Adaptive Event-Triggered Heterogeneous Consensus of Leader-Following Multi-Agent Systems With Nonlinear Dynamic Topology

Abstract

This paper is interested in the event-triggered leader-following consensus problem for discrete-time multi-agent systems subject to nonlinear dynamic topology. Considering that the on-board resources of each agent are generally limited, a novel adaptive event-triggered strategy is presented in this paper through establishing a nonlinear transformation law of consensus errors, which provides an effective way to weaken the impact of small fluctuations on triggering behaviors after the error systems converge. Then, an interval type-2 fuzzy model is introduced to describe the nonlinear dynamic topology where the dynamic characteristics contain both nonlinear time-varying law and uncertain parameter. In view of the discrepancies in adjacency relationships between different agents, the heterogeneous fuzzy-dependent controllers are employed to further decrease the consensus error. Meanwhile, some sufficient conditions are deduced to solve the designed controllers while ensuring that the consensus of multi-agent systems can be achieved with desired $H_{\infty }$ performance. Ultimately, advantages of the presented leader-following control strategy are illustrated by two examples.Note to Practitioners—Many practical control scenarios, such as formation of unmanned air vehicles, can be described by a leader-following consensus problem of multi-agent systems with limited onboard resources. To save the resource consumption of agent, a novel event-triggered scheme is provided, which has more potential to reduce the computational tasks to a greater extent. In addition, aiming to promote the convergence of tracking error, the agent will tend to strengthen connections with the one that have not yet reached consensus, and weaken connections with the others. Therefore, a nonlinear dynamic topology with time-varying weight is introduced during the design of heterogeneous controller, such that the adjacency relationship between agents can be adjusted online. The effectiveness of the obtained results is demonstrated by an actual experiment. It is expected that the presented strategy can be further extended to multi-agent systems where network structures are more complex, such as the one with multiple groups of agents.