摘要
面对瞬息万变的战场,如何有效地利用智能化技术实现计算机辅助决策,已经成为制约作战指挥控制技术发展的瓶颈。通过深入分析作战决策制定过程,将其转化为一个序列多步决策问题,使用深度学习方法提取包含指挥员情绪、行为和战法演变过程决策状态在内的战场特征向量,基于强化学习方法对策略状态行动空间进行搜索并对决策状态进行评估,直到获得最佳的行动决策序列,旨在实现未来战场"机脑对人脑"的博弈优势。
Faced with the rapidly changing battlefield situation,how to make effective use of intelligent technology to achieve computer-aided decision has become a bottleneck restricting the development of command and control technology.Through the in-depth analysis of the combat decision-making process,it is transformed into an issue of multistep sequential decision-making.Then the deep learning method is used to extract the characteristic vectors including the commander's mood,behavior and the decision-making state in the tactics evolution process.The reinforcement learning method is used to search in the action space of decision state and evaluate the decision state until obtaining an optimal action sequence decision,to gain the advantage at the game of machine versus human in the future battlefield.
出处
《空天防御》
2018年第1期31-35,共5页
Air & Space Defense
关键词
指挥控制
智能决策
深度学习
强化学习
Command and Control
Intelligent Decision-making
Deep Learning
Reinforcement Learning
