【作者】 陈磊；

【导师】 秦开宇；

【作者基本信息】 电子科技大学 ， 控制科学与工程， 2017， 博士

【摘要】 多智能体系统协调控制广泛应用于分布式计算、多机器人编队控制、多传感器融合等领域,各种不确定因素对其性能的影响一直是此领域的研究热点。本文主要围绕智能体系统跟踪控制问题中的通讯结构不稳定、信息不完全可知、时延及外部扰动等方面影响进行分析研究。首先,提出了联合连通变结构系统的李雅普诺夫函数分析判别方法。联合连通变结构是多智能体拓扑结构的一种特殊形式,可以模拟实际应用中通讯结构不稳定及智能体之间通讯受扰、被阻等情形;针对联合连通变结构拓扑图变化的特点,利用其通讯拓扑图拉普拉斯矩阵可分解为分块对角阵的特性,将非强连通系统等价分解成多个连通的子系统,并通过各子系统通讯拓扑矩阵特征值之和大于0的特点,实现李雅普诺夫函数的判别及跟踪误差收敛性的分析,解决了常规方法不能有效分析此类变结构系统的问题;同时进一步深入,在智能体之间通讯是单向的情况下(通讯拓扑图为有向图),证明了只需要领航者的信息在所处的子连通结构中是全局可达的,就能实现各智能体对领航者的跟踪。通过数值仿真,验证了该方法在无向图及有向图情况的有效性。在此基础上,进一步研究得出了通讯结构不稳定且领航者信息不完全可知情况的有界跟踪控制误差理论上界及其与相关参数关系。根据联合连通拓扑图的拉普拉斯矩阵可分解为分块对角阵的特点,将非强连通系统等价分解成多个连通的子系统,结合李雅普诺夫方法,研究各子系统结构变化与跟踪误差之间的关系,得到了智能体跟踪误差估计不等式,该不等式明确了智能体跟踪误差上界与通讯拓扑结构、目标加速度、时延等各参数之间的关系;利用跟踪误差估计方法研究联合连通变结构系统,能得出跟踪误差上界,具备较好的应用价值。通过数值仿真,验证了所提出的方法在存在时延及无时延情况下能实现目标的跟踪及跟踪误差上界的估计。接着,研究多智能体系统中的旋转目标跟踪控制问题,得出旋转跟踪控制方法及相关条件,并提出了一种基于位置信息的目标速度误差估计方法,解决存在速度未知情况的旋转跟踪控制问题。一是在无时延及存在时延两种情况下分别提出了基于速度与位置信息的旋转跟踪控制方法,并通过分析跟踪误差收敛情况证明该方法的有效性。在此基础上,针对无法对旋转领航者速度进行测量,存在信息不可知的系统,提出一种基于位置信息的速度误差估计方法,通过设定一个输入可知的中间变量,再利用变量与速度误差的关系,完成误差的估计及控制协议的设计。数值仿真证明,所提出的方法能在相应条件下实现对旋转领航者的跟踪。最后,研究了变结构系统及旋转跟踪系统的自适应控制问题,提出变结构系统的自适应控制分析方法,以及基于相互位置的自适应误差估计方法,获得了存在多种不确定因素变结构系统的跟踪控制条件,并得到了旋转跟踪控制中无需目标加速度信息的跟踪方法及条件。一是变结构系统中自适应控制问题的研究,通过系统的通讯拉普拉斯矩阵可分块的特点,将非强连通系统分解为多个连通子系统,结合变结构的分析方法及自适应控制方法,解决了联合连通变结构系统的自适应控制问题。同时,区别于传统的基于目标与跟随者之间的速度或位置误差的估计方法,采用智能体之间的相对速度、位置信息的估计方法,对未知信息、外部扰动进行估计。以此设计自适应控制协议,分别应用在动态目标与旋转目标的跟踪控制上,实现了两种情况下的自适应跟踪控制。通过数值仿真,分别验证了两类系统中,所提出的方法能实现对领航者的跟踪以及对未知变量的估计。更多还原

