多UCAV任务分配有限阶段MDP方法和算法  

Finite Stage MDP for Task Allocation in UCAVs Cooperative Control

作　　者：何鹏[1] 周德云[1] 王谦[1] 

机构地区：[1]西北工业大学电子信息学院,西安710129

出　　处：《火力与指挥控制》2012年第10期99-101,104,共4页Fire Control & Command Control

基　　金：航空科学基金资助项目(20080553019)

摘　　要：多UCAV(U nm anned Com bat A ir V eh icle)任务分配是无人机作战决策的重要内容,是一类求解困难的组合优化问题,目前尚无成熟高效的解决方案。通过马尔可夫决策过程(M DP)将任务分配问题化解为分阶段序列决策过程,并对当前阶段决策状态采用单一整数编码,提出了基于有限阶段M DP的求解算法。通过仿真实验表明:该算法可以快速精确地解决多UCAV任务分配的问题。Task allocation for multi-UCAV （Unmanned Combat Air Vehicle） is an important part of decision-making in operations with the UCAVs, and it is also a difficult combinatorial optimization problem which cannot be solved without mature and effective method. An algorithm based on the finite stage MDP is proposed by resolving the problem of allocation to be a sequence of phased decision-making process, and taking the current state of decision as single integral coding. The simulation results show that this algorithm for task allocation in UCAVs cooperation is effective and excellent.

关 键 词：无人作战飞机 马尔可夫决策 任务分配 组合优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]