基于虚拟原型的信息物理融合系统高效可信构造研究

As a key technology of the fourth industrial revolution, Cyber-Physical Systems (CPSs) are supposed to be able to accurately and efficiently conduct the sensing, actuation, communication and control within physical world. Since CPS designs involve both uncertain physical environments and parallel computing processes which interact with each other intensely, the design complexity of CPSs is extremely high. Due to the lack of expertise in both domains and the inherent difference between computing and physical processes, the process of CPS construction is very time-consuming and error-prone. ..Virtual prototype is a promising abstraction mechanism that plays an important role in reasoning the trustworthiness of CPS designs. It can be used to eclipse CPS complexity in early design phase. It enables the design space exploration as well as the co-validation of overall system functional behaviors. Therefore, the objective of this project is to construct trustworthy CPSs in an efficient way based on virtual prototypes. This project will mainly investigate how to optimize CPS designs and automate the CPS virtual prototype generation? This project will try to figure out how to promote the construction efficiency of CPSs which involve various resource constraints and uncertain physical environments. This project will investigate how to utilize virtual prototypes to reduce the overall validation efforts...In conclusion, as the “golden reference models” in CPS design flow, high-quality virtual prototypes can essentially guarantee the success of the trustworthy CPS construction. By exploring various efficient and effective virtual prototype oriented techniques, we believe that the outcomes of this project can not only significantly reduce the CPS construction time, but also can enhance the performance, reliability and predictability of constructed CPSs.

作为工业4.0的核心技术，信息物理融合系统（CPS）能够对物理世界精确地实施“感、执、传、控”。但由于需要同时考虑不确定物理环境与并发计算部件，CPS构造过程复杂度极高，最终产品可信难于保证。虚拟原型作为一种系统仿真手段，支持在设计早期进行精确与高效的设计空间探索与协同验证。本项目围绕CPS可信构造，研究如何基于虚拟原型来有效地优化CPS设计并降低CPS构造的总体时间。主要研究不确定环境与资源受限情况下虚拟原型的设计优化方法与自动化生成技术，提高CPS设计质量与开发效率；本项目的研究将为CPS的构造提供理论基础与技术支撑，在提高CPS可信的同时，大幅降低其设计与建模的时间。

作为工业4.0的核心技术，信息物理融合系统（CPS）能够对物理世界精确地实施“感、执、传、控”。但由于需要同时考虑不确定物理环境与并发计算部件，CPS构造过程复杂度极高，最终产品可信难于保证。虚拟原型作为一种系统仿真手段，支持在设计早期进行精确与高效的设计空间探索与协同验证。本项目围绕CPS可信构造，研究如何基于虚拟原型来有效地优化CPS设计并降低CPS构造的总体时间,主要研究不确定环境与资源受限情况下虚拟原型的设计优化方法与自动化生成技术，提高CPS设计质量与开发效率。本项目主要研究内容与贡献如下:.1)不确定环境下面向CPS定量评估方法。系统地提出了一整套不确定环境描述的AADL Annex，支持不确定连续变量与离散变量的描述，通过映射规则将不确定AADL转化为价格时间自动机，支持CPS体系结构层次性能的定量评估。基于机器学习，提出了一种CPS系统调度优化方法，能够有效约减搜索空间，提高不确定环境下CPS设计的性能。.2）资源受限情况下的并行高效综合方法。基于目前已有的分支界定方法，系统地提出了基于多核的并行空间约减方法来快速获取最优设计；创新地提出了两阶段搜索方法，支持最紧初始搜索空间的获取，支持多搜索任务间的协同并行搜索快速获得最优解。实验结果显示，我们提出的方法比现有最好的分支界定方法快10-100倍。.3）虚拟原型的高效测试用例生成。创新的提出了多种面向虚拟原型数据流的测试覆盖准则，并提出了自动化计算虚拟原型测试数据流覆盖的方法，支持高效的虚拟原型测试生成。.4）CPS基础设施的可信设计。针对目前云计算、物联网与异构嵌入式芯片的设计，充分考虑到外界环境的不确定因素，设计相关优化算法，提高其在不确定环境下的可靠性。.本项目的研究保证了在不确定环境下CPS系统级建模与分析的性能，节省系统级设计到底层实现间的综合时间，基于统计模型检验与机器学习方法提高了系统的整体评估与优化的效率，为CPS的构造提供理论基础与技术支撑。在提高CPS可信的同时，大幅降低其总体构造的时间。