Radar individual identification system with task adaptability

Abstract

The invention discloses a radar individual identification system with task adaptability. The system is used for designing a network structure for a specific radar individual identification task, and utilizing an optimal network structure for classification and identification of a radar individual radiation source. The system comprises a data preprocessing module, a radar individual identificationnetwork structure searching module, a radar individual identification model fine tuning training module and a radar individual identification module. According to the method, a network structure mostsuitable for a radar individual identification task can be searched from a predefined operation set according to the corresponding task, and the high-precision and high-robustness radar individual identification model with task adaptability is obtained.

Description

technical field [0001] The invention relates to the field of radar radiation source signal processing, the field of deep learning network structure search, and the field of feature extraction, in particular to a radar individual recognition system with task adaptability. Background technique [0002] "Know the whole situation at your fingertips", the individual identification of radar radiation sources is the key to "battlefield control" in electronic warfare, an important basis for complex electromagnetic situation analysis and threat estimation of enemy targets, and an important basis for integrated combat platforms such as electronic reconnaissance and support alarms core. However, the traditional deep learning feature extraction method has poor universality in the environment of strong electronic countermeasures, and the radar individual recognition model has random characteristics when the parameters of the recognition method itself are determined. The influence of arti...

AI Technical Summary

Problems solved by technology

[0003] In view of the problem that the design of the current radar individual identification system itself does not have task universality and the accuracy and confidence need to be improved, the purpose of the present invention is to provide a radar individual identification system with task adaptability, from the predefined operation set Search to obtain the most suitable network structure for radar individual recognition tasks, and obtain a highly accurate and robust radar individual recognition model with task adaptability, which is useful for the rapid design and accurate identification of radar individual recognition models in complex electromagnetic countermeasure battlefields is of great significance

Images