Working pattern recognition of phased array radar based on attention mechanism and feature fusion

In the modern battlefield environment, it is very important to realize the working pattern recognition of the enemy phased array radar accurately for subsequent accurate jamming and resource scheduling. In view of the existing radar mode recognition methods, there are problems such as insufficient use of feature parameters and difficulty in paying attention to the relationship between various features of radar signals. This paper provides a method of working pattern recognition for phased array radar based on attention mechanism and feature fusion. The radar signal pulse sequence is preprocessed by feature enhancement and feature fusion, and then fed into the deep learning neural network. We propose an improved Deep learning neural network called CBAM-DenseNet, which is a combination of DenseNet module and CBAM (Convolutional Block Attention Module). The algorithm can grasp the relation between each parameter of the radar signal, and then identify the radar working mode effectively. The experimental findings demonstrate that the identification precision of feature fusion and CBAM-DenseNet for each radar working pattern is more than 99%.