一种基于分层智能混合决策的多Agent框架

智能对于指挥决策模型非常重要,也是提高模拟训练和作战实验质量的关键。作战中指挥决策内容多,问题性质不一,对智能要求高。在分析指挥控制模型发展现状基础上提出模型对智能的需求,分析设定了智能的5个层次,提出一种混合Agent框架,并在框架中对每一层智能给出解决方案,框架兼具认知Agent、BDI Agent和刺激反应Agent的优长,对构建智能化的训练和实验系统有较大帮助。框架在指挥装备作战运用实验中得到应用,具有较好的理论意义和应用价值。

A Hybrid Agent Architecture Based on Hierarchical Intelligence

Modeling intelligence is very important for erecting C2 (Command and Control)decision model,which is a key factor to improving the quality of simulation training and combat experiment.In joint combat senario,construction of decision model are used to meet various problem,and claim for different level of intelligence.This article analyses the current situation and the future requirements of modeling intelligence.A hybrid Agent architecture is proposed and five levels of intelligence are contained as a solution.The architecture contains both the advantages of the BDI Agent and the IF-Then Agent.It benefits modeling training or experiment system with intelligence.This framework has been applied in C2 equipment combat application experiment,and it proved to have good significance and applied value.