计算机科学
事件(粒子物理)
控制理论(社会学)
控制器(灌溉)
动态规划
梯度下降
人工神经网络
跟踪(教育)
职位(财务)
贝尔曼方程
最优控制
控制(管理)
控制工程
数学优化
人工智能
工程类
算法
数学
量子力学
生物
农学
物理
经济
教育学
心理学
财务
作者
Lihua Dou,Siyuan Cai,Xiuyun Zhang,Xiaotong Su,Ruilong Zhang
标识
DOI:10.1016/j.jfranklin.2022.02.034
摘要
This paper is concerned with the distributed formation control problem of multi-quadrotor unmanned aerial vehicle (UAV) in the framework of event triggering. First, for the position loop, an adaptive dynamic programming based on event triggering is developed to design the formation controller. The critic-only network structure is adopted to approximate the optimal cost function. The merit of the proposed algorithm lies in that the event triggering mechanism is incorporated the neural network (NN) to reduce calculations and actions of the multi-UAV system, which is significant for the practical application. What’s more, a new weight update law based on the gradient descent technology is proposed for the critic NN, which can ensure that the solution converges to the optimal value online. Then, a finite-time attitude tracking controller is adopted for the attitude loop to achieve rapid attitude tracking. Finally, the efficiency of the proposed method is illustrated by numerical simulations and experimental verification.
科研通智能强力驱动
Strongly Powered by AbleSci AI