面向兵棋推演认知建模的交战火力单元智能预测方法  

Intelligent Prediction Method of Engaged Firepower Units for Cognitive Modeling of Wargames

作　　者：赵文栋 张明智[1] 贺筱媛[1] 郭圣明[1] ZHAO Wen-dong;ZHANG Ming-zhi;HE Xiao-yuan;GUO Sheng-ming(Joint Operations College,National Defence University,Beijing 100091;Unit 61267 of PLA,Beijing 101114,China)

机构地区：[1]国防大学联合作战学院,北京100091 [2]中国人民解放军61267部队,北京101114

出　　处：《指挥控制与仿真》2022年第4期42-47,共6页Command Control & Simulation

摘　　要：为适应智能化兵棋系统研究的需要,提出两种面向兵棋推演认知建模的交战火力单元智能预测方法。第一种是基于知识图谱表示学习的方法,它将知识图谱分解为属性子图,运用神经网络模型提取出各子图中火力单元的特征,通过计算目标与火力单元相关性进行预测。第二种是融合知识图谱与火力单元的运用行为的方法,它采用门控循环单元和注意力机制对火力单元的运用行为进行建模,将从知识图谱中提取的火力单元特征向量输入火力单元运用行为模型,以获得火力单元运用的行为特征,最终预测出将要交战的火力单元。实验表明,所提方法可以有效地预测出将要实施打击行动的火力单元。In order to adapt to needs of intelligent wargame system research,it proposes two intelligent prediction methods of engaged firepower units for cognitive modeling of wargames.The first approach is based on knowledge graph representation learning.It decomposes the knowledge graph into attribute subgraphs,the features of firepower units in each subgraph are extracted by using neural network model,the prediction is made by calculating the correlation between targets and firepower units.The second approach is based on the fusion of knowledge graph and the operational behavior of firepower units.It uses gated recurrent unit and attention mechanism to model the operational behavior of firepower units,the feature vector of firepower units extracted from the knowledge graph is input into the behavior model to obtain the behavior feature of firepower units,finally we predict which firepower units will be engaged.Experimental results show that the proposed methods can effectively predict the firepower units that will carry out the strikes.

关 键 词：火力单元预测 知识图谱 神经网络 表示学习 行为建模 

分 类 号：E911[军事] TP391.9[自动化与计算机技术—计算机应用技术]