Trajectory Prediction of Hypersonic Glide Vehicle Based on Empirical Wavelet Transform and Attention Convolutional Long Short-Term Memory Network

Hypersonic glide vehicle (HGV) has brought severe challenges to the existing radar system. Its trajectory prediction is a new problem affecting aerospace security, but it has not attracted extensive attention. The existing approaches provide unsatisfactory accuracy of HGV trajectory prediction due to the neglect of noise processing and the single prediction method. Here, we propose a HGV trajectory prediction method based on empirical wavelet transform (EWT) and attention convolutional long short-term memory (AConvLSTM) network to solve this problem. Firstly, aerodynamic acceleration components of HGV are selected as the prediction parameters, and the dynamic tracking model of HGV is established to estimate the prediction parameters. Secondly, we decompose and reconstruct the prediction parameters of HGV by the EWT and grey correlation degree to reduce noise interference to the prediction model. Finally, the prediction parameters after denoising are used to train the AConvLSTM network to predict the trajectory of HGV. Simulations show that the proposed method can effectively predict the trajectory of HGV using radar tracking data and achieve better prediction accuracy than the existing approaches.