高超声速滑翔飞行器机动状态识别方法研究

高超声速滑翔飞行器(HGV)的迅猛发展改变了传统的作战样式，开辟了军事斗争的新领域。对HGV的机动状态进行识别可以为威胁评估、轨迹预测和防御决策提供有力支撑。为提高HGV机动状态识别精度，该文提出一种基于注意力机制的卷积长短时记忆网络识别模型(AT-ConvLSTM)。在对HGV进行机动建模和特性分析基础上，将HGV在空间的机动状态分为8类，构造了对应的特征识别参数，建立了包含不同初始条件和控制模式下HGV机动轨迹的轨迹库。推导了从雷达跟踪信息到特征识别参数的转换步骤，使用提出的状态识别模型对HGV机动轨迹的时空特征进行提取，并通过SoftMax分类器输出机动状态分类。最后，通过仿真实验对模型性能进行验证。结果表明，所提状态识别模型能够有效在线识别HGV机动状态，具有较好的实时性和准确性。

Research on Maneuvering State Recognition Method of Hypersonic Glide Vehicle

The rapid development of Hypersonic Glide Vehicle (HGV) has changed the traditional combat style and opened a new field of military struggle. Identifying the maneuvering state of HGV can provide a powerful support for threat assessment, trajectory prediction and defense decision. In order to improve the accuracy of HGV maneuver state recognition, an HGV maneuver state recognition model based on ATtention Convolutional Long Short-Term Memory network (AT-ConvLSTM) is proposed. First, on the basis of maneuvering modeling and characteristic analysis of HGV, the maneuvering state of HGV in space is divided into eight categories, and the corresponding feature recognition parameters are constructed. A trajectory library containing HGV maneuvering trajectories under different initial conditions and control modes is established. Then, the conversion steps from radar tracking information to feature recognition parameters are deduced. The proposed state recognition model is used to extract the spatial features of HGV motion trajectory, and the maneuvering state is classified by the SoftMax classifier. Finally, the algorithm is verified by simulation experiments. The results show that the proposed method can effectively identify HGV maneuvering state online, which has good real-time and accuracy.