基于冗余度触发机制的多智能体系统可靠协调控制方法研究

In practical applications, the faulty components in some agents may result in the system performance deterioration, or may even lead to global cooperation failure of the multi-agent networks. Also, it is noted that the continuous communications between agents and their neighbors can increase waste of communication resources. To achieve the objective of cooperative fault-control and communication resource conservation, this project studies the reliable coordination control problem for multi-agent systems via topology-redundancy-triggered scheme. Both of the time-varying network disconnections and subsystem faults in agent networks are considered. Firstly, consider the case that the system states are all available and the communication networks satisfy the strong redundancy condition. Based on the distributed adaptive control technique, the topological redundancy property is used to design triggering condition and cooperative fault-tolerant control scheme. In addition, by introducing the multiple model adaptive control technique and "hard switching approaches", the design problem of reliable cooperative controller and event-trigger scheme for multi-agent systems subject to time-varying weak redundancy topology will be investigated. Furthermore, if the system states are unavailable, the dynamic output feedback reliable controller design approach via event-trigger rules will be further studied. Finally, the proposed reliable coordination control methods will be implemented on the relevant networked-robots simulation platforms to evaluate the effectiveness. This research plays an important role in the development of distributed control theory for multi-agent systems.

在多智能体系统实际工作中，子系统发生元器件故障和网络通讯故障不仅会影响子系统性能而且会导致全局协调目标的丧失。同时，智能体单元之间的连续通讯会造成网络资源浪费。本项目旨在针对不同故障模态下的多智能体系统，综合考虑通讯故障下时变网络拓扑的结构特征与子系统的故障特性，研究基于事件触发机制的可靠协调控制方法，保证系统完成容错协同目标，改善网络性能，并节约网络资源。首先，考虑时变网络强冗余的情形，拟利用拓扑冗余度信息结合分布式自适应技术，给出事件触发条件及协调容错控制设计方法。在此基础上，进一步研究时变网络弱冗余条件下的可靠协调控制问题，拟借鉴多模型自适应控制的思想，设计带有“硬切换”补偿机制的事件触发可靠协调控制方案。然后，针对状态信息部分可测的多智能体系统，研究不同故障模态下的事件触发输出协调容错控制问题。最后，将提出的控制方法应用到多遥感机器人网络仿真系统，由此提高本项目研究的实际意义。

本项目将复杂多智能体系统作为研究对象，针对复杂通讯网络限制下（如通讯故障、间歇传输、切换网络、传输时滞等）深入研究了分布式自适应驱动协同控制方法。首先，研究了一类非线性多智能体在通讯受限情况下的自适应一致性控制问题，考虑在切换拓扑条件下，通信存在间歇性通讯受限时提出了自适应神经网络协同控制算法，解决了非线性多智能体系统协同控制问题；其次，针对含有信息通讯时滞的耦合调控网络系统，基于非奇异M矩阵和非线性模型转换方法，给出了分布式耦合系统状态观测器设计方法和全局同步的分析条件；同时，针对通讯受限网络中的时滞现象，利用Lypunov矩阵方法并结合偏微分方程理论，给出了不确定系统指数稳定的充分必要条件，并设计了镇定控制器；在此基础上，研究了在非对称网络通讯环境中出现通讯中断（攻击）故障多智能体系统的自适应协调跟踪控制问题。利用子系统之间的信息交换设计分布式参考模型观测器，从而克服非对称拓扑结构矩阵与跟随智能体系统控制增益设计无法直接解耦的难点。