具有弱通讯的多智能体分布式自适应协同跟踪控制

针对具有弱通讯的高阶严反馈非线性多智能体系统,提出一种新颖的分布式自适应反演控制方法,并研究了该系统的协同跟踪控制问题.首先,以5个智能体作为被控对象,其中一个智能体以分布式结构组成"领航者-跟随者"编队模式,并且将领航者的运动速度作为整个编队系统的前行速度,其余智能体作为跟随者跟随领航者编队运动.其次,采用自适应反演...     

具有干扰的多自主体系统的群集运动与速度同步

针对具有干扰的二阶多自主体系统,提出了一种应用个体局部信息的分布式控制协议,给出了具有领航者的自主群体在受到外界干扰时取得群集运动与速度同步的充分条件。首先,对自主体设计一个干扰观测器。然后,基于观测器设计分布式控制协议,保证了具有领航者的自主群体能够取得群集运动与速度同步。最后通过例子进行仿真,结果表明结论的有效性。     

基于有限记忆的智能群体的群集运动与避障控制

在全局环境未知且存在静态障碍物的情况下,将群体扩展为具有有限记忆能力的智能群体,研究了该群体的群集运动及其避障行为.模型利用智能个体记忆单元的信息实时产生动态切换的拓扑网络结构;在人工势能结合速度一致策略的基础上,采用滚动窗口方法引入虚拟领航者,利用极限环方法解决群体避障问题,实现了快速且安全的群集运动控制.仿真结果验证了该模型的有效性.     

非线性多智能体网络的分布式包容控制

针对具有本质非线性动态的多智能体网络,研究分布式包容控制问题.假设只有部分个体已知领航者信息,依据相对位置和速度信息设计分布式控制律.基于代数图论、矩阵理论和Lyapunov稳定性分析方法,得出非线性网络实现渐近包容控制的充分条件.当跟随者之间有向强连通且每个跟随者都至少存在一个领航者与其通信,可选取合适的控制增益使得跟随者渐近收敛到由多个领航者所围成的静态凸包中.仿真实例验证了理论分析的正确性和有效性.     

全局未知环境下智能群体群集运动与避障控制

研究了在全局环境未知且存在静态障碍物的情况下,智能群体的群集运动控制问题.模型在人工势能结合速度一致策略的基础上,利用了滚动窗口的方法实时产生虚拟领航者引导群体向目标位置运行;并利用极限环方法解决了群体避障问题.基于滚动窗口方法构造的虚拟领航者,充分地利用了实时测得的局部环境信息,具有自适应性;极限环方法避障使得群体能够平滑地绕过障碍物,克服了传统人工势场法避障的局部最小问题.     

复杂网络特性与多智能体的一致性

对复杂网络特性与多智能体系统的一致性的关系进行研究,以BBV和BA网络模型为例,分别研究了度正相关性和度负相关性的网络模型中多智能体系统一致性的收敛速度.应用计算机仿真实例,分析了无尺度网络模型的平均距离和平均集群系数与多智能体系统一致性的影响,得到了复杂网络结构与多智能体系统一致收敛速度的关系.该研究对于构建一个多智...     

基于群体社会制度策略的主从多智能体汇聚

研究一阶主从多智能体网络的汇聚问题,基于多智能体网络群体的社会制度,提出3种汇聚策略,即民主、独裁和混合策略。假设在策略中,只有领导者知道全局任务,即汇聚到一个已知的目标点,跟随者只需要与领导者群体保持联系,从而能够到达目标点。研究发现,主从网络多智体个体在属性上可以分为目标智能体,领导者智能体,非孤立跟随者从智能体和孤立跟随者从智能体。研究结果表明民主策略容易实现所有智能体到达目标点,但移动速度较慢;在独裁策略作用下,智能体网络可以很快实现汇聚,但个别跟随者从智能体由于初始条件下感应不到环境中的领导者智能体而永远停留在原来的位置;引进混合策略后,即可以保证所有的智能体到达目标点而且又可以保持较快的到达速度。通过对模型和拓扑结构的分析,得出3种策略各自的特点,并以4个领导者和5个跟随者组成的主从智能体网络为例进行仿真。     

基于多智能体强化学习的动态频谱分配方法综述

认知无线电和动态频谱分配技术是解决频谱资源短缺问题的有效手段。随着近年来深度学习和强化学习等机器学习技术迅速发展, 以多智能体强化学习为代表的群体智能技术不断取得突破, 使得分布式智能动态频谱分配成为可能。本文详细梳理了强化学习和多智能体强化学习领域关键研究成果, 以及基于多智能体强化学习的动态频谱分配过程建模方法与算法研究。并将现有算法归结为独立Q-学习、合作Q-学习、联合Q-学习和多智能体行动器-评判器算法4种, 分析了这些方法的优点与不足, 总结并给出了基于多智能体强化学习的动态频谱分配方法的关键问题与解决思路。     

