基于一致性差分进化的分布式任务分配

Consensus-Based Differential Evolution for Decentralized Task Allocation

针对异构无人飞行器(UAV)集群在强约束条件下执行多类任务的最优分配问题,将自适应参数差分进化算法与分布式架构结构融合,形成了基于一致性机制的分布式差分进化(CBDE)算法,在处理既定目标任务分配时达到了集中式方法的优化效果。在算法实施过程中,集群个体作为具有计算能力的局部优化器,异步执行改进的自适应参数差分进化算法,并通过UAV间的通信拓扑,按照一定的决策规则,共享个体适应度和分配结果,最终实现全局一致的效果。与集中式方法的仿真对比发现,CBDE算法求解中小规模任务分配问题的时间更短,执行效率更高,而且灵活的扩展性非常适合向大规模集群任务分配问题推广。

Aiming at the problem of optimal assignment of multiple tasks in heterogeneous Unmanned Aerial Vehicle( UAV) swarm under strong constraint conditions,a Consensus-Based Differential Evolution( CBDE) algorithm based on consistency mechanism is formed by integrating adaptive parameter differential evolution algorithm with distributed architecture structure,which achieves the optimization effect of centralized method when dealing with the given target task assignment. During the implementation of the algorithm,as a local optimizer with computing capability,the individuals execute the improved adaptive parameter differential evolution algorithm asynchronously,and share the individual fitness and allocation results according to certain decision-making rules through the communication topology between UAVs,so as to achieve the effect of global consistency. The simulation results show that,compared with the centralized method,the CBDE algorithm has shorter operating time for medium and small-scale task allocation problem,higher execution efficiency and flexible scalability,which is also suitable for large-scale swarm task allocation problem.