An LSTM-based aerodynamic parameter recognition method for skip gliding targets

Abstract

High-speed skip gliding targets have become a new threat to national defense due to its high speed and large maneuverability. The trajectories of these targets have skip gliding characteristics, which cause serious model mismatch and reduce the trajectory estimation accuracy. Considering the trajectory estimation of skip gliding targets, an aerodynamic parameter recognition method is proposed based on LSTM to solve the model mismatch problem in trajectory estimation. First, the dynamic model of the targets is given, and the aerodynamic labels are defined by analyzing the recognition requirements. Then, considering the skip gliding characteristics of the targets, LSTM-based aerodynamic recognition networks are designed and trained by the available trajectory data. Finally, the aerodynamic recognition results and effect on trajectory estimation are analyzed in typical scenarios. Simulation results show that the proposed method can accurately recognize aerodynamic parameters and suppress the model mismatch, which improve the accuracy and dynamic performance of trajectory estimation.