In order to tackle the real network constraints such as unknown state time-varying delays and limited communication resources, this article proposes a dynamic event-triggered control framework to satisfy the distributed consensus of heterogeneous strict-feedback multi-agent systems. Each follower agent is subject to nonlinear-in-parameter uncertainties regardless of satisfying the Lipschitz condition. Having considered the follower’s uncertainties, the adaptive neural network strategy is used to approximate uncertain functions. To further simplify the controller form, the minimal learning parameter technique is introduced to evade the updated information of the neural network weight vectors. Graph-based Lyapunov–Krasovskii functionals are utilized to handle the unknown state time-delays. A command filter dynamic event-triggered control scheme is then introduced to remove the complexity explosion regardless of global information of network. Different from the existing triggering strategies for the strict-feedback multi-agent systems, the proposed triggering mechanism is proved to include a positive bounded dynamical variable which results in the exclusion of the Zeno phenomenon and reducing the communication load in the controller to actuator channels. That can lead to the save of energy and communication resources, which are meaningful for employing the proposed scheme in the real practical applications. It is eventually proved that under the proposed approach, all the closed-loop network errors converge to a small neighborhood of the origin. A simulation example is finally given to show the effectiveness of the control scheme.

Keywords