Dynamic Event-Triggered Output Tracking of Unknown Heterogeneous Systems Using Learning Algorithm

This paper investigates the output tracking control problem of heterogeneous linear multi-agent systems with a novel dynamic event-triggered control strategy. In contrast to existing observer methods, the learning algorithm is first developed and applied to the observer such that the each observer corresponding to each follower can provide an optimal estimation of the leader’s state by optimizing a specified cost function. Then, a controller consisting of the observer’s state and the agent’s state is designed and learned by a data-based off-policy learning algorithm to achieve the optimal output tracking control. Under the learned gain matrix, to reduce the communication burden for each agent, a model-free dynamic event-triggered control strategy for each agent is developed to realize the optimal event-triggered output tracking control without depending on any prior knowledge. Rigorous analysis shows that the proposed algorithms not only ensure the model-free output tracking control while saving the limited bandwidth, but also exclude Zeno behavior. Finally, a numerical example is provided to verify the theoretical analysis.