支持强化学习多智能体的网电博弈仿真平台
|
摘要
多智能体博弈仿真平台,可为智能网电作战单元提供智能体训练和验证环境,可为网电智能体提供任务场景、训练数据并评估训练结果.分析并设计了网电作战多智能体博弈仿真平台的主要功能、组成架构、逻辑架构和应用流程,最后对智能体模型设计架构、业务建模思路、奖励函数定义、典型业务场景等关键问题进行了分析研究,通过博弈仿真平台,可大大提升多智能体训练效率.
A multi-agent game simulation platform can provide a training and validation environment as well as task scenarios,training data and evaluation results for the intelligent network electric warfare unit.The main functions,composition structure,logical architecture and application flow of network-electric warfare multi-agent game simulation platform are analyzed and designed.Finally,some key issues are analyzed,such as the agent model design architecture,business modeling ideas,reward functions definitions,typical tactical scenarios.The simulation platform can greatly improve the training effciency of the multi-agent.
A multi-agent game simulation platform can provide a training and validation environment as well as task scenarios,training data and evaluation results for the intelligent network electric warfare unit.The main functions,composition structure,logical architecture and application flow of network-electric warfare multi-agent game simulation platform are analyzed and designed.Finally,some key issues are analyzed,such as the agent model design architecture,business modeling ideas,reward functions definitions,typical tactical scenarios.The simulation platform can greatly improve the training effciency of the multi-agent.
引文
1 ESEN A V.You only look twice:rapid multi-scale object detection in satellite imagery.[EB/OL].(2018-05-24)[2018-06-20].https://arxiv.org/abs/1805.09512?context=cs.CV.
2 GUPTA J K,EGOROV M,KOCHENDERFER M.Cooperative multiagent control using deep reinforcement learning[C]//International Conference on Autonomous Agents and Multiagent Systems.Berlin:Springer,2017:66-83.
3 Deepmind ai reduces google data centre cooling bill by 40[EB/OL].(2016-07-20)[2018-05-19].https://deepmind.com/blog/deepmind-aireduces-google-data-centre-cooling-bill-40/.
4 BUSONIU L,BABUSKA R,DESCHUTTER B.Acomprehensive survey of multiagent reinforcement learning[J].IEEE Transactions on Systems Man and Cybernetics Part C Applications and Reviews,2008,38(2):156.
5阳再清,许晓飞.海军战术导弹作战软件仿真测试平台设计[J].湖北航天科技,2003(3):30-33.
6黄克明,王涛,胡军.无人机作战仿真平台设计及其关键技术研究[J].兵工自动化,2016(1):90-92.
7阮武杰,赵伟华.基于HLA的直升机作战仿真平台研究[J].科技咨询导报,2007(27):9-11.
8顾闯,张强.基于Web Service的潜艇作战仿真平台框架研究[J].四川兵工学报,2009(11):60-63.
9张军,郭新钊,张红梅.一种作战实验仿真平台的实现方案[J].火力与指挥控制,2010,35(2):164-166.
10 SUKHBAATAR S,KOSTRIKOV I,SZLAM A,et al.Intrinsic motivation and automatic curricula via asymmetric self-play[EB/OL].(2017-05-15)[2018-04-27].http://arxiv.org/abs/1703.05407.
11刘全,翟建伟,章宗长,等.深度强化学习综述[J].计算机学报,2018,41(1):3-29.
12周来,靳晓伟,郑益凯.基于深度强化学习的作战辅助决策研究[J].空天防御,2018,1(1):45-49.
13徐志雄,曹雷,陈希亮.基于强化学习的无人坦克对战仿真研究[J].计算机工程与应用,2018,54(8):171-176.
14 GAO H L.Military image classification based on convolutional neural network[J].Application Research of Computers,2017(11):323-325.
15张文旭,马磊,贺荟霖,等.强化学习的地-空异构多智能体协作覆盖研究[J].智能系统学报,2018,13(2):48-53.
16 YANG Y D,LUO R,ZHOU M.Mean field multi-agent reinforcement learning[EB/OL].(2018-03-15)[2018-07-29].https://arxiv.org/abs/1802.05438.
2 GUPTA J K,EGOROV M,KOCHENDERFER M.Cooperative multiagent control using deep reinforcement learning[C]//International Conference on Autonomous Agents and Multiagent Systems.Berlin:Springer,2017:66-83.
3 Deepmind ai reduces google data centre cooling bill by 40[EB/OL].(2016-07-20)[2018-05-19].https://deepmind.com/blog/deepmind-aireduces-google-data-centre-cooling-bill-40/.
4 BUSONIU L,BABUSKA R,DESCHUTTER B.Acomprehensive survey of multiagent reinforcement learning[J].IEEE Transactions on Systems Man and Cybernetics Part C Applications and Reviews,2008,38(2):156.
5阳再清,许晓飞.海军战术导弹作战软件仿真测试平台设计[J].湖北航天科技,2003(3):30-33.
6黄克明,王涛,胡军.无人机作战仿真平台设计及其关键技术研究[J].兵工自动化,2016(1):90-92.
7阮武杰,赵伟华.基于HLA的直升机作战仿真平台研究[J].科技咨询导报,2007(27):9-11.
8顾闯,张强.基于Web Service的潜艇作战仿真平台框架研究[J].四川兵工学报,2009(11):60-63.
9张军,郭新钊,张红梅.一种作战实验仿真平台的实现方案[J].火力与指挥控制,2010,35(2):164-166.
10 SUKHBAATAR S,KOSTRIKOV I,SZLAM A,et al.Intrinsic motivation and automatic curricula via asymmetric self-play[EB/OL].(2017-05-15)[2018-04-27].http://arxiv.org/abs/1703.05407.
11刘全,翟建伟,章宗长,等.深度强化学习综述[J].计算机学报,2018,41(1):3-29.
12周来,靳晓伟,郑益凯.基于深度强化学习的作战辅助决策研究[J].空天防御,2018,1(1):45-49.
13徐志雄,曹雷,陈希亮.基于强化学习的无人坦克对战仿真研究[J].计算机工程与应用,2018,54(8):171-176.
14 GAO H L.Military image classification based on convolutional neural network[J].Application Research of Computers,2017(11):323-325.
15张文旭,马磊,贺荟霖,等.强化学习的地-空异构多智能体协作覆盖研究[J].智能系统学报,2018,13(2):48-53.
16 YANG Y D,LUO R,ZHOU M.Mean field multi-agent reinforcement learning[EB/OL].(2018-03-15)[2018-07-29].https://arxiv.org/abs/1802.05438.