一种无人机集群对抗多耦合任务智能决策方法

针对复杂场景下无人机集群对抗中协同目标分配和突防轨迹规划等多耦合任务的决策问题,提出了一种集群对抗多耦合任务智能决策方法.首先,针对无人机集群对抗中耦合任务多和决策空间大难题,结合集中式和分层式架构的优点,设计了面向多耦合任务的混合式深度强化学习架构,可提升多耦合任务间的协同性和集群对抗效能;其次,针对轨迹规划序贯决策...

An Intelligent DecisionMaking Method for MultiCoupling Tasks ofUAV ClusterCountermeasure

Aiming at the decision makingproblems of multi coupling tasks such as cooperativetarget assignment and penetration trajectory planning in UAV clustercountermeasure in complex scenes, an intelligent decision making method for multi coupling tasks inUAV cluster countermeasure is proposed. Firstly, aiming at the problems ofmulti coupling tasks and large decision making space in UAV cluster countermeasure, combined with the advantagesof centralized and hierarchical architectures, a hybrid deep reinforcementlearning architecture for multi coupling tasks isdesigned, which can improve the cooperation between the multi coupling tasks and the effectiveness of cluster countermeasure.Secondly, for the sparse reward problem of sequential decision making in trajectory planning, a trajectory construction method isdesigned. Thirdly, aiming at the scene uncertainty problem under the strongcountermeasure conditions, based on the UAV cluster red blue countermeasuresimulation platform, a red blue game training method based on multiple randomscenes is designed, which can enhance the generalization of the strategynetwork. Finally, by comparing with the traditional method, the centralized architecturemethod and the hierarchical architecture method, the simulation results showthat the effectiveness and the advanced nature of the proposed method areverified.