复杂环境下时变信号的分布式平均跟踪问题研究

In recent years, distributed multi-agent systems have received intensive attention, and have become a very hot multi-disiplinary research area. Due to the complexity of the external environment and the requirements of the research problems, agents need to handle external and time-varying information. It thus becomes an urgent issue to consider how to use the information effectively to better serve the global control objective of the system. This project studies the distributed average tracking problem in complex environments from the following perspectives: asymmetry of the network topology, state non-measurability, robustness to initialization conditions, discontinuous updates of the control inputs, and input saturation. Specifically, this project will introduce a scaling variable to solve the distributed average tracking problem under asymmetric network topologies; will construct observer-based algorithms to solve the distributed average tracking problem without state measurements; will use the Laplace transform to analyze the robustness of the system to the initialization conditions; will combine the ideas of sampled-data control and event-triggered control to solve the distributed average tracking problem with discontinuous control input updates; and will desgin low-gain feedback algorithms to solve the distributed average tracking problem with input saturations. The derived results will enrich the research topics on distributed average tracking, and provide theorectical foundation and technical support for its practical applications.

近年来，具有分布式特性的多智能体系统引起了研究者的广泛关注，并逐步成为涉及多学科交叉的前沿热点领域。由于智能体外部环境的复杂性及研究问题自身的需求，智能体需要处理一些外部的、时变的信息。如何有效地利用这些信息，使之更好地服务于系统的全局控制目标，是一个亟待解决的问题。本项目以复杂环境下的分布式平均跟踪问题为研究内容，从网络的非对称性、状态的不可测性、初始值的鲁棒性、控制输入的不连续更新与饱和限制等角度入手，拟引入尺度变量解决非对称网络的分布式平均跟踪问题；拟构造基于观测器的算法，解决状态不可测的分布式平均跟踪问题；拟通过Laplace变换分析算法关于初始值的鲁棒性；拟结合采样控制与事件触发控制的思想，解决控制输入不连续更新的分布式平均跟踪问题；拟构造低增益的分布式平均跟踪算法，解决控制输入的饱和问题。本项目的研究结果将丰富分布式平均跟踪问题的研究内容，并为其实际应用提供理论基础与技术支持。

近年来，具有分布式特性的多智能体系统引起了研究者的广泛关注，并逐步成 为涉及多学科交叉的前沿热点领域。由于智能体外部环境的复杂性及研究问题自身的需求， 智能体需要处理一些外部的、时变的信息。如何有效地利用这些信息，使之更好地服务于系 统的全局控制目标，是一个亟待解决的问题。按资助项目计划书的研究内容和研究计划，本课题已顺利完成，在多智能体系统的分布式平均跟踪以及协同控制领域取得了系统性的理论成果，包括：加速度有界参考信号的分布式平均跟踪、Euler-Lagrange系统的分布式平均跟踪、复合拉普拉斯函数在分布式平均跟踪的应用、弱速度信息条件下的分布式平均跟踪、分布式平均跟踪与编队控制的联系、异质有向网络的编队控制等。研究成果包括专利4项（授权1项，公开审理3项）；期刊论文15篇（包括3篇IEEE Transactions on Automatic Control、3篇Automatica、1篇IEEE Transactions on Cybernetics），会议论文14篇（包括1篇IEEE Conference on Decision and Control、2篇American Control Conference）。培养了硕士毕业生6名，其中杨承帅、赵建丰同学获得了福建省优秀硕士学位论文奖励。