The Application Research Of Distributed Event-Triggered Technology In Multi-Agent Consensus Control

More recently, there has been a tremendous surge of interested researchers from various disciplines of engineering and science in problems related to multi-agent networked systems with close ties to distributed cooperative control. This includes subjects such as consensus, synchronization of coupled oscillators, asynchronous distributed algorithms, collective behavior of flocks and swarms, algebraic connectivity of multi-agent networks, formation control for multi-robot systems, sensor fusion, random networks, dynamical graphs, complexity of coordinated tasks, optimization-based cooperative control, and consensus-based belief propagation in Bayesian networks and so on.The consensus problem serving as the fundamental branch of cooperative control plays an important role in real networked control. During the past decade, consensus problem has attracted increasing attention from various fields of science and engineering as a main issue of multi-agent networks. Though there are plenty of the previous research work, the previous works are all based on continuous-time feedback control techniques that have an obvious insurmountable deficiency: the designed control law requires real-time updates, which promotes network nodes to be equipped with high performance processors and high-speed communication channels. And more rigorously, it is assumed that each node can have continuous access to the states of its neighbors. In many practical applications, the nodes in a network transmit their own state values such as position, velocity, heading etc., to their individual neighbors not continuously but at some discrete time instants. Therefore, we eagerly need to develop a new approach to overcome the above problems while preserving the nice properties of stability and convergence.Event-triggered control mechanism offers a new point of view on how information could be sampled for control purpose. In a network, an autonomous node transmits its local state to its neighbors only when it is necessary, that is, only when a measurement of the local node state error reaches a specified threshold. Tabuada originally presented a triggering condition based on norms of the state and the state error, that is, the last measured state minus the current state of this node, where the measurement received at the controller is held a constant until a new measurement arrives. When this happens, the error is set to zero and starts increasing until it triggers a new measurement update.The main contribution of this dissertation is that it focuses on introducing the excellent event-triggered mechanism into the study of pinning control consensus in multi-agent networks and proposes some further improvements that can enhance the performance of the mechanism in practical implementation. The main points of this dissertation are summarized as follows:First, we briefly introduced the event-triggered mechanism in multi-agent networks, including centralized event-triggered control scheme and distributed event-triggered control scheme. In these previous works, though event-triggered strategy makes agents’ controllers avoid continuous communication, the cost is that each agent must collect its neighbors’ real-time states to calculate triggering function and then judge whether the event is triggered or not. In other words, the communication is still real-time and continuous at the point of view of agent individual.Second, aiming at the existing unavoidable weakness, we have proposed a novel event-triggered scheme for pinning control consensus of directed multi-agent networks with nonlinear dynamics. Note that this scheme is based on an improved and better event-triggered mechanism, in which the event judgment is only dependent on agent neighbors’ measurement states rather than real-time states, then the actuation update times and communication traffic are further reduced compared to previous event-triggered mechanism. Through this technology improvement, we solve the continuous-time feedback problem very well and achieve the purpose of communication reduction in pinning control consensus of multi-agent networks.In addition, considering limited communication, we further developed an alternative distributed iterative algorithm to determine each node’s event triggering time in advance, which avoids the judgments of event times and consumes less costly computing and communication resources to some extent.