Research Of Consensus Of Multi-Agent Systems Under Network Environment

Multi-agent systems are typical complex dynamical systems, which contain a large number of autonomous or semiautonomous subsystems(agents) to accomplish a task by local information exchange, collaboration and interaction of communication network. The consensus is the most fundamental problem in distributed cooperative control of multi-agent systems, and has attracted the great interest of academic and industries. Consensus problem is to design appropriate protocols such that agents in a network converge to a consistent state by local communication between agents in the absence of central coordination and global communications. The study of consensus problem not only helps to understand inherent law of complex systems, but also provides important theoretical support for new techniques and applications of multi-robot systems, complex sensor network and multi-aircrafts formation, and so on.Due to some factors such as network bandwidth and rate limitation, communication constraints including communication delays and data dropouts are inevitably introduced between agents through communication networks for information exchange. At the same time, a large number of computations for each agent can also produce self-delays. Therefore, considering communication constraints, self-delays and limited network resources, the dissertation investigates the consensus problem of multi-agent systems under network environment. The main contents and contributions of the dissertation are listed as follows.(1) For the impact on consensus with multiple deterministic time-delays under network environment, firstly the average consensus problem in directed network with communication delays is studied. Considering multi-agent systems with fixed/switching topology, the analytical form is employed to describe the quantitative relationship between directed graphs and maximum allowable delays to ensure the achievement of consensus. Secondly considering self-delay of each agent, a weighted consensus protocol is proposed for the consensus problem in undirected network with two different deterministic time-delays and fixed topology, and the constraint rules are given between two different maximum allowable delays and the asymptotic convergence to consensus.(2) For the impact on consensus with multiple time-varying delays under network environment, firstly the average consensus problem in directed network with communication time-varying delays is studied. Considering multi-agent systems with fixed/switching topology, a reduced dimension model with time-varying communication delays is established. Based on Lyapunov stability theory, the reciprocally convex approach is employed to obtain the sufficient condition of the achievement of consensus, and the upper bound of time-varying communication delays is given. Further considering heterogeneous time-varying communication delays and self-delays, a weighted consensus protocol is used and some sufficient delay-dependent conditions are achieved. Furthermore, based on free-weighting matrices, the relationship between the allowable upper bound of the heterogeneous delays is also obtained.(3) For the impact on consensus with time-varying communication delays and limited network resources under network environment, the centralized event-triggered average consensus problem in directed/undirected network is investigated. A comprehensive consensus protocol with centralized event-triggered mechanism and time-varying communication delays is proposed, which can decrease the numbers of interaction information between each agent and reduce communication burden. A hybrid multi-agent system is then modeled by taking into account centralized event-triggered mechanism and communication delays, and sufficient conditions of the achievement of consensus for every agent are obtained.(4) For the impact on consensus with communication constraints and limited network resources under network environment, the decentralized event-triggered average consensus problem in undirected network is investigated. A consensus protocol with decentralized event-triggered mechanism and time-varying communication delays is proposed firstly, which can further reduce the global information and communication burden of centralized event-triggered mechanism. Then a reduced dimension hybrid multi-agent system by taking into account decentralized event-triggered mechanism, communication delays and data dropouts within one framework is proposed, and some sufficient conditions for average consensus of each agent and an upper bound of communication delay and maximal allowable number of successive data dropouts(MANSD) are obtained, which can conveniently provide the relationship between the triggering parameters, communication constraints and the system stability.