Abstract

In this paper, an event-triggered neural network control method is proposed for autonomous surface vehicles subject to uncertainties and input constraints over wireless network. An event-triggered mechanism with three logic rules is employed to determine the wireless data transmission of states and control inputs. An event-driven neural network is applied to approximate the uncertainties using aperiodic sampled states. In addition, a predictor is employed to update the weights of neural network. An event-based bounded kinetic control law is applied to address the actuator constraints. The advantage of the proposed event-triggered neural network control approach is that the network traffic can be reduced while guaranteeing system stability and speed following performance. The closed-loop control system is proved to be input-to-state stable via cascade theory. The Zeno behavior can be avoided via the proposed event-triggered neural network control approach. A simulation example is provided to demonstrate the effectiveness of the proposed event-triggered neural network control approach for autonomous surface vehicles.

中文翻译：

---

无线网络上自主地面车辆的事件触发神经网络控制

摘要

本文提出了一种基于事件的神经网络控制方法，该方法适用于受不确定性和无线网络输入约束的自动驾驶水面车辆。具有三个逻辑规则的事件触发机制用于确定状态和控制输入的无线数据传输。应用事件驱动的神经网络使用非周期性采样状态来近似不确定性。另外，采用预测器来更新神经网络的权重。应用基于事件的有界动力学控制定律来解决执行器约束。所提出的事件触发神经网络控制方法的优点是可以减少网络流量，同时保证系统稳定性和速度跟踪性能。通过级联理论证明了该闭环控制系统是输入至状态稳定的。通过提出的事件触发神经网络控制方法可以避免Zeno行为。提供了一个仿真示例，以证明所提出的事件触发神经网络控制方法对自动水面车辆的有效性。