Abstract:

To address the limitations of traditional target trajectory prediction algorithms in multi-task adaptability and real-time performance, an analytical trajectory prediction method for entry gliding targets based on interactive multiple model tracking and dynamical mode decomposition （DMD） is proposed. Firstly, Singer model-based tracking model is established for entry gliding targets, enabling rapid estimation of target motion through interactive multiple model tracking algorithms. This provides essential target data support for subsequent feature extraction steps. Subsequently, inspired by the mode decomposition idea of mixed data sequences, a mode feature extraction method based on high-order DMD （HODMD） is designed. By analyzing multi-dimensional real-time tracking data of the target, this method extracts DMD mode features of target. Finally, dominant dynamical modes are selected based on the magnitude of mode amplitudes, and the target’s trajectory is reconstructed into an analytical expression with simple algebraic summation form. This enables fast analytical prediction of the target’s future trajectory. Simulation results of target trajectory prediction for the common aero vehicle-hypersonic? （CAV-H） demonstrate that the proposed method achieves high-precision analytical prediction of the target trajectory even when the target’s dynamical model is unknown. Compared to control variable estimation-based trajectory prediction methods, the proposed method offers superior trajectory prediction accuracy and computational efficiency advantages.

Key words: near-space re-entry gliding target, analytical trajectory prediction, interactive multiple model （IMM）, feature extraction, dynamical mode decomposition （DMD）