基于神经架构搜索及特征组合的便携式雷达个体识别方法
A portable radar individual identification method based on neural architecture search and feature combination
茆旋宇 1王欢 1王佩 1秦晋1
作者信息
- 1. 中国航天科工集团8511研究所,江苏 南京 210007
- 折叠
摘要
电磁环境日趋复杂、雷达辐射源体制愈发多样,面向现代电子对抗需求的辐射源个体识别任务充满严峻挑战,特别是针对便携式雷达目标,个体识别模型性能会随着环境、参数变化、调制方式变换等因素出现显著下降.针对上述问题,提出了一种基于神经架构搜索及特征组合的个体识别方法,利用神经架构搜索自适应组建深度神经网络模型,对原始信号、不同特征之间的组合进行结构化调整,提高模型面对便携式雷达信号时的识别性能.
Abstract
The electromagnetic environment is becoming increasingly complex,and the radar radiation source system is becoming more diverse.The individual identification task of radiation sources facing modern elec-tronic countermeasures is full of severe challenges,especially for portable radar targets.The performance of indi-vidual identification models will significantly decline with changes in environment,parameters,modulation meth-ods,and other factors.In response to the above issues,an individual identification method based on neural archi-tecture search and feature combination is proposed.By using neural architecture search to adaptively construct a deep neural network model,the combination between the original signal and different features is structurally ad-justed to improve the model's performance when confronted with portable radar signals.
关键词
神经架构搜索/个体识别/便携式雷达Key words
neural architecture search/individual identification/portable radar引用本文复制引用
出版年
2024
NETL
NSTL
万方数据