【作者】 王莉；

【导师】 陈增强；

【作者基本信息】 南开大学 ， 控制理论与控制工程， 2009， 博士

【摘要】 近年来,复杂动态网络受到来自物理、生物、数学、计算机、经济等不同学科领域的研究者越来越多的关注,其中基于群集智能的多智能体系统的协同控制已经成为一个研究热点。本论文对复杂动态网络的协同控制进行了多方面的研究,内容涉及到利用势函数的方法研究动态网络有leader和无leader时多智能体系统的运动。然后提出了基于图论和李亚普诺夫稳定性理论的方法来实现在复杂环境中多智能体系统在有限时间内达到预期状态的目的。最后利用卡尔曼滤波的方法,研究了多智能体系统的运动控制问题。每种研究方法都给出了有效的控制算法、稳定性分析及计算机仿真验证。研究结果有重要的理论意义与应用价值。论文主要创新性工作如下:(1)利用势函数的方法研究多智能体系统的协同控制,多智能体系统分为有leader和无leader的情况。多智能体系统的运动目的的是向预期位置聚集或编队并向目标位置移动。在势函数的构造中引入目标位置及环境信息对智能体的吸引能量,对预定运动目的的实现起到了积极的作用。(2)研究多智能体系统有限时间协同控制,基于图论知识和李亚普诺夫稳定理论的控制方法来实现多智能体系统在有限时间内达到预期状态的目的。考虑在多智能体之间存在通讯延时,进行了多智能体系统一致性算法的设计,保证多智能体系统在有限时间内实现状态一致。(3)研究多智能体系统有限时间协同控制,考虑在多智能体的运动环境中存在障碍的情况,多智能体系统的主要运动目的是能有效地避免与障碍物发生碰撞,并且绕过障碍物之后进行编队。通过计算机仿真,验证多智能体系统是否达到预期编队效果和避障的目的。(4)研究基于卡尔曼滤波的多智能体系统的运动控制。考虑了智能体间的信号在传输与检测过程中受到外来干扰与传感设备内部噪声的影响,使用卡尔曼滤波的方法预测和判断下一步迁移的位置,实现动态迁移的目的,利用这种机制可以缩短实现目标的时间。更多还原

【Abstract】 Recently, complex network has become an important interdisciplinary field, which attracts more and more researchers in physics, biology, mathematics, computer, etc, cooperative control of multi-agent system based on swarm-intelligence has become a hot topic in this area. In this paper, some research efforts are paid on the study on several problems in cooperative control of complex networks: movement of multi-agent system with leader and withoutleader is studied by using potential function approach. Then some control protocols based on algebraic graph theory and lyapunov stability theory are proposed to solve the finite-time reaching desired state of multi-agent system in complex entirvonment. Finally, movement of multi-agent system is studied by using kaermanfilter approach. Proper control and stability analysis and computer simulation are given in every research approach. The results have important theory significance and application value.The main innovative work and key points are listed as follows:(1) Movement of multi-agent system is studied by using potential function approach. The system are devided to two kinds: with leader and withoutleader. The movement object of system is to reach some desired or formation. In potential function, the attractive energy from desired situation or entirvonment to agents is used which has active to reach the object.(2) Finite-time cooperative control of multi-agent system is studied. some control protocols based on algebraic graph theory and lyapunov stability theory are proposed to solve the finite-time reaching desired state of multi-agent system. In the design of finite-time protocol, the communication time-delays are taken into account. The protocols can guarantee finite-time agreement of multi-agent system.(3) Finite-time cooperative control of multi-agent system is studied. Some obstacles are in the entirvonment. The movement object of system to. obstacle avoidance for multi-agent system and reach some desired formation is the main movement object. Through numerical simulations to show the effectiveness of the theoretical analysis results. (4) Movement of multi-agent system based on kaermanfilter approach is studied. The effect of entirvonment disturbance and sensor noise are taken into account. So kaermanfilter approach is used to predict the next situation of some agent to reduce the movement time.更多还原