基于强化迭代学习的分布式无人机编队控制研究

Research on Distributed Unmanned Aerial Vehicle formation Control Based on Reinforcement Iterative Learning

在无人机飞行过程中,基站主机所定义的编队信息会对与飞行器相关的转向行为造成一定的影响,故为保证无人机飞行器的稳定飞行状态,基于强化迭代学习对分布式无人机编队控制算法展开研究;计算强化学习函数的具体数值,通过迭代处理的方式,实现对迭代值概率系数的分布表示,完成强化迭代学习模型的设计;以此为基础,定义无人机编队拓扑结构,并求解信息迁移指标的具体数值,实现基于强化迭代学习的无人机编队信息迁移;在无人机编队控制器的配合下,建立分布式编队信息集合,并联合其中的编队数据样本,求解UAV控制参数,以便后续实现对分布式无人机编队的精准控制;再根据行进编队建模条件,完善控制算法执行流程,完成基于强化迭代学习的分布式无人机编队控制方法的设计;实验结果表明,在强化迭代学习模型的影响下,无人机偏航角始终保持在0°~90°的数值范围之内,表示飞行器按照基站主机所定义的编队信息飞行,能够始终保持较为稳定的运动状态,符合实际应用需求。

During the flight of unmanned aerial vehicles(UAVs),the formation information defined by the base station host has a certain impact on the aircraft s turning behavior.To ensure UAV stable flight,a distributed UAV formation control algorithm based on reinforcement iterative learning is researched.The specific numerical values of the reinforcement learning function are computed,and the distribution representation of the iterative value probability coefficient is achieved through iterative processing,implementing the design of the reinforcement iterative learning model.Based on this foundation,the UAV formation topology structure is defined,and the specific numerical results for the information migration indicators are solved to achieve the UAV information migration based on reinforcement iterative learning.With the cooperation of the UAV formation controller,a distributed formation information collection is established,and formation data samples are combined to solve the UAV control parameters,enabling the precise control of the distributed UAV formation.Furthermore,the control algorithm execution process is enhanced based on the modeling condition of marching formation,completing the distributed UAV formation control method based on reinforcement iterative learning.Experimental results show that under the influence on the reinforcement iterative learning model,the yaw angle of the UAVs remains within the range of 0°~90°,indicating that the aircrafts fly the formation information defined by the base station host with a relatively stable motion state,which is in line with practical application requirements.