Abstract
<jats:p>Neural network models for automated analysis of spectral characteristics of electronic warfare (EW) signals have been investigated, which is a relevant direction for increasing the efficiency of detection, classification, and counteraction systems against modern threats. The main architectures of artificial neural networks, in particular, convolutional and recurrent networks, as well as their application for processing and recognition of spectral features of signals, have been analyzed. Methods for preparing training datasets, features of data preprocessing, and model optimization to improve classification accuracy are considered. Special attention is paid to the comparison of traditional spectral analysis algorithms with neural network approaches, and their advantages and limitations in complex electronic environments are identified. The results of experimental studies are presented, demonstrating the effectiveness of neural networks for automating the processes of EW signal analysis. The materials of the article may be useful for developers of electronic intelligence, monitoring, and protection systems, as well as for specialists implementing modern artificial intelligence technologies in the field of electronic warfare Keywords: Neural network models, spectral analysis, electronic warfare, automation, artificial intelligence, signal classification, convolutional neural networks, spectral characteristics, signal processing, electronic intelligence.</jats:p>