面向适应性和差异性的锂离子动力电池组建模和状态估计方法研究

To realize the request that the high-accurate states estimation of the battery management system (BMS) to lithium-ion battery pack, even under the entire life-cycle and dynamic operating temperatures, concerning problems that, the singleness and poor adaptability of traditional modelling, and the difficulty of tracking the inconsistency of cells in the pack, the project is to investigate: 1) Analyze the application boundary of equivalent circuit model, fractional order impedance model and electrochemical model and a novel modelling method facing strong adaptability will be presented in order to meet the strict requirement to modelling accuracy under complex cycles. 2) Dissect the mechanism between modelling accuracy and that of SOC estimation and according to the theory of uncertainty optimization, a multi-models probability fusion estimation approach will be proposed to elevate the reliability and stability of SOC estimation. 3) Explore the inner relation of capacity degradation and characteristic variation, construct the response surface of open circuit voltage, reinforce the sensitivity of equivalent circuit model to capacity degradation and bring up a new method with respect to high accuracy estimation for the available capacity of batteries. 4) Analyze the multi-dimensions evolving regularity of performance degradation of battery pack in series, build the differentia model concerning the propagation mechanism of uncertainties, and based on the theory of uncertainty optimization, the quantization of uncertainties will be accomplished step by step. The research results will promote the accuracy and reliability of states estimation for the battery pack, provide the theoretical foundation of delicacy management of BMS, which owns significant and realistic meanings.

为实现在全寿命周期动态工作温度下，电池管理系统对动力电池组高精度状态估计的目标，针对当前动力电池建模方法单一，模型适应性不高，成组后单体间不一致性难以追踪的问题，研究包括：1）分析等效电路模型、分数阶阻抗模型以及电化学模型的适用边界，提出动力电池强适应性建模方法，满足复杂工况对模型准确度的高要求；2）剖析模型准确度对荷电状态估计精度的作用机理，依据信息融合理论，提出多模型概率融合估计方法，提升荷电状态估计可靠性和稳定性；3）探究电池容量衰退与其特性改变的内在关联，建立开路电压响应面模型，强化等效电路模型对容量衰退的敏感度，提出高精度可用容量估计新方法；4）解析串联动力电池组性能衰退多维演化规律，构建计及不一致性传播机制的单体间差异模型，基于不确定性优化理论，逐步实现不一致性的准确量化。研究成果将提升对动力电池组状态估计精度和可靠性，为电池管理系统精细化管理提供理论基础，具有重要的现实意义。

车载动力电池组具有强非线性特征，在工程应用中存在模型建不精，状态估不准的问题，制约着电动汽车的应用与发展。本项目重点研究了动力电池组的建模和状态估计。.针对动力电池单体的建模的问题，基于传统等效电路模型和电化学模型提出降阶电化学模型计算方法和具有温度和老化意识融合的电池模型，解决了普通模型适应性不强的问题。.针对动力电池SOC-SOH估计中非线性强，噪声干扰大的问题，本项目提出了具有强非线性跟踪能力的UKF算法和强鲁棒性PID观测器融合估计算法，能够较好的保障算法的跟踪精度，抑制噪声干扰。.针对动力电池组建模和状态估计问题，提出一种基于曲线特征量提取的电池筛选方法，有效减小了电池成组的一次不一致性；针对电池组使用过程中由于系统不一致性导致常用的电池组SOC估计失效的问题，建立了实时量化不一致性的电池组SOC估计算法，实现了电池组荷电状态的准确估计。.此外，搭建了基于xPC Target的动力电池在环仿真实验平台和BMS算法验证平台，完成了电化学模型、状态估计算法的硬件验证。.在圆满完成既定目标的前提下，进一步开展了动力电池快速充电策略研究，并提出一种考虑温度和老化的充电方法。在该项目支持下，共发表论文5篇，其中SCI论文3篇，EI论文2篇，申请发明专利3件，培养硕士研究生4名，博士研究生4名。