基于深度置信网络和双谱对角切片的低截获概率雷达信号识别

发布时间：2018-03-24 20:48

本文选题：低截获概率雷达　切入点：深度学习　出处：《电子与信息学报》2016年11期

【摘要】：基于深度置信网络(DBN)对信号双谱对角切片(BDS)结构特征进行学习,实现低截获概率(LPI)雷达信号识别。该方法首先建立基于受限玻尔兹曼机(RBM)的DBN模型,对LPI雷达信号的BDS数据进行逐层无监督贪心学习,然后运用后向传播(BP)机制在有监督学习方式下根据学习误差对DBN模型参数进行微调,最后基于该BDS-DBN模型实现未知信号的分类和识别。理论分析和仿真结果表明,信噪比高于8 d B时,基于BDS和DBN的识别方法对调频连续波(FMCW),Frank,Costas,FSK/PSK 4类LPI信号的综合识别率保持在93.4%以上,高于传统的主成分分析加支持向量机法(PCA-SVM)和主成分分析加线性判别分析法(PCA-LDA)。
[Abstract]:Based on deep confidence network (DNN), the structural features of signal bispectral diagonal slicing (BDSs) are studied, and the low probability of interception (LPI) radar signal recognition is realized. Firstly, a DBN model based on constrained Boltzmann machine is established. The BDS data of LPI radar signal are studied layer by layer, and then the parameters of DBN model are fine-tuned according to the learning error by the backward propagation (BP) mechanism under the supervised learning mode. Finally, the classification and recognition of unknown signals are realized based on the BDS-DBN model. The theoretical analysis and simulation results show that the signal-to-noise ratio is higher than 8dB. Based on BDS and DBN, the synthetic recognition rate of FM CW FMCW / FSK-PSK4 LPI signals is more than 93.4%, which is higher than that of traditional principal component analysis (PCA) plus support vector machine (PCA) and principal component analysis (PCA) plus linear discriminant analysis (PCA-LDAA).
【作者单位】：空军工程大学航空航天工程学院;解放军95357部队;
【基金】：国家自然科学基金(61372167) 航空科学基金(20152096019)~~
【分类号】：TN957.51

【相似文献】