【Abstract】 Cooperative control for multi-agent systems has been investigated quite intensively and applied to many fields, such as distributed computations, formation control of multi-robot, multi-sensor networks. The study of the impact of various uncertain factors on the performance is the research hotspot of this field. This treatise analyzes and researches the tracking-control problems of multi-agent systems such as unstable communication structure, incomplete information, time-delay and external disturbance.First of all, this thesis proposes the function analysis discriminant method of Joint unicom variable structure system, basing on Lyapunov function. Jointly-connected topologies are a special form of communication topology, and can simulate some of the actual situation in the practical engineering, such as instability or interference of the communication. In the light of the characteristics of jointly-connected topologies, by using the characteristics of the Laplacian matrix of the communication topology, the Laplacian matrix can be decomposed into block diagonal matrix, so the non strongly connected system can be decomposed into several connected subsystems, then according to the characteristics of the eigenvalues of the communication topology matrix of each subsystem is greater than 0, determining the Lyapunov function,analyzing the convergence of tracking error, which solves the problem that the conventional method can not effectively analyze this kind of variable structure system.Further, when the communication between agents is unidirectional (communication topology is directed graph), it is proved that when the leader’s information is globally reachable in the sub-connected structure, various intelligent agents’ tracking of the leader could be realized. Through numerical simulation, the effectiveness of the proposed method in undirected graphs and directed graphs is proved.On this basis, it is concluded from further studies that the upper bound of the tracking error and the relationship between tracking error and some parameters, under the condition that the communication structure is unstable and the leader information is not completely known. According to the characteristic that the Laplacian matrix of the jointly-connected topology can be decomposed into a block diagonal matrix, the non-strongly connected system can be decomposed into several sub-connected subsystems. And through using the Lyapunov method, the relationship between the changing structure of each subsystem and tracking error is studied, with the inequality of tracking error estimation being obtained. The inequality defines the relationship between the upper bound of the tracking error and some parameters, such as the communication topology, the target acceleration and the time delay. Through studying the connected variable structure system based on tracking error estimation, the upper bound of tracking error could be obtained, which has a good application value. Through numerical simulation, the results show that the proposed method can achieve the leader-tracking and estimate the upper bound of the tracking error,with and without time delay.Then the rotating tracking control problem of the multi-agent system is studied,with the methods and conditions of rotation tracking control being obtained. Besides, a method of target velocity error estimation method based on position information is proposed, which solves the rotating tracking control problem with unknown velocity.First, the control strategies of rotating tracking based on velocity and position information with and without time-delay have been put forward, and by analyzing the convergence of tracking error, the validity of the method has been proved. On this basis,to solve the uncertainty information problem that the velocity of the rotating leader cannot be measured, a velocity error estimation method based on position information is proposed. Through setting the inputting of a knowable variable, and using the relationship between the variable and the velocity error, the error estimation and design of control protocol have been achieved. The numerical simulation shows that the proposed method can solve the rotating tracking control in the corresponding conditions.Finally, the adaptive control problem of variable structure system and rotating tracking system is studied. An adaptive control method for variable structure systems is proposed, and based on the adaptive error estimation method of relative position, the tracking control conditions for variable structure systems with multiple uncertainties are obtained, with the method and condition of rotating tracking control without acceleration measurement being established. In the study of adaptive control of variable structure systems, using the characteristics that the Laplacian matrix of the jointly-connected topology can be decomposed into a block diagonal matrix, the non strongly connected system is decomposed into several connected subsystems. Through combining with the analysis method of variable structure system and adaptive control method, the adaptive control problem of multi-agent system with jointly-conntcted topology is solved. Simultaneous,different from the traditional method of error estimating that is base on the velocity or position error between the leader and the follower, the method for estimating based on the relative velocity and position information among agents is applied to estimate the unknown information and the external disturbance. Then the adaptive control protocol is designed by using the relative velocity and position information, which is applied to the tracking control of active and rotating target,realizing the adaptive tracking control in two cases. Through the numerical simulation, it is respectively testified in the two kinds of systems that the proposed methods could achieve leader-following and estimation of unknown variables.更多还原