Broadband electromagnetic interference source identification based on fusion of Convolutional Block Attention Module and Resnet

Aiming at the broadband electromagnetic interference identification problem faced by civil aviation communication and navigation, this paper proposes a wideband electromagnetic interference source identification method based on the fusion of Convolutional Block Attention Module (CBAM) and residual network. Firstly, the frequency domain data of various types of interference sources are subjected to simple moving average (SMA) smoothing to avoid some background-induced fluctuations in the original data; then the one-dimensional spectral data of the interference sources are converted into two-dimensional grayscale images; finally, the two-dimensional grayscale images are used as inputs for neural network to classify and recognize the interference sources. In order to improve the recognition accuracy and efficiency, this paper incorporates the Convolutional Block Attention Module into the residual network, and the final experimental results show that the broadband interference source recognition model designed in this paper has an average accuracy of 98.42% for the recognition of a variety of different sources of interference, and realizes the effective recognition of broadband electromagnetic interference sources.