基于VMD分解和多域联合分布的雷达辐射源识别

Radar emitter identification based on VMD and multi-domain joint distribution

针对现有雷达辐射源个体识别存在的识别准确率低、难以区分同型号雷达、缺乏实采信号验证等问题,提出了一种基于变分模态分解(variational modal decomposition,VMD)和时频能多域特征联合的雷达辐射源个体识别方法。首先对雷达信号进行VMD,然后对得到的模态分量分别在时域、频域和时频域提取13种特征参数,最后结合k近邻分类器实现雷达个体识别。采用实际采集的船用导航雷达信号对所提方法和现有的其他两种方法进行了对比验证,实验结果表明所提方法在识别准确率、算法鲁棒性方面优于现有两种方法,具有一定的工程应用价值。

Aiming at the problems of high computational complexity,insufficient ability to the same type radar and insufficient experiment with actual signal in the existing algorithms,a radar emitter individual identification method based on variational modal decomposition(VMD)and joint distribution feature of time-frequency-energy is proposed.The VMD is carried out on the radar signal,and 13 kinds of characteristic parameters in the time,frequency and time-frequency domain are obtained for the multiple modal components.Finally,the radar individual identification is carried out with the k-nearest neighbor(k-NN)network.The method proposed is verified with actual marine navigation radar signal,and the experimental results show that the method proposed is superior to the two existing typical methods in terms of recognition accuracy and algorithm robustness,and has practical engineering application value.