【作者】 刘丹；

【导师】 罗小元；

【作者基本信息】 燕山大学 ， 控制理论与控制工程， 2012， 硕士

【摘要】 随着传感技术、通信技术的发展，多智能体系统受到了研究学者越来越多的关注，在许多领域获得了广泛的应用。多智能体群集凭借其分布式协作特点，成为多智能体系统协调控制的一个重要方向。多智能体系统的通信拓扑连通性往往会对智能体之间的相互作用产生巨大的影响。因此，本文在分析多智能体系统能控性与通信拓扑连通性之间关系的基础上，对多智能体系统目标追踪群集控制进行了研究。首先，针对切换拓扑下多leader多智能体系统，研究了多智能体系统能控性与通信拓扑之间的关系。通过引入并图的概念，分析了切换拓扑下系统能控性与其并图对应的固定拓扑下系统能控性之间的关系，提出了基于拓扑连通性的多智能体系统结构能控的充分必要条件，并进一步提出了多智能体系统能控的充分条件。其次，研究了以目标追踪为目的的多智能体群集与避障控制。针对质点型智能体模型的局限性，考虑智能体的实际半径，基于群集与避障规则设计了不同的人工势场。同时，为了保证通信拓扑的连通性，将目标作为leader，为每个智能体提供导航反馈。在此基础上提出了多智能体系统目标追踪群集运动的控制算法，使具有一定实际大小的智能体能够顺利完成目标追踪、群集与避障的行为目的。最后，为了克服群集算法通信能耗大的缺陷，研究了基于拓扑优化的多智能体目标追踪群集控制。在保证通信拓扑连通性的基础上，设计了一种基于最优刚性图的分布式拓扑优化策略。通过局部通信拓扑生成最优刚性图，并选择部分智能体为已知目标信息的智能体，从而减小了系统的通信复杂度。此外，进一步提出了基于拓扑优化策略的目标追踪群集控制算法，使得移动智能体群以较小的能耗完成以目标追踪为目的的群集运动。更多还原

【Abstract】 With the development of sensor and communication technologies, multi-agentsystems have been widely used in many areas and received a great attention from scholars.Flocking has become an important issue of cooperative control for multi-agent systemsdue to its distributed coordination. Communication topology connectivity has essentialinfluence upon the interactions among agents. As a consequence, this paper analyses therelationship between controllability of multi-agent systems and their communicationtopology connectivity, and further investigates the flocking control of target pursuit formulti-agent systems.Firstly, this paper investigates the relationship between controllability andcommunication topology connectivity for multi-agents systems with multiple leaders andswitching topologies. In this case, the concept of union graph is introduced to reveal therelationship between controllability of multi-agent systems with switching topologies andcontrollability of the corresponding systems with union graph. A necessary and sufficientcondition for structural controllability of the switching multi-agent systems is presented.Moreover, a sufficient condition for controllability of multi-agent systems with switchingtopology is proposed.Secondly, flocking and obstacle avoidance control for multi-agent systems with targetpursuit is studied. Considering the physical sizes of agents and model limitation, two kindsof artificial potential are established based on the rules about flocking and obstacleavoidance. In addition, virtual target is viewed as leader and provides navigation feedbackfor every agent to guarantee the topology connectivity. Based on these control inputs, thispaper proposes a control algorithm for flocking in target pursuit for multi-agent systems.With this algorithm, the multi-agent systems that are constituted with agents of physicalsize can successfully complete the behaviors of target pursuit, flocking and obstacleavoidance.Finally, to overcome the defect of vast communication consumption of the abovealgorithm, flocking control of target pursuit based on topology optimization for multi-agent systems is studied. Based on the guarantee for communication topologyconnectivity, a distributed topology optimization scheme based on optimally rigid graph ispresented. This optimization scheme generates optimally rigid graph for each agent withits neighboring flockmates and selects partial agents as the informed units to reduce thecommunication complexity. In addition, this paper proposes a flocking algorithm to makethe mobile agents result in behaviors of flocking and target pursuit with less energyconsumption.更多还原