Hypersonic Vehicle Trajectory Classification Using Improved CNN-LSTM Model

Hypersonic vehicle has become a hotspot of aircraft research because of its high speed and complex maneuvering mode. The classification of the trajectory of hypersonic vehicle is of great significance to the trajectory prediction and interception of hypersonic vehicles. This paper proposes a neural network structure combining Convolutional Neural Network (CNN) and Long Short-Term Memory Network (LSTM) for the classification of hypersonic aircraft flight trajectories. The classification experiment of the hypersonic vehicle trajectories generated by simulation shows that the model has good performance under the condition of introducing observation noise. In addition, this paper gives a comparison between Hypersonic Convolutional Neural Network (HCNN) and proposed model to show the advantages of the proposed model in classification.

I. Introduction

Hypersonic vehicle usually refers to aircrafts, missiles and other winged or wingless vehicle that have a cruising speed of more than 5 times the speed of sound. Its unique aerodynamic shape and high speed enables it to perform complicated maneuvers, to shorten the duration of flight, to provide good crypticity and surpass conventional aircraft in many aerospace fields.