Time Domain Attention Mechanism Based Multi-Functional Radar Working Mode Recognition
Wenbo Li, Yang‐Yang Dong, Lidong Zhang, Chunxi Dong · 2023
With the rapid development of radar technology, multi-functional radars have been widely used in land, sea, air, and space. Due to the ability of multi-functional radar (MFR) to vary parameters freely with the change in battlefield scene, traditional classification network based radar working mode recognition method can no longer handle well. Therefore, a multi-function radar working mode recognition method using attention mechanism is proposed. Utilizing the Time Domain Attention Mechanism (TDAM) to assign different weights to pulse amplitude and waveform sequences at different times, enhancing the level of attention to key time data. Then, a concatenate convolutional neural network (CNN) is used for feature extraction and classification. In consideration of the receiver sensitivity and reconnaissance duration, simulation results have shown the proposed method is superior to convolutional neural networks (CNN) and convolutional autoencoder (CAE) methods.