Research On Consensus Value And Cooperative Control Networked Multi-Agent Systems

Based on algebraic graph, matrix theory and control theory, the consensus value and cooperative control in networked of multi-agent systems are investigated. The coordinate and cooperation in group is realized by information exchange of communication topology, which is composed by each agent. Because each agent's decision depend on its state and coupling of others online states, the target of each other would come forth either conflict or consensus phenomena in group.The problem of consensus convergence and consensus value of multi-agent continuous systems are investigated by graph Laplacian matrix. For non-balanced topology, the system will globally asymptotically converge to the convex hull of initial states. By solving a set of linear equations, the convex combinations of equilibria is gotten, and finally obtain the consensus value of continuous system. If the topology is fixed and time-invariant, the consensus value of the linear time-delay system is also the convex hull of the initial states and is identical to the LTI system. Furthermore, the convergence convexity characteristic of linear discrete-time system is also investigated. If the topology is fixed and time-invariant, the consensus value of discrete-time systems also globally asymptotically converges to the convex combination of initial states, which is identical to continuous systems.In multi-agent systems, the algebraic connectivity is an important concept, which is a parameter to measure the stability, robustly, and convergence. Based on the connectivity theory of digraph, the relationship between graph Laplacian matrix and algebraic connectivity is investigated and the computer method is obtained.The unfalsified control is one type of model-free adaptive control. The control strategy enables one to design a controller that is consist with a performance objective and measured past input-output online data without plant models or assumptions on the plant. The theory works by eliminating candidate controllers that do not satisfy performance specification by evolving experimental data. The controller which remained to the last is a controller which satisfy a performance specification. Based on unfalsified control theory, the trajectory tracking control of mobile agent is analyzed. By adopted unfalsified control strategy to nonholonomic dynamic system directly, the tracking desired trajectory is achieved smartly and accurately. Furthermore, by adopted adaptive PID control algorithm of unfalsified control strategy to feedback linearization system, the mobile agent can track the desired trajectory rapidly.The output synchronization of mobile agent's aim is two aspects: making the speed of each agent consensus and tracking desired trajectory in group. Using adjacent information and incidence matrix, a decentralized control law is acquired, which is composed of alignment control and cohesion control. Finally, the stability of group is proved by the decentralized control law and the synchronization control objective is realized.