摘要
针对无人机(UAV)集群协同搜索算法中灵活性不足、重要目标搜索效率差等问题,提出一种基于分级信息素的无人机集群协同搜索(HP-CS)算法。首先,面向复杂任务场景构建多维度的环境模型、目标模型等,通过精细化建模有效还原真实任务场景的关键特征;同时,改进传统的信息素模型,提出分级信息素模型;其次,利用知识图谱基于已知目标信息对目标的重要程度进行分级,重要目标对应高级别信息素,以此实现重要目标的优先搜索;最后,基于分布式架构构建协同搜索算法,无人机单机状态不会影响整个集群。仿真结果表明,所提算法能够实现对任务区域内目标的高效搜索,同时在目标重要程度不同的任务场景中表现良好,在单一目标类型搜索与多元目标类型搜索实验中表现显著优于对比算法,证明所提算法能提高目标尤其是重要目标的搜索效率。
To address the limitations of unmanned aerial vehicle(UAV)swarm cooperative search algorithms,such as insufficient flexibility and low search efficiency for critical targets,this paper proposed a hierarchical pheromone-based cooperative search algorithm for UAV swarm(HP-CS).Firstly,this paper achieved high-fidelity modeling of complex missions through multidimensional representations of environments and targets,preserving critical operational features.Meanwhile,it improved the traditional pheromone model by introducing a hierarchical pheromone model.Secondly,the algorithm leveraged knowledge graphs to classify the importance level of targets based on known target information.It linked high-priority targets to high-level pheromones,enabling prioritized search for critical objectives.Finally,the distributed-architecture-based cooperative search algorithm ensured that individual UAV states remain decoupled from the global swarm state.Simulation results demonstrate that the proposed algorithm efficiently searches targets within the mission area and maintains robustness in scenarios with targets of varying importance.In both single-target and multi-target search experiments,it significantly outperforms comparison algorithms,which proves it enhances search efficiency,especially for critical targets.
作者
李旭东
陈俊升
刘恒川
Li Xudong;Chen Junsheng;Liu Hengchuan(College of Software,Nankai University,Tianjin 300457,China;Tianjin Key Laboratory of Operating System,Nankai University,Tianjin 300457,China)
出处
《计算机应用研究》
北大核心
2026年第2期393-402,共10页
Application Research of Computers
基金
国家科技大项目(2018YFB0204304)。
关键词
无人机集群
协同搜索
分级信息素
知识图谱
分布式搜索
UAV swarm
cooperative search
hierarchical pheromone
knowledge graph
distributed search
