基于长短时神经网络的目标意图识别

来袭目标意图识别是战场态势认知的重要部分。为充分利用探测到的空中来袭目标运动状态信息的时间相关性来提高意图识别精确度，本文提出一种基于长短时记忆神经网络(LSTM)的敌方空中目标作战意图识别方法。该方法首先利用仿真推演平台根据4种常见意图想定推演来袭目标数据，对生成数据进行清洗以及滑窗处理从而得到有效样本集，利用长短时记忆神经网络对生成样本集进行学习形成敌方空中目标作战意图识别模型。实验结果表明，利用长短时记忆神经网络来学习4种常见意图数据的运动及时间相关特征信息，预测准确率最终可达92%，取得了比传统分类器更好的效果。

Identification of target's combat intention based on Long Short Term Memory network

The identification of enemy air target''s combat intention is an important part of battlefield situation awareness. In order to improve the accuracy of Intention Identification by making full use of the temporal correlation of the detected air target motion state information,an identification of enemy air target''s combat intention method is proposed based on Long Short Term Memory(LSTM) network in this paper. Firstly, the simulation platform is utilized to deduce large number of air enemy target data according to four different intentions. Then, the generated data is cleaned and slide window processed to obtain the effective sample set. Finally, the LSTM network is adopted to learn the generated sample set and form the combat intention identification model of enemy air targets. Experimental results show that the prediction accuracy of network can reach 92% by using the LSTM network to learn the time-related characteristic information of data, and the result is better than that of traditional classifier.