系统仿真学报 ›› 2017, Vol. 29 ›› Issue (1): 136-143.doi: 10.16182/j.issn1004731x.joss.201701019

• 仿真系统与技术 • 上一篇    下一篇

基于多分辨率的multi-Agent武器装备体系作战仿真研究

闫雪飞, 李新明, 刘东, 李亢   

  1. 装备学院复杂电子系统仿真实验室,北京 101416
  • 收稿日期:2015-04-24 修回日期:2015-06-15 出版日期:2017-01-08 发布日期:2020-06-01
  • 作者简介:闫雪飞(1989-), 男, 内蒙赤峰, 博士生, 研究方向为作战仿真与体系评估; 李新明(1965-), 男, 湖南益阳, 博士, 研究员, 研究方向为体系评估。
  • 基金资助:
    “核高基”重大专项支持课题(2013ZX010 45-004-002)

Battle Simulation Study on Multi-Agent Weapon System-of-Systems Based on Multi-Resolution

Yan Xuefei, Li Xinming, Liu Dong, Li Kang   

  1. Science and Technology on Complex Electronic System Simulation Laboratory of Equipment Academy, Beijing 101416, China
  • Received:2015-04-24 Revised:2015-06-15 Online:2017-01-08 Published:2020-06-01

摘要: 针对基于multi-Agent的武器装备体系作战仿真应用中,multi-Agent的决策与协作行为异常复杂而现有模型不能很好应对的问题,提出一种基于多分辨率建模与multi-Agent理论相结合的作战仿真模型。将战术级协作任务交给高分辨率Agent,将战略级决策任务交给具有宏观决策优势的低分辨率聚合Agent,其中,高层指挥Agent基于强化学习进行决策,底层作战Agent基于OA环(observe-action)进行协作,进而大幅降低了计算复杂度,并提高了决策收益。此外,基于面向Agent编程思想分别设计了高分辨率Agent和聚合Agent。基于JAVA语言进行了体系仿真验证。

关键词: Agent, 武器装备体系, 作战仿真, 多分辨率, Q-learning

Abstract: A model based on Multi-Resolution modeling technology with multi-Agent theory was put forward since the problem of the decision and cooperation of traditional multi-Agent system is too complicated in the field of battle simulation of weapon SOS. The tactic-level cooperation missions based on OA circle(observation-Action )was given to the high-Resolution Agents and the strategy-level decision missions based on reinforcement learning was given to the low-Resolution Agents. Besides, the class structure of high-Resolution Agent and Aggregation Agent based on the Agent oriented programming(AOP) was designed. At last, the ideas were validated through battle simulation for Weapon SOS based on Java language.

Key words: Agent, weapon system-of-systems, battle simulation, Multi-Resolution technology, Q-learning

中图分类号: