Consensus Of Multi-agent Systems With Communication Constraints

In recent years, the coordination problem of multi-agent systems is a hot topic and has received spreading attention from various fields of science, including control engineering, mathematics, physics, robots and artificial intelligence. The consensus of multi-agent systems is the most basic and important problem. The information exchange between each two agents is the key to the consensus problem. Usually, the actual multi-agent systems are often in complex environments, for example, the systems may have communication noises or the connection may be broken because of the external disturbance and meanwhile the resource in the systems may be limited. In order to solve the problems in the communications and save more energy, the thesis studies the consensus of multi-agent systems with communication constraints by using the graph theory, matrix theory and control theory. The main contents of this thesis are given as follows.Firstly, the group consensus of multi-agent systems with communication noises and different control tasks is studied. The models of the agents are one-order discrete-time and the process of communication contains noises. Then the Lyapunov function is constructed for the error systems of the multi-agent systems. Furthermore, the sufficient conditions under which the group consensus can be reached are obtained. Then some simulations are given to show that if the sufficient conditions are satisfied, the group consensus of multi-agent systems can be achieved.Secondly, the thesis studies the leader-following consensus problem of nonlinear multi-agent systems under switching topologies with communication failure. The systems have several topologies including a null graph which can be taken as communication failure. And then the consensus protocols for followers are given. By analyzing the systems some sufficient conditions are obtained. Lastly, the simulations show that if the time of communication failure satisfies the conditions in the theorem the consensus can be achieved and the controlling gain can influence the speed of the consensus.Thirdly, the consensus of nonlinear multi-agent systems based on event-triggered control is considered, which can save more energy in the networks. In this protocol, each follower communicates with its neighbors only when its state measurement error is above certain threshold. Similarly, some sufficient conditions are given by analyzing the protocols. In simulation, different values are used to illustrate the validity of the theorem.Finally, a summary has been done for all discussions above. The further and more deeply researching works are presented for the problem of multi-agent systems.