车用动力电池不一致性多尺度特征聚类模型及均衡预测控制研究

Balancing is an important way to improve the performance and lifetime of the battery pack. Power batteries have the characteristics of nonlinearity, strong time-variation and close coupling, and their cells show inconsistency of parameters. Due to the complex working conditions of electric vehicle, the inconsistency extending of cells can be aggravated, and even leads to significant differences. Aiming at solving the problems of the evolution of battery inconsistency enlarging to the unbalanced state and their balancing control, this project concentrates on the following contents: 1) The power batteries' external characteristics, SOC, and other parameters will be explored affected by the factors such as aging degree, charge and discharge rate, and so on, and the unbalanced path model of the power batteries parameters will be constructed, and the coupling parameters as well as the balancing switching mechanism will also be explored. 2) A joint estimation methodology of states and parameters of power battery will be developed, and a clustering by fast search and find of density peaks (CFSFDP)-based approach will be proposed to analyze the multi-scale features of inconsistency in the power batteries, and in addition, an adaptive local density clustering core as well as distance from data points belonging to other clusters will be explored to perfect the model of the multi-scale features of inconsistency. 3) To optimize the balancing trigger function, a primary feature identification algorithm for the inconsistency of the power batteries will be developed, and an isolated bidirectional Cuk converter-type centralized active balancing control system will be designed, and its time/energy consumption minimization-based balancing predictive control will be implemented. A model of multi-scale features of inconsistency and an isolated bidirectional Cuk converter-type centralized active balancing control will be expected, and the research achievements and contributions will provide scientific value and theoretical significance for improving battery management and balancing efficacy.

均衡是提升电池组性能与寿命的重要途径。动力电池具有非线性、强时变、强耦合等特征，单体电池间各种参数存在差异，而电动汽车工况复杂，导致差异加剧，甚至出现显著性不同，针对不一致性扩大至不均衡状态的演化与一致性均衡等问题：1) 探索老化程度、充放电倍率等因素对动力电池外特性、SOC等参数的影响规律，建立各参数的失衡路径模型，探究参数耦合与均衡启闭机制；2) 研究动力电池模型参数与状态联合估计，基于密度峰值和距离聚类分析不一致性多尺度特征，并通过自适应局部密度聚类中心与截断距离完善不一致性多尺度特征聚类模型；3) 构建动力电池不一致性主特征判据，优化均衡启闭触发函数，设计一种隔离型双向Cuk变换器集中式主动均衡控制系统，研究计及时间/能量耗散最小的均衡预测控制。预期提出动力电池不一致性多尺度特征聚类模型与一种隔离型双向Cuk变换器集中式均衡系统，对完善电池管理与提高均衡效能具有科学价值和理论意义。

发展新能源汽车是我国国家战略，动力电池及其管理系统是新能源汽车的关键技术。动力电池具有非线性、强时变、强耦合等特征，单体电池间各种参数存在差异，而新能源汽车工况复杂，导致差异加剧，动力电池使用或管理不当会进一步影响整车续驶里程与使用期限。因此，研究动力电池不一致性成因及均衡管理技术提高电池组一致性，延长其使用寿命，对提升动力电池乃至整车性能有重要意义。本项目针对电池组不一致性扩大至不均衡状态的演化问题，开展了性能测试及寿命衰退实验，分析了不一致性扩大成因、失衡路径以及演变规律，对动力电池组进行多尺度特征密度峰值聚类，建立不一致性评判依据，提出了一种基于信息熵-灰色关联分析的电池组不一致性评价方法。其次，本项目基于动力电池RC等效电路模型建立了改进的平方根二阶中心差分变换卡尔曼滤波SOC估计方法，在HPPC和UDDS工况下估计误差均在2%内。为进一步提高变工况条件下的状态估计精度，本项目研究了基于模型-数据双驱的电池状态SOX联合估计方法，建立了GRU深度-迁移学习SOX联合估计模型，实现了在电池老化情况下的状态估计误差在5%内。最后，为实现动力电池组一致性均衡及均衡效率优化，本项目提出了基于隔离变换器的模块化均衡架构和分层控制系统，设计了考虑电池组间SOC差异和电池组内单体间SOC差异的能量分配策略，提出了考虑均衡电流自适应追踪和阈值时变约束的分层MPC均衡管理策略，建立了电池组均衡系统预测模型，构建了电池组均衡时间或能量消耗最小的目标函数，求解均衡优化问题获取最佳均衡电流值。采用均衡器电流自适应滑模追踪控制，解决能量转移路径变化和电池均衡器工作模式频繁切换导致的外部负载干扰突变和参数摄动等问题。并通过最佳均衡电流自适应追踪控制和电池组均衡管理等实验，验证了提出的基于隔离变换器的均衡系统样机的可行性。综上，本项目研究的动力电池组均衡方法有望应用于电动汽车电池组的一致性管理与优化。

内容获取失败，请点击重试