摘要
三体对抗场景下,对来袭导弹制导律的准确识别可以为其轨迹预测提供有力支持。针对此场景中来袭导弹运动与目标运动呈现半合作特征的特点,在收集导弹轨迹数据时加入不同目标机动,使数据更加贴合实际。提出了一种基于长短期记忆(Long Short-Term Memory,LSTM)神经网络的制导律识别方法,对来袭导弹制导律的类型进行识别。设计了一种加入注意力机制的LSTM神经网络,提高了网络模型的自适应能力以及泛化能力,使识别准确率及识别精度大幅度提高。实验结果证明,此方法识别准确率较高,且识别所需时间小,可以满足弹上使用需求。
In the three-body confrontation scenario,the accurate recognition of the guidance law of incoming missile can provide strong support for its trajectory prediction.In view of the semi-cooperative characteristics of incoming missile movement and target movement in this scenario,different target maneuvers are added when collecting missile trajectory data to make the data more realistic.A guidance law recognition method based on long short-term memory(LSTM)neural network is proposed to identify the type of guidance law for incoming missile.A LSTM neural network with attention mechanism is designed to improve the adaptive ability and generalization ability of the network model,and the recognition accuracy and precision are greatly improved.The experimental results show that this method has high recognition accuracy and short recognition time,which can meet the needs of on-board use.
作者
袁则华
崔颢
徐琰珂
王龙
周桃品
Yuan Zehua;Cui Hao;Xu Yanke;Wang Long;Zhou Taopin(China Airborne Missile Academy,Luoyang 471009,China;National Key Laboratory of Air-based Information Perception and Fusion,Luoyang 471009,China)
出处
《航空兵器》
CSCD
北大核心
2024年第6期57-63,共7页
Aero Weaponry
关键词
神经网络
制导律识别
LSTM
三体对抗
注意力机制
neural network
guidance law identification
LSTM
three-body confrontation
attention mechanism
