多因素耦合的虚拟电池用锂离子电池组分数阶时频域建模研究

Due to the existing problems of poor repeatability, time-consuming, requiring many peripheral equipment and low security during experimental tests for real batteries, the replacement of real batteries with virtual batteries is of great significance in accelerating the development of battery management system strategy, charging machine test development and motor control strategy formulation. The extreme simplicity of the virtual battery model and its poor adaptability under many application conditions have become the bottleneck restricting its replacement of the real battery for the research and development of battery application technology.How to capture the dynamic characteristics of battery under multi-factor coupling conditions in time-frequency domain plays a critical role in establishing a systematic model with high precision and good applicability for the virtual battery. A fractional-order time-frequency domain model for lithium-ion battery is proposed by researching the physical and chemical mechanism inside the battery. Based the developed model, the structural composition of 1s pulse impedance in time domain is revealed, and the equivalent relationship between the time-domain impedance and frequency-domain impedance is found. We propose a fast and efficient parameter identification method for the developed model via combining time-domain and frequency-domain data. The evolution rule of model parameters with temperature, C-rate and SOC is put forward. What’s more, the compensation mechanism and boundary conditions that make the multi-condition compounding response satisfy the superposition principle are uncovered, and the comprehensive mechanism of parameters is constructed. Ultimately, a new fractional-order time-frequency domain systematization model of lithium-ion battery pack used for the virtual battery system is proposed, which is suitable for multi-factor coupling conditions, and a complete scheme for the testing, modeling and parameter identification of lithium-ion battery used for virtual battery systems is developed.

由于实际电池存在测试重复性差、耗时长、需要外围设备多、安全性低等问题，虚拟电池替代实际电池对加快电池管理系统策略研发、充电机测试开发和电机控制策略制定等具有重要意义。虚拟电池用模型过于简单、适用工况过于单一已成为制约其替代实际电池进行应用研发的瓶颈。如何模拟电池在时频域多因素耦合工况下的动态特性是建立精度高、适用性好的虚拟电池用系统化模型的核心问题。项目通过开展电池内部物理化学机理的研究，建立锂离子电池分数阶时频域模型，揭示时域脉冲1s阻抗的结构化组成关系，给出时域阻抗与频域阻抗的等效关系，提出时域与频域相结合的快速、高效的参数辨识方法；提出模型参数随温度、倍率和SOC的衍变规律，揭示多工况复合响应符合叠加原理的补偿机制和边界条件，构建参数综合机制；提出适用于多因素耦合工况的虚拟电池用锂离子电池组分数阶时频域系统化模型，形成一套用于虚拟电池系统的锂离子电池测试、建模和参数辨识方法系统方案。

采用高精度虚拟电池测试电池管理系统存在可重复性好、切换快、测试成本低等优势，对加快策略开发及测评具有重要意义。但模型适用工况单一、动态特性差已成为制约虚拟电池技术发展的瓶颈。针对时频域多因素耦合工况下电池动态特性影响因素多、非线性强、难以高精度模拟的难题，项目从单体到系统，从理论到产品，层层深入开展研究：.首先，针对分数阶模型倍率工况适用性单一问题，通过不同幅值电化学阻抗谱（EIS）测试，构建了反映电池内部特性的分数阶模型，推演了模型参数与激励幅值的相关性规律，建立了适用于多倍率工况的分数阶频域模型。第二，针对EIS低频半波周期长、开路电压（OCV）影响大的问题，提出了基于OCV校正的修正EIS数据，在低频时域区将模型误差降为常规的50%以内。第三，兼顾整数阶模型计算量低与分数阶模型精度高的优点，建立模拟多时间尺度电池特性的变阶数模型，实现了短长时间尺度特性的高精度模拟。第四，基于连续短时脉冲激励下的暂态极化电压特征，提出了改进的整数阶电池模型参数辨识方法，突破了虚拟电池单体模型精度和计算量无法兼顾的难题，提高了模型适用性。第五，针对高频纹波应力下电池电压特性难以准确模拟的难题，考虑时域工况的稳态特性和频域工况的暂态特性，建立了兼顾模型精度和仿真时间的电池高频特性模型。第六，采用基于多元核密度估计-蒙特卡洛抽样和变分自编码器的方法，考虑参数的分布特性及相关性，提出了电池不一致性建模方法，实现了虚拟电池组与真实数据同分布的参数重构，生成任意串数的电池组参数，同时采用多种方法对比研究。第七，依托硬件通过模型下载建立了适用于多工况的高精度虚拟电池系统，通过外围电路的拓展，建立了电池管理系统测评用硬件在环（HiL）仿真测试系统。.项目攻克了锂离子电池多因素耦合工况时频域仿真难题，建立了高精度电池组模型，技术成果已在民用领域获得批量应用，在国防领域的技术支撑也已开展，应用潜力巨大。