基于动态权重的知识积累与灵巧干扰识别方法

A Knowledge Accumulation and Smart Jamming Recognition Method Based on Dynamic Weights

针对低干噪比条件下灵巧干扰识别准确率不高与干扰特征难以积累导致的网络需要重新训练问题,本文将信号的平滑伪Wigner⁃Ville分布的时频二维图像作为输入,提出了一种基于动态权重的知识积累(Dy⁃namic Weighted Knowledge Accumulation method based on Convolutional Neural Network,DWKA⁃CNN)灵巧干扰识别方法,利用通道特征注意力机制,提升了低干噪比下模型的干扰识别能力,通过均值最近邻分层屏蔽网络权重,实现了在单一网络中的知识积累,与当前典型基于深度学习的灵巧干扰识别方法相比,无需每次重新训练即可学习多项干扰识别任务。并且仿真实验表明,与现有典型算法相比,该算法模型在7种雷达灵巧干扰分类数据集上的平均识别准确率显著提升,在低干噪比条件下分类性能优秀。

Aiming at the problem of low accuracy of smart jamming recognition under low JNR and the need to retrain the network due to the difficulty of accumulating interference features,this paper takes the time⁃frequency two⁃dimensional image of the smooth pseudo Wigner⁃Ville distribution of the signal as the input,and proposes a dynamic weighted knowledge accumulation method based on convolutional neural network(DWKA⁃CNN).The method uses the channel feature attention mechanism to improve the jamming recognition ability of the model under low JNR,and achieves knowledge accumulation in a single network by shielding the network weights through mean⁃nearest⁃neighbor hierarchy,which allows learning multiple jamming recognition tasks without retraining each time,compared with the cur⁃rent typical deep⁃learning⁃based dexterous jamming recognition methods.Simulation experiments show that,compared with the existing typical algorithms,the algorithm model significantly improves the average recognition accuracy on sev⁃en radar smart jamming classification datasets,and the classification performance is excellent under low JNR conditions.