Research on the Identification Method of ZPW-2000 Frequency Shifted Signal based on Auxiliary Classifier Generative Adversarial Network

As the track frequency shifted signals will cause signal mistranslation when subjected to interference. Therefore, it is very important to achieve the correct recognition of lowfrequency information in the case of noise, but the signal mistranslation caused by noise interference in the field is episodic, and a large number of datasets cannot be used for model training. In this paper, we propose to use Auxiliary Classifier Generative Adversarial Network (ACGAN) to achieve the data expansion of the signal, which provides the data basis for the classification and recognition of the signal. The results show that the method in this paper can effectively assist the training of the signal classification and recognition model, and can improve the classification and recognition accuracy from $\mathbf{9 0. 5 5 \%}$ to $\mathbf{9 7. 4 6 \%}$.