通讯受限下CPS异构网络的限时协同与结构动态优化方法研究

Based on the characteristics of network heterogeneity, node autonomy, and distributed structure for CPS, this project will deeply investigate the problems of dynamic switching models, finite-time cooperation control algorithms and structure optimization for CPS heterogeneous networks with limited communication, and establish the corresponding principles and methods. By the communication characteristics of typical CPS heterogeneous networks, one will first analysis the main communication limiting factors of CPS heterogeneous networks and their impacts on network cooperation behaviors. Then, according to the evolution rule of information transmission dynamics for actual multi-agent complex dynamical networks, one will further construct a general dynamic switching model for CPS heterogeneous networks oriented with optimizing network communication performance. From the view of actual CPS heterogeneous networks under limited communication, one will finally give the finite-time cooperation control algorithms and the dynamic optimization schemes of network structure, which can guarantee the optimal performance of lowest energy consumption, balanced communication load as well as real-time communication, and apply it to the actual multi-robot coordination system. In theory, the investigation results will help us deeply understand and reveal the intrinsic links and laws between these communication limiting factors and the finite-time cooperation for CPS heterogeneous networks, and provide new ideas and methods for solving the crucial technique of CPS networks. In practice, the investigate results will also guide us to design and implement some typical CPS networks with limited communication such as multi-robot intelligent systems and intelligent traffic systems and so on.

基于信息物理融合系统(CPS)的网络异构性、节点自治性、结构分布性等特点，深入研究通讯受限下CPS异构网络的动态切换模型、限时协同控制算法及结构优化问题，建立相应的理论和方法. 根据典型CPS异构网络的通讯特点，分析CPS异构网络在信息通讯中的主要受限因素及其对网络协同行为的影响. 根据实际多智能体复杂动态网络中信息传播的动力学演化规律，以优化网络通讯性能为导向建立具有一定普适意义的CPS异构网络的动态切换模型. 结合实际的CPS异构网络，给出在通讯受限下使网络达到能耗最低且保证负载均衡和实时通讯的限时协同控制算法与结构动态优化方案，并应用到实际多机器人协同系统中. 在理论上深刻理解和揭示这些通讯受限因素与CPS异构网络限时协同之间的内在联系与规律，为解决CPS网络的关键技术提供新的思路和方法，在实践上指导多机器人智能系统和智能交通系统等通讯受限下典型CPS异构网络的设计与实现.

能源的日趋匮乏与需求的急剧上升促使微电网的建模与CPS协同控制成为未来复杂系统智能控制的一个新的聚焦点. 本项目深入分析了CPS异构网络中参考信息不完全可知、通讯链路不确定以及存在非一致时变通讯时延等典型的通讯受限因素，构建了微网簇导向的CPS双层通讯网络切换拓扑结构，并研究了受控系统的分布式限时协同及结构动态优化问题. 研究结果表明: (1)对于参考信息不完全可知的CPS系统，分布式限时追踪策略可有效实现系统的协同；结合迭代学习的间歇通讯机制可有效抑制不确定通讯链路对受控CPS系统产生的影响；相较于通讯时延的上界大小，其变化率对CPS系统的稳定性能的影响更大. (2)基于簇导向的双层通讯网络切换拓扑结构，当其上层与底层通讯网络的采样时间之比小于等于某一上界时可保证受控CPS系统的分层分布式协同控制性能；牵制控制和一致性控制协议分别对于底层和上层通讯网络的稀疏性结构优化和鲁棒性能提升起到了至关重要的作用；拓扑切换滞留时间在一定程度上决定了稀疏网络拓扑结构的优化准则，进而影响了变结构网络带宽阈值的选取. (3)理论结果应用于通讯受限环境下的低压微电网CPS系统，不仅在一定程度上有效缓解了电压调节和无功共享之间的固有矛盾，而且促使微电网簇内和簇间各分布式电源均实现了电压/频率调节以及有功/无功功率共享；相较于传统的微电网控制理论和框架，受控系统对上述各通讯受限因素的鲁棒性能大幅度提升. 本项目首次针对微网系统提出了簇导向的双层通讯网络拓扑架构及相应的多尺度限时协同优化控制方法，研究结果不仅具有深刻的理论意义，而且为未来微电网簇CPS异构网络架构的设计与实现提供了重要的科学支撑.