面向复杂多任务的异构无人机集群分组调配

Heterogeneous UAV swarm grouping deployment for complex multiple tasks

针对复杂多任务下的异构无人机(unmanned aerial vehicle,UAV)集群分组调配问题,提出一种基于改进K均值和延迟接受(deferred-acceptance,DA)算法的先聚类后匹配方法。在任务聚类分组环节,通过离群点检测和固定初始聚类中心的方法来提高K-means聚类的精度,并设计余量裕度下的分组均衡性调整策略,在最优性的前提下提高分组的均衡性。在集群匹配分组环节,改进了DA算法,通过任务倾向的偏好列表快速生成预中选方案,并设计两阶段冲突消除来保证匹配的稳定性和收敛性。仿真实验表明,所提方法能够快速有效地解决复杂多任务下的UAV集群分组调配问题,具备良好的最优性和时效性。

A method of clustering before matching based on the improved K-means and deferred-acceptance(DA)algorithm is presented to solve the group deployment problem of heterogeneous unmanned aerial vehicle(UAV)swarm for complex multiple tasks.During the task clustering grouping stage,the approach of outlier detection and fixed initial cluster centers is exploited to increase the K-means clustering accuracy,and the grouping equalization adjustment strategy under margin is designed to enhance the grouping equalization based on the optimality condition.In the swarm grouping stage of matching,DA algorithm is developed by the preference list of task preferences to quickly generate a pre-selected scheme,and a two-stage conflict resolution is designed to ensure the matching stability and convergence.The simulation results show that the proposed method can solve the UAV swarm grouping deployment problem for complex multiple tasks quickly and effectively,and possess good optimality and timeliness.