位置依赖型多智能体系统分布式协同优化

Distributed consensus-based cooperative optimization problems in multi-agent systems have received much attention in recent years. In this project, distributed cooperative optimization problems in multi-agent systems with several different types of agent dynamics will be studied under a consensus framework. The models of position-dependent communication channels will be built and the distributed cooperative optimization methods will be proposed based on the models. Moreover, improvement in optimization methods will be explored by considering the characteristics of position-dependent communications. Focusing on the subject of distributed cooperative optimization in multi-agent systems, the aim of this project is to improve several existing results and investigate several new problems. During the model-building process, the distance constraints and delays will be comprehensively taken into account. The velocity information acting on agents with second-order dynamics will be considered when designing the optimization methods. The attention will be also paid to reducing the negative effects of distance constraints in the optimization task. The problems investigated in this project are based on the coordination control of multi-agent systems, which is a hot topic in the control community, and confirm to the development trend of this area.

以多智能体系统一致性为基础的分布式协同优化问题近年来受到了广泛的关注。本项目拟针对几类具有不同个体动态的多智能体系统，在分布式一致性的框架下研究分布式协同优化问题，建立位置依赖型通信信道模型，提出有效的分布式协同优化方法，并针对智能体通信依赖位置的特点提出改进方法。本项目旨在围绕多智能体系统分布式协同优化这一主题，改进现有结果并探索新问题，继而得到一系列较成体系的理论成果。本项目在信道建模中综合考虑智能体通信的距离制约和时延现象，在优化方法设计上着重探索速度信息对二阶动态智能体行为的影响，在改进方法上致力于降低通信的距离约束对优化任务的影响程度。本项目的选题根植于多智能体系统的协调控制这一控制科学领域的热点问题，符合该学科的发展趋势。

本项目研究一阶、二阶及一般线性多智能体系统的一致性及优化问题。针对一阶和二阶连续多智能体系统，采用一类分布式PD协议，研究网络通信时延影响下的包围控制问题，得出参数条件及网络通信时延的最大容许值；针对一阶离散多智能体系统，采用一类分布式PD协议，研究概率量化器作用下的量化一致性问题，证明在合适的参数下，系统能达到概率意义下的一致；针对二阶离散多智能体系统，采用一类基于参考模型的协议，以避免使用网络的图的连通度这一全局信息，证明在该协议下系统达到一致，并进一步研究概率量化器的影响，证明系统能达到概率意义下的一致；针对一般线性离散多智能体系统，分别采用基于状态反馈的静态协议和基于分布式观测器的输出反馈协议，研究量化操作对系统一致性的影响，证明在所设计的参数下，系统达到概率意义下的一致性。本项目的研究完善并发展了多智能体系统的分布式一致性及优化的算法设计方法体系，推动了分布式系统与控制理论的发展。