摘要
In highly dynamic and confrontational environments, the motion of aerial targets exhibits distinct spatiotemporal coupling characteristics and intricate maneuver patterns. The strategic behaviors of aerial targets are influenced by their motion states and dynamic interactions between opposing sides, which makes traditional trajectory prediction approaches insufficient for supporting subsequent maneuver strategy inference. To address this issue, this paper proposes an aerial target maneuver strategy inference framework integrating the Spatiotemporal Attention Transformer (STA-Transformer) and matrix game theory. First, a trajectory behavior recognition model based on STA-Transformer is established. Maneuver-aware positional encoding and a dual-pathway spatiotemporal feature separation mechanism are introduced to separately model temporal evolution patterns and correlations among multi-dimensional state variables, thereby enhancing the representation capability of spatiotemporal maneuver patterns under complex motion conditions. Second, based on predicted trajectory states, a dynamic threat payoff function is constructed by incorporating target motion characteristics, radar operating conditions, and tactical interactions. A matrix game model between aerial targets and ground radars is further formulated. Linear programming is adopted to solve the mixed-strategy Nash equilibrium, enabling the transformation from trajectory behavior perception to maneuver strategy inference. Simulation results demonstrate that under the test scenarios and experimental settings established in this work, the proposed method achieves accuracy of 95.3% and 96.4% for trajectory pattern recognition and maneuver strategy inference, respectively, outperforming comparative approaches. The findings verify the effectiveness of the proposed framework in aerial target behavior modeling and strategy inference, and provide a feasible scheme for integrating data-driven behavior modeling and game-based strategy inference in complex air defense scenarios.
| 源语言 | 英语 |
|---|---|
| 期刊 | IEEE Transactions on Aerospace and Electronic Systems |
| DOI | |
| 出版状态 | 已接受/待刊 - 2026 |
学术指纹
探究 'A Spatio-Temporal Attention Transformer and Matrix Game Model for Aerial Target Maneuver Strategy Prediction' 的科研主题。它们共同构成独一无二的学术指纹。引用此
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver