摘要
飞行机动识别技术在飞行训练考核、飞行数据挖掘等需要对飞行数据进行评估、理解的领域有重要的应用价值。针对现有飞行机动识别方法过于依赖领域知识的问题,考虑飞行数据的时序特征,提出一种基于序列模式识别的飞行机动识别的方法,实现飞行机动的实时自动识别。针对基于普通前向神经网络、循环神经网络和双向循环神经网络的飞行机动识别算法,通过仿真实验进行了识别效果对比分析。结果表明,循环神经网络方法能够识别出飞行数据中的序列模式,在降低领域知识依赖的同时,达到较高的识别精度。
Flight maneuver recognition technology has great application value in the field of flight training assessment,flight data mining,etc.To solve the problem that the existing methods rely too heavily on domain knowledge,a flight maneuver recognition method based on sequential pattern recognition is proposed considering the temporal characteristics of flight data,which is able to achieve real-time automatic recognition.Simulation experiments are conducted to compare the performance of three algrithms based on nerual network,recurrent nerual network and bi-direction recurrent nerual network,respectively.The results show that,the proposed method is able to identify sequential patterns in flight data.While reducing the dependence on domain knowledge,it performs higher recognition accuracy.
作者
周亚楠
聂帅
ZHOU Ya-nan;NIE Shuai(China Academy of Electronics and Information Technology,Beijing 100141,China)
出处
《中国电子科学研究院学报》
北大核心
2020年第6期539-546,共8页
Journal of China Academy of Electronics and Information Technology
关键词
飞行机动识别
序列模式识别
循环神经网络
LSTM
flight maneuver recognition
sequential pattern recognition
recurrent neural network
LSTM
作者简介
周亚楠(1990-),男,安徽人,博士,工程师,主要研究方向为空战仿真、体系仿真、仿真试验技术等,E-mail:zhouyanan1990@163.com;聂帅(1989-),男,山东人,博士,工程师,主要研究方向为体系仿真、赛博空间仿真等,。
