资源受限网络控制系统的必要性通讯与分布式控制协同研究

Control over a communication network has attracted more and more interests in international academe.The general adopted time-triggered single-packet communication scheme in a networked control system is not suitable to a large-scale system (LSS), since LSS is spatially distributed and with multi-sensors and multi-actuators. For a large-scale system connected via a limited bandwidth network, the following three points will be addressed in this project: 1) necessary communication scheme with the characteristic of discrete sampling and event-triggered communication will being studied for saving the limited network resources and guaranteeing the desired control performance; 2) based on the spatially distributed characteristic of large-scale systems, a communication threshold-error dependent distributed control modeling will being constructed based on the information of subsystem itself and its' directly connected subsystems; and 3) since the performance of an NCS depends not only on the design of the control algorithms but also on the scheduling of the share network resources, separately considering the control and communication in the open literature will inevitably bring some conservativeness. To tackle this problem, the co-design between the control and communication will being carefully studied. Based on the proposed theory and technique, we will implement the co-design experiment between control and network on a practical test bed- - -two pendulums coupled by a spring to verify the proposed theorem and method; exploit the corresponding remote control and communication modules for networked large-scale systems; and construct a practical demonstrated plat for networked large scale systems. At last, we will propose a novel and effective necessary communication scheme with the characteristic of discrete sampling and event-triggered communication for saving the limited communication resources, a distributed control modeling for a large-scale system under network environments, and a co-design method between the communication and control for preserving the desired performance while occupying fewer communication resources.

通过网络实现闭环控制越来越受到国际学术界的重视。网络控制系统中通常采用的单包传输和基于等周期的时间驱动传输策略对于网络资源受限的具有多传感器和执行器的大规模系统不再适用。针对资源受限网络环境下的大规模系统，本项目从三个方面进行深入研究：1）为达到期望控制性能而占用较少网络资源的目标，研究基于"离散采样，事情驱动传输"的必要性通讯策略；2）基于大规模系统空间分布性特征，研究利用子系统本身和直接相连子系统信息并嵌入必要性通讯策略的通讯阈值误差依赖的分布式控制模型；3）基于已有通讯策略与控制算法分别设计的问题，研究通讯策略与分布式控制协同设计方法。基于提出的理论方法与技术路线，针对并联倒立摆等复杂大规模对象进行网络与控制协作实验，开发出适应大规模系统网络控制的远程控制模块和相应的控制与通讯软件，建立网络控制应用示范平台；在网络控制系统通讯策略、分布式控制、通讯与控制协同设计等方面提供有效方案。

本项目对具有网络资源限制的网络化系统必要性通信及分布式控制协同策略进行了较为深入的研究。针对“离散采样，采样点事件驱动”的信号传输策略，设计了QoS约束下的异步多包必要性通讯逻辑；基于异步多包必要性通讯机制，建立了融合必要性通讯策略的大规模系统分布式控制模型；同时考虑必要性通讯参数和分布式控制参数，实现了网络化大规模系统分布式控制与信号传送策略的联合优化设计，揭示了网络通讯策略与控制参数对NCS综合性能影响。在本项目资助下，发表SCI期刊论文17篇，会议论文12篇；授权发明专利2项，登记软件著作权1项；培养研究生6名，其中博士（联合培养）2名，硕士4名，项目完成预期目标，按期结题。