面向数字孪生战场的智能体建模框架构建  

Construction of Agent Modeling Framework for Digital Twin Battlefield

作　　者：吴云超 傅琛 张宁馨 WU Yunchao;FU Chen;ZHANG Ningxin(Science and Technology on Information Systems Engineering Laboratory,Nanjing 210023,China)

机构地区：[1]信息系统工程重点实验室,南京210023

出　　处：《指挥信息系统与技术》2022年第4期19-25,31,共8页Command Information System and Technology

基　　金：装备发展部“十三五”预研课题资助项目。

摘　　要：针对现有军事对抗推演平台缺乏智能算法训练支撑能力的问题,分析了智能博弈技术的需求以及游戏对抗领域对智能决策问题的实现方法,研究了面向数字孪生战场的仿真数据模糊化以及任务分解等技术,构建了面向强对抗孪生战场的决策智能体建模框架,设计了数字孪生战场决策模型输入和输出层模型结构,为数字孪生战场中智能决策模型训练研发提供参考。Aimed at the problem of the existing military confrontation inference platform lack of the intelligent algorithm training support ability,the demand of intelligent game technology and the realization method for functional decision making problem in the field of game confrontation are analyzed.The technologies of the simulation digital fuzzy and the task decomposition for digital twin battlefield are studied.The modeling framework of decision agent for strong confrontation twin battlefield is constructed.The input and output layer model structure of digital twin battlefield decision model is designed.Thus it can provide reference for the training and development of intelligent decision model in digital twin battlefield.

关 键 词：强化学习 智能博弈 算法训练 作战决策 数字孪生 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程] TP391[自动化与计算机技术—控制科学与工程]