ACPS（汽车信息物理系统，Automotive Cyber Physical Systems）数据能反映汽车的重要特征，有效利用数据设计系统以提高汽车的整体性能是目前的研究热点之一。鉴于ACPS数据在结构和采集方式上均存在较大差异，如何筛选数据并让其发挥应有的作用是一个非常重要的问题。本课题着眼于数据驱动的建模与验证问题：研究预处理技术，通过计算获得精确的嵌入维度与时间延迟构建系统的状态空间；提出建模方法的同时引入反馈机制，通过迭代不断缩小离散数据与连续模型之间的误差，建立系统精确模型；将系统数据模块化，研究关联维度与汽车可靠性之间的关系。课题研究将解决关于ACPS的若干理论和关键技术问题，为未来基于数据驱动的建模及验证技术提供参考与借鉴。

The data of ACPS (Automotive Cyber Physical Systems) reflect the important characteristics of automobiles, and it is one of hot spots in current research to improve the overall performance of automobile by effective use of data design system. On the basis of the great differences in the structure and the collection method of ACPS data, it is a very important issue on how to filter the data and play its due role. This topic focuses on modeling and verification based on data. First, we research the preprocessing technology and construct the state space of system through calculating the embedding dimension and time delay. Secondly, a modeling method with feedback mechanism is introduced to reduce the difference between the discrete data and continuous model by iteration. Finally, the topic researches the relationship between correlation dimension and vehicle reliability on the modularity data. This project will solve several theoretical and key technical problems about ACPS and provide a reference for future data-driven modeling and verification technology.