Attention U-Net在雷达信号图像化分选中的应用研究

Research on The Application of Attention U-Net to Image Sorting of Radar Signals

针对海战场复杂电磁环境对雷达信号分选的挑战,采用改进的U-Net网络结合注意力机制提出新的分选方法。首先,将脉冲描述字转化为图像序列以适应深度学习处理。通过优化U-Net架构,融入注意力机制,有效提升模型对关键脉冲特征的识别与提取能力,实现像素级分类。通过此方法,系统能够精准搜索并归类所有雷达脉冲。实验证明,在海战场复杂电磁环境中,该方法显著提升了雷达信号分选准确率,提供了一种应对强干扰环境下的高效解决方案。这一研究成果证实了Attention U-Net在雷达信号智能分选中的优越性和实用性。

In order to face the challenge of radar signal sorting in the complex electronic environment of naval battlefield,this paper proposes a new sorting method by using an improved U-Net network combined with attention mechanism.Firstly,pulse description words are transformed into image sequences to adapt to deep learning processing.By optimizing U-Net architecture and integrating attention mechanism,the model can effectively improve the recognition&extraction ability of key pulse features and realize pixel-level classification.In this way,the system can accurately search and classify all radar pulses.Experimental results show that the proposed method significantly improves the accuracy of radar signal sorting in the complex electromagnetic environment of naval battle field,and provides an efficient solution to deal with strong interference environment.This research result confirms the superiority and practicability of Attention U-Net in radar signal intelligent sorting.