【作者】 杨勇；

【导师】 樊渊；

【作者基本信息】 安徽大学 ， 控制理论与控制工程， 2018， 硕士

【摘要】 近年来随着分布式网络化系统的迅速发展,多智能体系统的研究也逐渐受到了来自众多领域研究者们的关注。与之前传统的单一控制系统相比,多智能体系统拥有低廉的运行成本、极高的置信度、容易实现和维护等许多优势,所以会受到广泛应用。可是因为这整个系统中的每个个体只具有有限的计算能力、能量和通信范围也是有限的,除此之外整个系统的通信网络带宽和资源也受到约束,这就使得对于多智能体系统的控制和设计变得比较复杂。因此本文将引入事件触发机制,在减少通信资源和计算资源消耗的同时确保系统达到一致性。本文的主要研究内容如下:首先在多智能体分布式事件触发机制的基础上引入一个新的内部动态变量,该动态变量的取值不仅受到智能体当前状态的影响,还受到智能体之前状态的影响。此前的事件触发机制中触发条件只与系统的当前状态和当前状态下与邻接智能体的状态差值有关,这种触发条件有一定的约束性,通过内部动态变量的引入系统的触发条件将更加具有可操作性。其次在线性多智能体系统一致性分析中提出一种基于输出反馈的事件触发机制,该触发条件可以用于某些系统中状态测量值比较难得到的系统,而在大部分控制系统中,系统的输出值总是比较容易得到。因此基于输出反馈的事件触发研究的应用范围将会更加广阔。最后事件触发机制对多智能体系统的控制产生了极大的便利,但在理论研究过程中关于系统无芝诺行为的证明方法选取上还存在一定的难度,本文基于此提出了一种基于时间和事件混合触发的机制,该触发机制能够避免系统中的智能体发生芝诺行为从而达到控制目标的要求。更多还原

【Abstract】 In recent years,with the rapid development of distributed network system,the analysis of multi-agent systems have been extensively investigated by various researchers from multiple areas.Compared with the traditional single control system previously,multi-agent systems are famous for low operational cost,high reliability,easy to implement and maintain and other superiority.Thus why these systems have been widely used.However,Because the agents in multi-agent systems have limited computing power,energy and communication range.In addition,the bandwidth and resources of the communication network in the whole systems are also constrained,which makes the control and design of the multi-agent system become more complex.This paper,we have been introduced the event-triggered mechanisms to reduce the cost of communication and computing resources,meanwhile guaranteeing the consensus of multi-agent systems.These are the major researches as follow:Firstly,A new internal dynamic variable is introduced on the basis of multi-agent distributed event-triggered mechanism.The value of the dynamic variable is affected not only by the current states of the agents,but also by the previous states of the agents.In the previous event-triggered control,the triggering condition of the agents are only affected by the current states and the error of the the agent without any connection with the previous states of the agents,the trigger conditions have certain constraints.By introducing an internal dynamic variable of the systems the trigger conditions will be more operable.Secondly,in the consensus analysis of linear multi-agent system,a trigger mechanism for output feedback has been proposed.The trigger conditions can be used to the systems,which measurement of system is difficult to get,but in most of the control systems,the output value of the system is always easier to get.Therefore,the application of event triggered control based on output feedback will be widerly used.Finally,event triggered mechanism in multi-agent systems’ control has got great convenience,but it is difficult in the process of the theory research methods of proof about zeno’s behavior.This paper proposes a mechanism based on the hybrid triggering control,the trigger mechanisms can avoid the agent occur zeno behavior in the systems,so it achieves the control objectives of the requirements.更多还原