基于信息获取与融合的电池系统健康状态估计方法研究

State of health (SOH) of the battery system can directly affect the safe and stable operation of the electric vehicle. Thus, this project aims to solve the problem of accurate and robust estimation of the SOH for the battery system through the theory and technology of information science and control science. Based on the information acquisition, processing and utilization mechanism with information fusion theory system, this project is focusing on battery system information active acquisition method, information combination processing method, battery system modeling and SOH estimation algorithm. The main works of this project can be planed as follows. Firstly, based on the research of the acquisition approaches of information with different sources, layers and time-scales, an active information acquisition method system will be built and a sufficient, effective and reliable information acquisition mechanism for the real battery system will be realized. Secondly, by taking into account of the large amount types of sensors, large amounts of measured data and multifarious information time scales of the battery system, the combined processing of multi-source, multi-level, multi-scale battery information is achieved based on the information fusion technology. Thirdly, the mathematical model describing the degradation of battery performance is established. On the basis of the degradation model, the battery SOH will be estimated by using the close-loop algorithms and decision-level information fusion methods. With the help of the above-mentioned works, the overall research results can improve the information acquisition efficiency of the battery system, and promote the accuracy and robustness of the battery SOH estimation results. Moreover, these results are expected to provide new ideas for the acquisition, processing and control of the multi-sources, multi-layers and multi-time scale information system.

电池系统的健康状态将直接影响电动汽车的安全、稳定运行。本项目旨在利用信息科学、控制科学的理论和技术，解决电池系统健康状态的准确、鲁棒估计问题。项目拟以电池系统信息获取、处理与利用机制为切入点，基于信息融合理论体系，围绕电池信息主动式获取与组合处理方法、电池系统模型构建与健康状态估计算法开展研究，主要研究内容有：拟研究不同来源、层次、尺度电池信息的获取手段，构建主动式信息获取方法体系，实现充分、高效、可靠的电池信息获取；针对电池系统中传感器种类多、测量数据量大、信息尺度杂的特点，结合信息融合技术，实现对多源、多层次、多尺度电池信息的组合处理；建立描述电池性能衰退的数学模型，应用闭环估计算法和决策级信息融合方法，实现对电池系统健康状态的估计。上述研究的开展将有助于提高电池系统信息获取效率，以及电池健康状态估计的准确性和鲁棒性，并有望为多源、多层次、多尺度信息获取、处理和利用的研究提供新思路。

电池系统健康状态是影响电动汽车的安全、可靠运行的重要因素。本项目重点针对动力电池系统行为描述和状态估计问题，利用信息、计算机和控制科学的理论和技术，通过机理分析、数据挖掘和信息融合等手段，开展电池系统数据处理、特征提取和状态估计方法的研究。具体地，通过主动控制电池系统在充电过程的运行模式，利用电池管理系统获取此过程的电压、温度、电流等多源、多层次和多尺度的数据，在数据特性分析的基础上针对性的设计信息处理和特征生成方法；根据数据驱动的电池行为仿真模型的需求，在前述生成特征的基础上，制定特征重要性和影响力的评价体系，利用相关性分析、重要性分析、最优化算法等手段从不同角度对已生成的特征进行评价并筛选，提取在具体场景下最适合刻画电池老化过程的关键特征；分析电池老化过程中各项重要物理参数的变化规律和演变机制，结合筛选的特征，利用数据驱动模型及其融合模型对电池老化过程进行建模并量化描述，运用非线性回归算法、深度神经网络等手段实现对电池健康状态的准确估计，最后通过非线性滤波算法对估计结果进行二次修正和优化，实现对电池健康状态估计效果的再次提升。在本项目的支持下，项目负责人在IEEE Transactions on Power Electronics、IEEE Transactions on Energy Convention、Energy、Journal of Power Sources等期刊发表高水平论文15篇，实审发明专利12项。在本项目研究内容和成果的基础上，项目负责人凭借在动力电池健康状态估计方面的研究入选了2020年度安徽省科协青年科技人才托举计划。本项目相关研究的开展有助于提高电池系统状态估计的准确性和鲁棒性，对加快电动汽车领域的信息化和智能化进程有着重要的作用，并为多源、多层次、多尺度信息的处理和利用提供了新思路和新方法。