一种空中飞行器行为事件的判别方法及装置
By converting radar echo sequences into two-dimensional time-frequency maps and combining them into three-dimensional time-frequency map sequences, and combining them with three-dimensional convolutional neural networks, the problem of accurately identifying aerial vehicle behavior events was solved, and high-precision behavior event recognition was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- BEIJING INST OF ENVIRONMENTAL FEATURES
- Filing Date
- 2025-10-24
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to accurately identify behavioral events of aircraft, such as turns, climbs, dives, rolls, and cobra maneuvers. The one-dimensional non-stationarity of radar echo sequences makes effective analysis difficult.
By performing time-frequency analysis on radar echo sequences to generate two-dimensional time-frequency maps, combining them into a three-dimensional time-frequency map sequence, and using a three-dimensional convolutional neural network for discrimination, accurate discrimination of aerial vehicle behavior events can be achieved.
It significantly improves the accuracy of identifying complex behavioral events of airborne vehicles, achieving a recognition accuracy of 95.6% for maneuvering events and 97.1% for non-maneuvering events, with an overall accuracy of 96.4%.
Smart Images
Figure CN121477150B_ABST
Abstract
Citation Information
Patent Citations
Radar moving target detection and classification integration method based on time-frequency graph convolutional neural network
CN109407067A
LDoS attack detection method based on FSWT time-frequency distribution
CN112788057A