Abstract This paper addresses the Dynamic Output Feedback Control (DOFC) problem for the network-based unmanned marine vehicle (UMV) systems, where the controller is located at the land-based console. The output measurements on the system are transmitted to the land-based console via a communication network. A novel event-based transmission mechanism is devised to alleviate the communication burden, where the threshold of the event function is intelligently updated by a Q-learning algorithm. Since the communication delay is inevitably present in the networked controller design, a time-delay system model is established which incorporates both event-based transmission and communication delay. With help of the Lyapunov stability theory and the extended reciprocally convex matrix inequality approach, a sufficient condition is obtained which ensures asymptotic stability of the tracking error system while a prescribed H ∞ disturbance attenuation performance is met. Finally, the simulation study on a benchmark system shows that the proposed controller design not only attenuates the Yaw Velocity (YV) error variation and the Yaw Angle (YA) oscillation, but also reduces the communication load effectively. Compared with the existing results in literature, it is shown that the proposed control algorithm can provide a better tracking performance.

中文翻译：

---

具有 Q-learning 的基于事件的智能输出反馈控制，用于无人船系统

摘要 本文解决了基于网络的无人船（UMV）系统的动态输出反馈控制（DOFC）问题，其中控制器位于陆基控制台。系统上的输出测量值通过通信网络传输到陆基控制台。设计了一种新颖的基于事件的传输机制来减轻通信负担，其中事件函数的阈值由 Q 学习算法智能更新。由于网络控制器设计中不可避免地存在通信延迟，因此建立了一个包含基于事件的传输和通信延迟的时延系统模型。借助李雅普诺夫稳定性理论和扩展互逆凸矩阵不等式方法，在满足规定的H ∞ 扰动衰减性能的同时，获得了保证跟踪误差系统渐近稳定性的充分条件。最后，对基准系统的仿真研究表明，所提出的控制器设计不仅可以减弱偏航速度 (YV) 误差变化和偏航角 (YA) 振荡，还可以有效降低通信负载。与文献中已有的结果相比，表明所提出的控制算法可以提供更好的跟踪性能。同时也有效地降低了通信负荷。与文献中已有的结果相比，表明所提出的控制算法可以提供更好的跟踪性能。同时也有效地降低了通信负荷。与文献中已有的结果相比，表明所提出的控制算法可以提供更好的跟踪性能。