协同控制笔记

协同控制笔记

图论

假设集合A={a, b}，集合B={0, 1, 2}，则两个集合的笛卡尔积为{(a, 0), (a, 1), (a, 2), (b, 0), (b, 1), (b, 2)}。

矩阵理论：

一个矩阵，如果其每一行的非对角元的模之和都小于这一行的对角元的模，即称该矩阵是严格对角优势或强对角优势（ strictly diagonally dominant ）的。

 

若矩阵仅满足每一行的非对角元的模之和都小于等于这一行的对角元的模，但至少有一行的非对角元的模之和严格小于这一行的对角元的模，则称改矩阵是弱对角优势的。

 

这类矩阵有着广泛的实际背景，如很多微分方程边值问题的离散化方程的系数矩阵往往具有上面的性质，因此对这类矩阵的研究是十分重要的。这类矩阵还有一些重要性质,例如，若矩阵A是严格对角优势或不可约弱对角优势的，则 A是非奇异的；若A还是埃尔米特矩阵，且对角元皆为正数，则A是正定的。又如用直接法或迭代法解系数矩阵为对角优势矩阵的线性代数方程组时，可以保证算法的稳定性或收敛性。

词汇

Identically equal 恒等

Infimum、supremum 下、上确界

Equilibrium State 平衡状态

Monograph：专著

Interferometers：干涉仪

Combat：战斗

Surveillance：监视

Reconnaissance systems：侦查

Catastrophic：灾难性的

Instantiations：例示

Maintenance：维修、维持、保持

The rendezvous time, the length of a perimeter：约定的时间、边缘的长度

Swarm：群

Formation control：编队控制

Attitude alignment：姿态校正

Flocking：群列

Piecewise：分段的

Dwell times：驻留时间

Convex hull：凸包

Asynchronous：异步

Kronecker product 

Sampling instant：采样时刻

Embedded ：嵌入式的

Hazardous ：有害的、危险的

maneuver 熟练而谨慎的移动

rendezvous：约会

vice versa 反之亦然

contiguous 毗邻的

weighted power mean 加权幂平均

uniformly bounded, consecutive time intervals 一致有界的，连续的时间间隔

Part I Overview of Consensus Algorithms in Cooperative Control 

开头第一小段是介绍这一章的全文概述

协同控制的主要问题

a critical problem for cooperative control is to design appropriate distributed algorithms such that the group of vehicles can reach consensus on the shared information in the presence of limited and unreliable information exchange and dynamically changing interaction topologies. 

The abundance of embedded computational resources in autonomous vehicles enables enhanced operational effectiveness through cooperative teamwork in civilian and military applications.

协同控制的困难：

First, the research objective is to develop a system of subsystems rather than a single system. Second, the communication bandwidth and connectivity of the team are often limited, and the information exchange among vehicles may be unreliable. It is also difficult to decide what to communicate and when and with whom the communication takes place. Third, arbitration between team goals and individual goals needs to be negotiated. Fourth, the computational resources of each individual vehicle will always be limited.

中心协同控制的问题：

A centralized coordination scheme relies on the assumption that each member of the team has the ability to communicate to a central location or share information via a fully connected network. As a result, the centralized scheme does not scale well with the number of vehicles. The centralized scheme may result in a catastrophic failure of the overall system due to its single point of failure. A centralized coordination scheme relies on the assumption that each member of the team has the ability to communicate to a central location or share information via a fully connected network. As a result, the centralized scheme does not scale well with the number of vehicles. The centralized scheme may result in a catastrophic failure of the overall system due to its single point of failure.

分布式算法的作用：

distributed algorithms need to be developed to ensure that the team is converging to a consistent view of the coordination information. distributed algorithms need to be developed to ensure that the team is converging to a consistent view of the coordination information.

通信拓扑结构会是时变的：

The communication topology may be time varying due to vehicle motion or communication dropouts.

连续时间一致性算法公式：

，i=1…n

where a_ij(t) is the (i,j) entry of adjacency matrix An ∈ Rn×n associated with Gn at time t and x_i is the information state of the ith vehicle. Setting a_ij = 0 denotes the fact that vehicle i cannot receive information from vehicle j. Although the above equation ensures that the information states of the team agree, it does not dictate a specified common value

如果通信拓扑固定，增益不变，则最终稳态值是初始状态的线性组合，且通常是凸组合。

离散时间一致性算法公式P8(Macbook下是 P22)：

，i=1…n

连续时间一致性充要条件和离散时间一致性充要条件：p9-10（Mac22页）

1.This result implies that achieves consensus if and only if the directed communication topology has a directed spanning tree or the undirected communication topology is connected. 

2.As a result, under a time-invariant communication topology with constant gains , achieves consensus if and only if either the directed communication topology has a directed spanning tree or the undirected communication topology is connected 

The directed graphs： 

动态通信拓扑【文献192】

一致性的李雅普诺夫分析

【文献144】：For continuous-time consensus algorithm (1.1), [144] considers the Lyapunov candidate V (x) = max{x1,...,xn} − min{x1,...,xn}. It is shown 

in [144] that the equilibrium set span{1n} is uniformly exponentially stable if there is an interval length T > 0 such that, for all t, the directed graph 

of has a directed spanning tree 

【文献145】：[145] shows that discrete-time consensus algorithm (1.2) is uniformly globally attractive with respect to the collection of equilibrium solutions span{1n} if and only if there exists K ≥ 0 such that the union of the directed communication topologies has a directed spanning tree across each interval of length Kh, where h is the sample time. 

有时延的连续一致性计算公式，这里存在两种情况

一种是信息传输和过程都有时延，【文献158】

，i=1…n

一种是只有信息传输时延变化，【文献144】

，i=1…n

离散的也是加上一个时延系数，【文献230、258】

[258] shows sufficient conditions for consensus under dynamically changing communication topologies and bounded time-varying communication delays. 

异步一致性【文献31,35,67,68,140 】

由于异步，因此heterogenous vehicles, time-varying communication delays, and packet dropout must be taken into account in the same asynchronous consensus framework. 

1.2.3 Synthesis and Extensions of Consensus Algorithms 

fastest distributed linear averaging (FDLA) problem 在文献【259】中被定义

连续FDLA问题是找到一个权重W保证最快到达一致值

离散FDLA问题，因为W矩阵对称，可以转移成数值可解半定问题【文献259】

【文献106】with the motivation that the algebraic connectivity of the Laplacian matrix characterizes the convergence rate of the consensus algorithm（利用拉普拉斯矩阵的代数连通性刻画了一致性算法的收敛速度。）

【文献52】focuses on synthesizing a decentralized state feedback control law that guarantees consensus for the closed-loop system without disturbances as well as synthesizing a state-feedback controller that achieves not only consensus but optimal H2 performance for disturbance attenuation. （分散式状态反馈控制）

Consensus Algorithms for Single-integrator Dynamics 

Consider information states with single-integrator dynamics given by 

A continuous-time consensus algorithm is given by 

Hence，

连续时间算法达到一致渐进的充要条件：(状态转移矩阵趋向于同一常数)

证明这个充要条件的主要知识就是：

(书本27页)

几个重要引理：
 

其中1： 

则定理有：