基于事件触发机制的异质多智能体系统分布式协同控制

Distributed cooperative control is a fundamental problem in the applications of multi-agent systems consisting of large-scale and low-cost agents. As one important tool, event-triggered strategy can be used to reduce the communication load and the energy consumption. The agents in the considered multi-agent systems can have nonidentical dynamics and even different state dimensions. Furthermore, the agents dynamics are not limited to linear systems. Such heterogeneity and nonlinearity of the systems are coupled with the nonlinearity introduced by the event-triggering mechanism, which becomes a main challenge. The project aims to study the output consensus problem, the cooperative signal tracking problem at the presence of disturbances for a class of heterogeneous nonlinear multi-agent systems. For this purpose, first, problem formulations of such cooperative control problems with time-invariant and symmetric network are established based on event-triggering communication. Second, by constructing novel distributed dynamic compensators synthesized with event-triggering mechanism, the original cooperative control problems are converted to the stabilization problem of a class of closed-loop augmented systems. To solve the stabilization problem, feasible event-triggered control schemes are proposed by combining both event-triggered strategy and periodic sampling strategy together. In this way, solvability conditions of the original cooperative control problems are obtained. Third, the following two cases are considered: the communication network is time-varying and asymmetric; the system dynamics contain some uncertainty. In both two cases, event-triggered control schemes are developed. The project will provide the theoretical support to the applications of resource-limited multi-agent systems.

分布式协同控制是解决大规模、低成本系统在应用中的关键问题。事件触发策略是有效减少网络通信负担和能源消耗的重要手段。本项目中的智能个体允许有不同的动态方程乃至状态维数，且动态方程不再局限于线性或线性化处理。系统本身的异质性和非线性与事件触发引起的非线性耦合，是本项目的主要难点。针对一类异质非线性多智能体系统的输出一致性问题和干扰下的协同信号跟踪问题，首先，本项目将建立在时不变、对称网络下基于事件触发通信的协同控制问题模型；其次，通过构造集成了事件触发机制的分布式动态补偿器，将原问题转化成一类闭环的增广系统的镇定性问题，通过融合事件触发采样和周期采样两种方式，设计可执行的基于事件触发机制的控制算法解决该镇定性问题，从而获得原问题的可解性条件；最后，研究时变、非对称通信网络和系统方程含不确定参数的影响，提出基于事件触发的控制算法。本项目为多智能体系统在资源受限环境下的应用提供理论支持。

分布式协同控制是解决大规模、低成本系统在应用中的关键问题。事件触发策略是有效减少网络通信负担和能源消耗的重要手段。本项目针对非对称通信网络下的多智能体系统分布式事件触发控制展开研究。在本项目的支持下，主要进行了如下几方面的研究：一、针对非对称通信网络下多智能体系统基于事件触发通信下的一致性问题，设计了一类新型的基于动态事件触发机制的协同控制算法，进一步放大采样间隔的同时排除了芝诺行为；二、针对非对称通信网络下、系统方程含不确定参数的多智能体系统一致性控制问题，提出了基于事件触发通信的输出反馈分布式控制算法，通过集成事件触发机制的分布式观测器，将原问题转化为误差增广系统的镇定性问题，并得到问题可解性条件和控制器设计方法；三、针对一类异质非线性多航天器系统带干扰下的姿态协同跟踪问题，提出了一种具有性能可调节函数的事件触发通信机制和控制方法；四、针对二阶多智能体系统的一致性问题，提出了一种新型重置控制算法。相比于传统静态控制算法和动态控制算法，此控制算法下的被控系统具有更快的收敛速度和更好的系统性能。基于本项目的支持，共发表高水平论文9篇，其中SCI检索英文论文7篇（包括顶级期刊Automatica、IEEE Transactions on Cybernetics），高水平中文期刊2篇（包括《中国科学：信息科学》）；另有2篇英文论文处于审稿阶段；授权发明专利2项。该项目研究为非对称通信网络下异质多智能体系统事件触发控制的实际应用提供了理论支撑和技术支持。