基于贝叶斯理论的动力电池系统参数与状态联合估计方法研究

The electric vehicle as a kind of zero-emission transportations is a good solution for energy and environment issues. The power battery system is the core component of the electric vehicle. The safe and efficient management issues of power battery system have always been a hot topic in academia. The battery system is different from the general industrial control objects. It belongs to a class of complex electrochemical system. On the one hand, the battery states are easily affected by factors like environment and aging issues. On the other hand, based on the previous research of the proposer, there exists a coupling relationship between battery model parameters and the system states, which cause it mutually restricted between the accuracy and reliability of the state estimation and the accuracy of the battery model parameters. To solve the above problems, this project is facing vehicle power battery system. Based on the study of the internal and external characteristics of vehicle power battery system, a full cycle life model of power battery system considering environment and aging characteristics will be established. Based on this model and Bayesian theory, a non-Gaussian noise-driven state estimation theory will be developed, to achieve the joint estimation of battery model parameters and system states with high accuracy and robustness. Furthermore, research on the prediction of power state and health state at multi-timescales will be performed, to provide a theoretical basis for the safe, healthy and efficient operation of the power battery system.

电动汽车作为一种零排放的交通工具是解决环境与能源问题的一个优选方案。动力电池系统是电动汽车的关键部件，其安全高效管理问题一直是学术界关心的热点问题。电池系统不同于一般的工业控制对象，其本身属于一类复杂的电化学系统，一方面，电池的状态易受环境、老化等因素影响。另一方面，申请人前期研究发现电池的模型参数与系统状态之间存在耦合关系，状态估计的精确度和可靠度与电池模型参数的准确性相互制约。针对上述问题，本项目面向车载动力电池系统，通过对车载动力电池系统的内、外特性研究，建立考虑环境与老化特征的动力电池系统全生命周期模型。在该模型的基础上，基于贝叶斯理论，开展由非高斯噪声驱动的非线性动态系统状态估计理论研究，实现高精度、高鲁棒性的电池模型参数和系统状态联合估计。进一步开展多时间尺度功率状态与健康状态预测问题的研究，为动力电池系统的安全、健康、高效运行提供理论基础。

加快发展节能与新能源汽车技术对中国实现汽车产业跨越式发展，从汽车大国迈向汽车强国具有重大战略意义。动力电池系统的管理和控制问题关系着电动汽车产业化和市场化的成败。该项目针对动力电池精细化建模和状态估计问题，从基础理论和关键技术应用两方面深入开展了系统性的研究，主要成果如下：首先，提出了以动力锂电池系统为代表的复杂非线性时变系统的建模、参数与状态联合估计理论方法，构建了基于贝叶斯理论的多尺度SOX闭环状态观测器，针对新能源汽车动力系统建立了基于分布式网络架构的智能化监测与管控系统，实现了对锂电池系统的智能监测和管控。其次，提出了基于动态贝叶斯推理及特征匹配的故障诊断、溯源和隔离方法，建立了基于高斯过程回归模型的电池全生命周期健康预测理论体系，实现了锂电池系统全寿命周期的故障诊断和健康管理，为锂电池系统的安全可靠运行提供支撑和保障。最后，提出了自终端至云端的智能网络协同制造平台架构，实现了基于信息物理系融合的电池系统生产/制造/服务全域和供应/营销/服务全链一体化协同，推动了人工智能、大数据等技术在新能源汽车领域的深度应用。.项目实施过程中，在国内外重要期刊上发表论文30篇，其中SCI论文26篇，申请发明专利6项，项目团队成员多次受邀在国际重要学术会议上报告研究成果，培养硕、博士研究生5名，相关成果获教育部技术发明二等奖，第34届世界电动汽车大会Excellent Paper Award。