoiding waste pollution emissions on the environment. But the study on the dynamic characteristics of lithium ion battery and its aging process generally only in battery capacity loss rate is less than 20% of the stage, the relevant research results cannot be simply extrapolated to the life cycle of the battery, and thus cannot be in the full life cycle to ensure reliable diagnosis of the loss mechanism of the battery capacity and battery replacement time of cascade The effective prediction point. In addition, as online time judging retired battery, battery health with existing vehicle (SOH) estimation method is also facing the temperature range and the actual work environment of electric vehicle does not match the problem. In order to solve the above problems, this paper from the research on diagnosis capacity loss mechanism of lithium ion battery with of battery capacity loss in the whole life cycle prediction problems and car issues targeted on stage SOH battery on-line estimation, the main research contents are as follows: firstly, aiming at the existing for open circuit voltage loss diagnosis mechanism (OCV) in the aging model of battery aging period, the curve of OCV the fitting gradually distortion, which led to the diagnosis conclusion is not reliable, non uniform compression characteristics of aging OCV improved model curve of electrode potential is proposed based on the model through the establishment of the battery. Non uniform compression conversion function between the electrode and the electrode can coordinate the inherent coordinate system, characterization of electrode materials of single particle size change and particle size distribution change of the electrode potential curve of single-phase region and two-phase region accounted for the effect, so that the OCV aging model in the whole life cycle to maintain good fitting accuracy the battery in the 1C. Through the current rate, temperature 25 C cycling aging life OCV characteristics under experimental conditions show that, compared with the existing models, this model will be RMS error OCV fitting the whole life cycle from 11 m V reduced to 2 mV. second, according to the existing capacity loss prediction model is only suitable for the characterization of battery aging capacity early changes, which led to the replacement time reasonable planning problem of cascade in the whole life cycle of the battery was established based on the stress diffusion The theory of circular cloth Li loss - active material loss (LLILAM) composite capacity loss model is proposed based on the existing LLI loss model, the stress distribution by using the theory of diffusion of spherical particles, fatigue fracture effect model of particle electrode of lithium intercalation process by reciprocating the stress change caused by the characterization the coupling of aging conditions (temperature and current ratio) and the quantitative relationship between LAM cell loss rate, and then obtained the ability to predict battery capacity loss in the whole life cycle under the trajectory. Through the battery at 1C current ratio, the temperature of 40 DEG C and 0.5C current ratio, 25 degrees aging show that under the experimental conditions. Compared with the existing models, this model prediction error tolerance in the capacity loss of 5%, the capacity loss prediction model suitable for the capacity loss rate of less than 20% stage extended to the entire life of the battery Cycle. In addition, the existing battery state of Health (SOH) estimation methods are confined to work at room temperature conditions, problems caused and electric vehicle does not match the actual working environment, put forward a kind of lithium ion battery health online for wide temperature range estimation method. This method uses solid electrolyte interface film formation process, can be recycled lithium consumption caused by the increase in the ohmic resistance function and the relationship between the amount of battery capacity loss, rule model combined with the temperature change on the ohmic resistance components of resistance, in principle, to break the existing battery SOH online estimation of temperature range limitations. The experiment proved that this method has the same accuracy in existing methods (error less than 5%) under the premise of the temperature range of SOH estimation by 20 DEG ~30 DEG to -10 DEG ~50 DEG. Broaden

【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TM912

【相似文献】