交互式多智能体进化算法及其应用

将多智能体技术应用于交互式进化计算领域，提出一种新的交互式多智能体进化算法。该方法让固定在网格上的相邻智能体之间展开竞争和熏优智能体本身的自学习，来提高智能体的能量，从而使得算法获得较强的全局收敛能力和局部搜索能力。算法以较快的进化速度收敛，并且用户每次只需要选择2个左右最感兴趣的个体，使得评价具有易操作性和轻松性，从而有效缓解用户的疲劳。服装设计的仿真实验证了算法的可行性和有效性。     

VGE多用户协作建模与虚拟演示

提出将多智能体技术集成到虚拟地理环境（VGE）中,以有效地虚拟表达面向“人”GIS。主要研究VGE中基于遗传算法（GA）和多智能体系统（MAS）的多用户之间的协作。分析和研究了基于GA的多用户集体操作行为学习的算法,建立多智能体与对象之间动态交互的动力学作用模型和运动策略;最后以进化的3个智能体如何协作完成群体推立方体箱任务为例,基于V-Realm Builder 2.0设计了VGE中3用户集体推箱的虚拟世界原型,并用Simulink 6.0进行建模与虚拟演示。     

Cluster-Delay Consensus for Second-Order Nonlinear Multi-agent Systems

This paper investigates the second-order nonlinear multi-agent systems subject to the cluster-delay consensus. The multi-agent systems consist of leader and agents, whose dynamics are second-order nonlinear. The objective is that the agents track the leader asymptotically with different time delays, i.e., the agents in different groups reach delay consensus, while the agents in the same group reach identical consensus. To guarantee the cluster-delay consensus for the second-order multi-agent systems, a new control protocol is proposed. Then some corresponding conditions for cluster-delay consensus are derived by using Lyapunov directed method and matrix theory. Finally, the effectiveness of the theoretical analysis results are verified by some numerical simulations.     

Fixed-Time Leader-Following Formation Control of Fully-Actuated Underwater Vehicles Without Velocity Measurements

This paper is concerned with formation control of fully-actuated underwater vehicles (FUVs), focusing on improving system convergence speed and overcoming velocity measurement limitation. By employing the fixed-time control theory and command filtering technique, a full state feedback formation algorithm is proposed, which makes the follower track the leader in a given time with all signals in the system globally practically stabilized in fixed time. To avoid degraded control performance due to inaccurate velocity measurement, a fixed-time convergent observer is designed to estimate the velocity of FUVs. Then the authors give an observer-based fixed-time control method, with which acceptable formation performance can be achieved in fixed time without velocity measurement. The effectiveness and performance of the proposed method are demonstrated by numerical simulations.

     

Distributed tracking control of second-order multi-agent systems under measurement noises

This paper considers a leader-following tracking control problem for second-order multiagent systems（MASs） under measurement noises and directed communication channels.It is assumed that each follower-agent can measure the relative positions and velocities of its neighbors in a noisy environment.Based on a novel velocity decomposition technique,a neighbor-based control law is designed to realize local control strategies for these continuous-time agents.It is shown that the proposed consensus protocol can guarantee that all the follower-agents track the active leader.In addition,this result is extended to a more general case with switching topologies.Finally,a numerical example is given for illustration.     

复杂系统的适应维生方法

复杂适应系统理论是复杂性科学的一个重要分支,它通过系统主体间的适应性造就稳定性,并通过单个主体的规则选择和多个主体之间的学习和应变来达到复杂系统的适应和维生。复杂适应系统理论通过主体间的适应性来分析复杂组织在涌现生成之后主体之间是如何组织在一起,并维持这个组织系统的稳定性,它为我们分析复杂组织的维生机制提供了科学的方法工具。     

Leader-following coordination of multi-agent systems with information feedback

Based on the strategy of information feedback from followers to the leader, flocking control of a group of agents with a leader is 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. The results of simulation are provided to illustrate that information feedback can improve the performance of the system.     

营销代理商与企业的营销活动分析

对完全信息下的营销代理商与企业的营销活动进行了研究.论文先对企业无分销代理商时,经理和营销部门建立主从对策模型进行分析,再对企业有分销代理商时,经济、营销部门和分销代理商建立二级主从对策模型进行分析.对两种情况下的结果进行了比较,说明营销部门把一部门产品委托给代理商销售,对经理、营销部门和分销代理商都是有利的,这也间接论证了代理商参与销售活动的机制和原因.     

Leader selection for strong structural controllability of single-integrator multi-agent systems

This paper addresses the leader selection problem for strong structural controllability (SSC) of multi-agent systems (MASs). For a path-bud graph, it is proved that only one leader is required to guarantee the SSC of MASs. For a special type of topologies, based on the partition of the topology into disjoint pathes and path-buds, it is proved that the MASs is strongly structurally controllable if the root nodes of the pathes are selected as leaders. For general topologies, an algorithm is provided to determine the agents that can behave as leaders. For some special topologies, the minimum number of leaders guaranteeing the robust strong structural controllability (RSSC) of MASs is also obtained. Two examples are given to verify the effectiveness of the results.