摘要
作战推演技术在指挥训练、作战指挥、战法研究等领域中有着广泛的应用。传统的作战推演技术过于依赖与人在环路,推演效率低下,难以覆盖真实对抗中的各种可能情况。近年来广泛研究的深度强化学习技术,只能扮演战术级以下的智能蓝军用于模拟训练,因为其即不能胜任战役级决策的复杂度,也无法融入红队指挥官的智慧。于是提出一种混合智能的作战推演应用模式,同时融入人的智慧和深度强化学习训练的智能模型,用于态势演变预测和复盘分析、方案推演评估和探索优化。结合典型应用场景开展了实例化研究与概念验证,具有一定参考价值。
Wargame as a tool,has vast applications in training,command & control,and combat research domains.Traditional wargame technologies greatly rely on human wisdom in the loop.Its low efficiency makes it impossible to cover the various possible situations in real combats.Deep reinforcement learning technology,widely researched in recent years,can only play the role of the intelligent blue army below the tactical level in training systems,because it can neither adapt to the complexity of theater-level decision-making,nor integrate the red team commander's wisdom.This paper proposes a hybrid intelligence wargame application mode,which integrates human wisdom and AI models trained by deep reinforcement learning algorithms.It has applications in situation evolution prediction and replay analysis,COA(course of actions) development & analysis,and so on.Combined with a typical scenario,a case study and concept verification have been carried out,which has a certain reference value.
作者
金欣
王新年
周芳
丁冉
JIN Xin;WANG Xin-nian;ZHOU Fang;DING Ran(Science and Technology on Information System Engineering Laboratory,Nanjing Research Institute of Electronic Engineering,Nanjing Jiangsu 210007,China)
出处
《计算机仿真》
2024年第9期1-6,257,共7页
Computer Simulation
关键词
混合智能作战推演
态势演变预测
方案制定与分析
Hybrid intelligence
Wargame
Situation evolution prediction
COA development&analysis
