基于混合MAS的保障任务分布规划设计与研究

Design and Research on Distributed Support Task Planning and Scheduling Based on Hybrid Multi-Agent System

任务自动规划是当前指控系统的重点研究内容,针对网络化指控环境下的信息分布特征和任务传播流程,建立了面向网络化指挥控制的保障任务自动迁移规划模型.以决策节点属性和任务需求为基础,研究了保障任务规划过程的混合多智能体(MAS)模型,在任务发起节点建立主控agent,设计了移动agent用于任务的迁移与本地计算,并以任务执行节点的资源信息模型进行局部求解,采用基于图最短路径搜索的任务耗费时间计算方法,建立了任务完成度模型,最后通过序列比较实现了分布式任务规划.通过建立的原型系统进行仿真实验,表明该方案能够完成任务的自动规划处理,具有可行性.

Task automated planning and scheduling is current focus in the area of command and control system. By taking the C2 environment of information distribution and task processes into account, a support task automated planning and scheduling （TAPS） model is proposed, which is oriented to the network command and control （C2）. Based on the decision-nodes＇ properties and tasks＇ requirements, a hybrid multi-agent system （MAS） model is studied, whose control agent （CA） is initialized on a beginning decision-node, and some mobile agents （MAs） are designed for migration and local computation. A local optimal solution is constructed by a MA at an implementing decision-node, which gives a task completion degree （TCD）. The time cost is calculated based on the model of shortest path searching problem on a weighted undirected graph. TCD＇s collection is used for intercomparison of tasks＇ local solutions, and finally a task＇s distributed planning and scheduling procedure is implemented. Experiments on the developed prototype system show that the method for the TAPS is practical.