【作者】 王冬梅；

【导师】 方华京；

【作者基本信息】 华中科技大学 ， 控制理论与控制工程， 2009， 博士

【摘要】 近年来，分布式大规模网络化群体系统的协调控制问题逐渐成为国内外智能系统理论界研究的热点。这源于自然界中广泛存在的群体现象，大规模群体系统不仅可以描述和解释大多数生物群体的群体协调行为和白组织现象，而且在工程上具有广阔的应用前景。因此，研究大规模群体系统的理论与应用具有重大的意义。本论文在前人的研究基础上，对现有生物群体行为的涌现机制进行深入理解，研究分析群体系统群集行为以及与之相关的动态特性问题。首先综述了大规模群体系统的发展历程，以及当前国内外关于群集行为建模与控制方法的理论研究和应用研究的现状。然后采用人工势能结合速度一致的策略，提出了一类带虚拟领航者的自适应的群集运动控制模型并分析了固定／切换两种拓扑结构下系统的稳定性，讨论了群体系统中局部外界环境和个体自身内部因素对群集行为的影响。随后，在此基础上，将该问题推广到有障碍物空间，探讨了全局环境未知且存在静态障碍物的情况下群集运动控制与避障问题；并将个体扩展为具有记忆能力的智能个体，提出了一类基于记忆能力的大规模智能群体群集运动和避障控制策略，使群体能安全地穿过障碍区。群体对应的拓扑结构为连通图是系统收敛必要条件；本文对连通性控制问题进行了深入研究，分析了连通性保持的实质；针对任意初始状态的群体系统提出了两种连通性控制策略；并采用相应的策略分别在系统对应的初始邻接图连通和不连通的两种条件下实现了连通性保持和群集运动控制，从而扩宽放松群集运动收敛的条件。时延是网络系统中普遍存在的现象；本文以跟踪动态的领航者的群集现象为研究对象，研究分析了带时延的群集运动控制的稳定性问题。首先采用了时域和频域稳定性分析方法分析了固定时延和时变时延系统的稳定性，得出了带相同时延的群体系统收敛的条件。然后，采用频域法分析带多时延的群体系统的稳定性，该方法不要求时延的导数已知，从而放松拓宽了以往研究的限制条件；得出了系统按衰减度为α＞0的指数稳定的条件；并采用线性矩阵不等式方法(LMI)进一步探讨了带参数扰动的多时延的群集系统，得出了系统稳定运行的条件。最后，通过Lyapunov方法分析估计系统的跟踪误差，探讨了时延对系统性能(跟踪误差)的影响。反馈机制是提高系统性能的有效方法；本文采用两种策略将反馈机制引入到有领航者的群集运动控制中，分析了该策略的鲁棒性以及容错性。模型中领航者按事先确定的轨迹运行，同时领航者需根据跟随者的状态信息自适应的调节自身的运行跟踪轨迹。研究分析表明该策略具有容错性，且减小了系统的跟踪误差，提高了系统的鲁棒性。本文最后针对基于最小外接圆方法的群集模型的局限性，在前人的研究基础上，采用求解任意维空间点集的凸壳的方法，将该模型推广到n维空间中，对n维空间中群体的聚集以及迁徙行为进行建模和稳定性分析。更多还原

【Abstract】 Recently, distributed coordination control of the scale networked swarm system has been a particularly active topic in intelligent system field. The effort has been inspired by common swarming phenomena existing in nature everywhere. The scale networked swarm system not only can describe and explain most of the coordinated behaviors and self-organization phenomena of biological swarms, but also has potential applications in engineering. Thus, it is signification to study the theory and applications of Large-scale swarm systems.In this dissertation, through understanding the emergent mechanism of biological swarm, the further investigations on flocking control and the relevant dynamic characteristic of swarm behavior will be given.Firstly, the history, the present status about the theory and applications research for swarm behavior are overviewed. Then, based on the strategy of combining artificial potential with velocity consensus, flocking control model based on virtual leaders which are chosen adaptively from the surrounding is built; the stability analysis for the fixed and switching swarm system are given; and the influence of local interior and exterior surrounding are discussed.After that, we extend or combine existing methods to obstacle avoidance and flocking control in complex environments; then, we produce an intelligent swarm by giving individual agents some limited memories, and further discuss the obstacle avoidance and flocking control of the intelligent swarm. The result shows that the intelligent swarm can more efficient to achieve flocking motion in obstacle environment.A common assumption in distributed coordination control problems is the connectedness of the underlying network. Further research is discussed in this dissertation. At first, we analyze the essential of connectivity maintenance. Then, two distributed control strategies for connectivity maintenance are proposed. At last, two kind of systems with arbitrary initial states are analyzed, i.e., connected system and disconnected system. Flocking motion with connectivity maintenance can be achieved for each of them. It is proved that flocking problem can be solved under a more relaxed condition. Time-delay is ubiquitous in network system. For flocking behavior with an active leader, we discuss the stability of flocking motion with time delay. At first, we adopt a frequency-domain approach to this problem and discuss the stability of system with fixed time delay and time-varying delay. We derive sufficient conditions for the system under the same communication delays. Then, the stability of flocking motion with multiple time-delay is discussed with a frequency-domain approach that needs no derivative of time. A more relax convergence condition is obtained. Moreover, we further discuss the stability of uncertain systems with multiple time delays and perturbations by employing a linear matrix inequality method, and obtain the stability condition. At last tracing error is estimated by Lyapunov method, the time- delay influence on tracing error is discussed.The strategy of information feedback is efficient to improve the performance of system. Two strategies of information feedback from followers to the leader are introduced to the flocking motion; the robustness and tolerance of the proposed method are studied. The leader tracks a pre-defined trajectory and at the same time the leader uses the feedback information from followers to the leader to modify its motion .The advantage of this control scheme is that it reduces the tracking errors and improves the robustness of the team cohesion to followers’ faults.At last, for the limition of the swarm model based on minimal circumcircle method, we extend the model to n-dimensional spaces; study the aggregating swarm behavior and flocking behavior; and analysis the stability of system.更多还原