Consensus Of Linear Multi-agent Systems With Disturbance

A multi-agent system is a complex dynamical system composed of many differ-ential or difference equations and connected via an information transmission network.Multi-agent systems are a class of mathematical models for many practical systems,which play an important role for analysis and control of the collective behaviors ap-peared in some natural or artificial systems. For example, the cooperative control of unmanned aerial vehicles, synchronization of fish school or bird flock, coordination of mobile robots, attitude consensus of a group of satellites and coordination tasks of distributed sensor networks. For a general nonlinear multi-agent system, it is usually difficult to analyze consensus theoretically. A effective method is linearization of the system models around the consensus equilibria. Then linear multi-agent systems have been paid great attention from the community of control theory. The control theory of linear multi-agent systems combines the classical linear control system theory with the algebraic graph theory and develops the networked linear control systems.Linear control systems with disturbance are a kind of important control systems considered in linear system theory. Many classical control problems address external disturbance such as output regulation, disturbance rejection, disturbance decoupling and filter problem. This thesis investigates the consensus problem of linear multi-agent systems with disturbance. By designing a consensus control protocol, the con-sensus is realized for linear multi-agent systems with disturbance. First of all, for linear multi-agent systems with disturbance, a consensus control protocol based on full-order state observers is proposed and a sufficient condition for consensus based on linear matrix inequality is obtained. In this case, it is assumed that the informa-tion transmission networks can transmit estimated states and estimated disturbance.Second, a consensus protocol based on function observers is designed and a sufficient condition for consensus is obtained.The results obtained in this thesis generalize the corresponding parts of linear system theory, enrich and develop the control theory of multi-agent systems.