A Trajectory Prediction Algorithm for HFVs Based on LSTM

Aiming at the trajectory prediction problem of hypersonic flight vehicles skip trajectory, a prediction model of trajectory reachable area based on LSTM is proposed. According to the periodic jump characteristics of the skip trajectory, firstly, the trajectory is expressed as a time-dependent approximate expression consisting of a linear attenuation term and a sine term of amplitude attenuation. Then the dynamic characteristics of the parameters in the approximate expression are analyzed and the range of the parameters is calculated. Finally, according to the parameter range identified by the LSTM network and the trajectory expression, the final predicted trajectory reachable region is calculated. Due to the consideration of the prediction error caused by the drastic change of altitude and density in the jump trajectory, the trajectory accessible region prediction algorithm based on LSTM has more advantages than the previous trajectory prediction algorithm. The simulation results show that the trajectory reachable region prediction algorithm based on LSTM requires less prior knowledge and significantly improves the accuracy of trajectory prediction.